Add pandas and datasets for benchmarks

2025-10-21 07:13:52 +08:00 · 2025-06-04 06:51:59 -07:00
1894 changed files with 89073 additions and 208794 deletions
--- a/.buildkite/generate_index.py
+++ b/.buildkite/generate_index.py
@ -8,8 +8,7 @@ template = """<!DOCTYPE html>
 <html>
    <body>
    <h1>Links for vLLM</h1/>
-        <a href="../{x86_wheel_html_escaped}">{x86_wheel}</a><br/>
-        <a href="../{arm_wheel_html_escaped}">{arm_wheel}</a><br/>
+        <a href="../{wheel_html_escaped}">{wheel}</a><br/>
    </body>
 </html>
 """
@ -22,25 +21,7 @@ filename = os.path.basename(args.wheel)

 with open("index.html", "w") as f:
    print(f"Generated index.html for {args.wheel}")
-    # sync the abi tag with .buildkite/scripts/upload-wheels.sh
-    if "x86_64" in filename:
-        x86_wheel = filename
-        arm_wheel = filename.replace("x86_64", "aarch64").replace(
-            "manylinux1", "manylinux2014"
-        )
-    elif "aarch64" in filename:
-        x86_wheel = filename.replace("aarch64", "x86_64").replace(
-            "manylinux2014", "manylinux1"
-        )
-        arm_wheel = filename
-    else:
-        raise ValueError(f"Unsupported wheel: {filename}")
    # cloudfront requires escaping the '+' character
    f.write(
-        template.format(
-            x86_wheel=x86_wheel,
-            x86_wheel_html_escaped=x86_wheel.replace("+", "%2B"),
-            arm_wheel=arm_wheel,
-            arm_wheel_html_escaped=arm_wheel.replace("+", "%2B"),
-        )
+        template.format(wheel=filename, wheel_html_escaped=filename.replace("+", "%2B"))
    )
--- a/.buildkite/lm-eval-harness/configs/Meta-Llama-3-8B-QQQ.yaml
+++ b/.buildkite/lm-eval-harness/configs/Meta-Llama-3-8B-QQQ.yaml
@ -0,0 +1,12 @@
+# For vllm script, with -t option (tensor parallel size).
+# bash .buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh -m HandH1998/QQQ-Llama-3-8b-g128 -b 32 -l 1000 -f 5 -t 1
+model_name: "HandH1998/QQQ-Llama-3-8b-g128"
+tasks:
+- name: "gsm8k"
+  metrics:
+  - name: "exact_match,strict-match"
+    value: 0.419
+  - name: "exact_match,flexible-extract"
+    value: 0.416
+limit: 1000
+num_fewshot: 5
--- a/.buildkite/lm-eval-harness/configs/models-large.txt
+++ b/.buildkite/lm-eval-harness/configs/models-large.txt
@ -3,3 +3,4 @@ Meta-Llama-3-70B-Instruct.yaml
 Mixtral-8x7B-Instruct-v0.1.yaml
 Qwen2-57B-A14-Instruct.yaml
 DeepSeek-V2-Lite-Chat.yaml
+Meta-Llama-3-8B-QQQ.yaml
--- a/.buildkite/lm-eval-harness/run-lm-eval-gsm-hf-baseline.sh
+++ b/.buildkite/lm-eval-harness/run-lm-eval-gsm-hf-baseline.sh
@ -2,7 +2,7 @@
 # We can use this script to compute baseline accuracy on GSM for transformers.
 #
 # Make sure you have lm-eval-harness installed:
-#   pip install git+https://github.com/EleutherAI/lm-evaluation-harness.git@206b7722158f58c35b7ffcd53b035fdbdda5126d#egg=lm-eval[api]
+#   pip install lm-eval==0.4.4

 usage() {
    echo``
--- a/.buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh
+++ b/.buildkite/lm-eval-harness/run-lm-eval-gsm-vllm-baseline.sh
@ -3,7 +3,7 @@
 # We use this for fp8, which HF does not support.
 #
 # Make sure you have lm-eval-harness installed:
-#   pip install git+https://github.com/EleutherAI/lm-evaluation-harness.git@206b7722158f58c35b7ffcd53b035fdbdda5126d#egg=lm-eval[api]
+#   pip install lm-eval==0.4.4

 usage() {
    echo``
@ -46,6 +46,6 @@ while getopts "m:b:l:f:t:" OPT; do
 done

 lm_eval --model vllm \
-  --model_args "pretrained=$MODEL,tensor_parallel_size=$TP_SIZE,add_bos_token=true,trust_remote_code=true,max_model_len=4096" \
+  --model_args "pretrained=$MODEL,tensor_parallel_size=$TP_SIZE,distributed_executor_backend=ray,trust_remote_code=true,max_model_len=4096" \
  --tasks gsm8k --num_fewshot "$FEWSHOT" --limit "$LIMIT" \
  --batch_size "$BATCH_SIZE"
--- a/.buildkite/lm-eval-harness/test_lm_eval_correctness.py
+++ b/.buildkite/lm-eval-harness/test_lm_eval_correctness.py
@ -18,14 +18,12 @@ RTOL = 0.08

 def launch_lm_eval(eval_config, tp_size):
    trust_remote_code = eval_config.get("trust_remote_code", False)
-    max_model_len = eval_config.get("max_model_len", 4096)
    model_args = (
        f"pretrained={eval_config['model_name']},"
        f"tensor_parallel_size={tp_size},"
        f"enforce_eager=true,"
        f"add_bos_token=true,"
-        f"trust_remote_code={trust_remote_code},"
-        f"max_model_len={max_model_len}"
+        f"trust_remote_code={trust_remote_code}"
    )
    results = lm_eval.simple_evaluate(
        model="vllm",
--- a/.buildkite/nightly-benchmarks/README.md
+++ b/.buildkite/nightly-benchmarks/README.md
@ -7,11 +7,11 @@ This directory contains two sets of benchmark for vllm.
 - Performance benchmark: benchmark vllm's performance under various workload, for **developers** to gain clarity on whether their PR improves/degrades vllm's performance
 - Nightly benchmark: compare vllm's performance against alternatives (tgi, trt-llm and lmdeploy), for **the public** to know when to choose vllm.

-See [vLLM performance dashboard](https://hud.pytorch.org/benchmark/llms?repoName=vllm-project%2Fvllm) for the latest performance benchmark results and [vLLM GitHub README](https://github.com/vllm-project/vllm/blob/main/README.md) for latest nightly benchmark results.
+See [vLLM performance dashboard](https://perf.vllm.ai) for the latest performance benchmark results and [vLLM GitHub README](https://github.com/vllm-project/vllm/blob/main/README.md) for latest nightly benchmark results.

 ## Performance benchmark quick overview

-**Benchmarking Coverage**: latency, throughput and fix-qps serving on A100 (the support for FP8 benchmark on H100 is coming!) and Intel® Xeon® Processors, with different models.
+**Benchmarking Coverage**: latency, throughput and fix-qps serving on A100 (the support for FP8 benchmark on H100 is coming!), with different models.

 **Benchmarking Duration**: about 1hr.

@ -28,34 +28,16 @@ See [vLLM performance dashboard](https://hud.pytorch.org/benchmark/llms?repoName
 ## Trigger the benchmark

 Performance benchmark will be triggered when:
-
 - A PR being merged into vllm.
 - Every commit for those PRs with `perf-benchmarks` label AND `ready` label.

-Manually Trigger the benchmark
-
-```bash
-bash .buildkite/nightly-benchmarks/scripts/run-performance-benchmarks.sh
-```
-
-Runtime environment variables:
-
- `ON_CPU`: set the value to '1' on Intel® Xeon® Processors. Default value is 0.
- `SERVING_JSON`: JSON file to use for the serving tests. Default value is empty string (use default file).
- `LATENCY_JSON`: JSON file to use for the latency tests. Default value is empty string (use default file).
- `THROUGHPUT_JSON`: JSON file to use for the throughout tests. Default value is empty string (use default file).
- `REMOTE_HOST`: IP for the remote vLLM service to benchmark. Default value is empty string.
- `REMOTE_PORT`: Port for the remote vLLM service to benchmark. Default value is empty string.
-
 Nightly benchmark will be triggered when:
-
 - Every commit for those PRs with `perf-benchmarks` label and `nightly-benchmarks` label.

 ## Performance benchmark details

 See [performance-benchmarks-descriptions.md](performance-benchmarks-descriptions.md) for detailed descriptions, and use `tests/latency-tests.json`, `tests/throughput-tests.json`, `tests/serving-tests.json` to configure the test cases.
-> NOTE: For Intel® Xeon® Processors, use `tests/latency-tests-cpu.json`, `tests/throughput-tests-cpu.json`, `tests/serving-tests-cpu.json` instead.
->
+
 ### Latency test

 Here is an example of one test inside `latency-tests.json`:
@ -78,7 +60,7 @@ Here is an example of one test inside `latency-tests.json`:
 In this example:

 - The `test_name` attributes is a unique identifier for the test. In `latency-tests.json`, it must start with `latency_`.
- The `parameters` attribute control the command line arguments to be used for `vllm bench latency`. Note that please use underline `_` instead of the dash `-` when specifying the command line arguments, and `run-performance-benchmarks.sh` will convert the underline to dash when feeding the arguments to `vllm bench latency`. For example, the corresponding command line arguments for `vllm bench latency` will be `--model meta-llama/Meta-Llama-3-8B --tensor-parallel-size 1 --load-format dummy --num-iters-warmup 5 --num-iters 15`
+- The `parameters` attribute control the command line arguments to be used for `benchmark_latency.py`. Note that please use underline `_` instead of the dash `-` when specifying the command line arguments, and `run-performance-benchmarks.sh` will convert the underline to dash when feeding the arguments to `benchmark_latency.py`. For example, the corresponding command line arguments for `benchmark_latency.py` will be `--model meta-llama/Meta-Llama-3-8B --tensor-parallel-size 1 --load-format dummy --num-iters-warmup 5 --num-iters 15`

 Note that the performance numbers are highly sensitive to the value of the parameters. Please make sure the parameters are set correctly.

@ -86,13 +68,13 @@ WARNING: The benchmarking script will save json results by itself, so please do

 ### Throughput test

-The tests are specified in `throughput-tests.json`. The syntax is similar to `latency-tests.json`, except for that the parameters will be fed forward to `vllm bench throughput`.
+The tests are specified in `throughput-tests.json`. The syntax is similar to `latency-tests.json`, except for that the parameters will be fed forward to `benchmark_throughput.py`.

 The number of this test is also stable -- a slight change on the value of this number might vary the performance numbers by a lot.

 ### Serving test

-We test the throughput by using `vllm bench serve` with request rate = inf to cover the online serving overhead. The corresponding parameters are in `serving-tests.json`, and here is an example:
+We test the throughput by using `benchmark_serving.py` with request rate = inf to cover the online serving overhead. The corresponding parameters are in `serving-tests.json`, and here is an example:

 ```json
 [
@ -104,6 +86,7 @@ We test the throughput by using `vllm bench serve` with request rate = inf to co
            "tensor_parallel_size": 1,
            "swap_space": 16,
            "disable_log_stats": "",
+            "disable_log_requests": "",
            "load_format": "dummy"
        },
        "client_parameters": {
@ -121,8 +104,8 @@ Inside this example:

 - The `test_name` attribute is also a unique identifier for the test. It must start with `serving_`.
 - The `server-parameters` includes the command line arguments for vLLM server.
- The `client-parameters` includes the command line arguments for `vllm bench serve`.
- The `qps_list` controls the list of qps for test. It will be used to configure the `--request-rate` parameter in `vllm bench serve`
+- The `client-parameters` includes the command line arguments for `benchmark_serving.py`.
+- The `qps_list` controls the list of qps for test. It will be used to configure the `--request-rate` parameter in `benchmark_serving.py`

 The number of this test is less stable compared to the delay and latency benchmarks (due to randomized sharegpt dataset sampling inside `benchmark_serving.py`), but a large change on this number (e.g. 5% change) still vary the output greatly.

@ -136,23 +119,6 @@ If you do not see the table, please wait till the benchmark finish running.
 The json version of the table (together with the json version of the benchmark) will be also attached to the markdown file.
 The raw benchmarking results (in the format of json files) are in the `Artifacts` tab of the benchmarking.

-The `compare-json-results.py` helps to compare benchmark results JSON files converted using `convert-results-json-to-markdown.py`.
-When run, benchmark script generates results under `benchmark/results` folder, along with the `benchmark_results.md` and `benchmark_results.json`.
-`compare-json-results.py` compares two `benchmark_results.json` files and provides performance ratio e.g. for Output Tput, Median TTFT and Median TPOT.  
-If only one benchmark_results.json is passed, `compare-json-results.py` compares different TP and PP configurations in the benchmark_results.json instead.
-
-Here is an example using the script to compare result_a and result_b with Model, Dataset name, input/output length, max concurrency and qps.
-`python3 compare-json-results.py -f results_a/benchmark_results.json -f results_b/benchmark_results.json`
-
-|   | Model | Dataset Name | Input Len | Output Len | # of max concurrency | qps  | results_a/benchmark_results.json | results_b/benchmark_results.json | perf_ratio        |
-|----|---------------------------------------|--------|-----|-----|------|-----|-----------|----------|----------|
-| 0  | meta-llama/Meta-Llama-3.1-8B-Instruct | random | 128 | 128 | 1000 | 1 | 142.633982                             | 156.526018                             | 1.097396 |
-| 1  | meta-llama/Meta-Llama-3.1-8B-Instruct | random | 128 | 128 | 1000 | inf| 241.620334                             | 294.018783                             | 1.216863 |
-
-A comparison diagram will be generated below the table.
-Here is an example to compare between 96c/results_gnr_96c_091_tp2pp3 and 128c/results_gnr_128c_091_tp2pp3
-<img width="1886" height="828" alt="image" src="https://github.com/user-attachments/assets/c02a43ef-25d0-4fd6-90e5-2169a28682dd" />
-
 ## Nightly test details

 See [nightly-descriptions.md](nightly-descriptions.md) for the detailed description on test workload, models and docker containers of benchmarking other llm engines.
@ -160,9 +126,9 @@ See [nightly-descriptions.md](nightly-descriptions.md) for the detailed descript
 ### Workflow

 - The [nightly-pipeline.yaml](nightly-pipeline.yaml) specifies the docker containers for different LLM serving engines.
- Inside each container, we run [scripts/run-nightly-benchmarks.sh](scripts/run-nightly-benchmarks.sh), which will probe the serving engine of the current container.
- The `scripts/run-nightly-benchmarks.sh` will parse the workload described in [nightly-tests.json](tests/nightly-tests.json) and launch the right benchmark for the specified serving engine via `scripts/launch-server.sh`.
- At last, we run [scripts/summary-nightly-results.py](scripts/summary-nightly-results.py) to collect and plot the final benchmarking results, and update the results to buildkite.
+- Inside each container, we run [run-nightly-suite.sh](run-nightly-suite.sh), which will probe the serving engine of the current container.
+- The `run-nightly-suite.sh` will redirect the request to `tests/run-[llm serving engine name]-nightly.sh`, which parses the workload described in [nightly-tests.json](tests/nightly-tests.json) and performs the benchmark.
+- At last, we run [scripts/plot-nightly-results.py](scripts/plot-nightly-results.py) to collect and plot the final benchmarking results, and update the results to buildkite.

 ### Nightly tests

@ -172,6 +138,6 @@ In [nightly-tests.json](tests/nightly-tests.json), we include the command line a

 The docker containers for benchmarking are specified in `nightly-pipeline.yaml`.

-WARNING: the docker versions are HARD-CODED and SHOULD BE ALIGNED WITH `nightly-descriptions.md`. The docker versions need to be hard-coded as there are several version-specific bug fixes inside `scripts/run-nightly-benchmarks.sh` and `scripts/launch-server.sh`.
+WARNING: the docker versions are HARD-CODED and SHOULD BE ALIGNED WITH `nightly-descriptions.md`. The docker versions need to be hard-coded as there are several version-specific bug fixes inside `tests/run-[llm serving engine name]-nightly.sh`.

 WARNING: populating `trt-llm` to latest version is not easy, as it requires updating several protobuf files in [tensorrt-demo](https://github.com/neuralmagic/tensorrt-demo.git).
--- a/.buildkite/nightly-benchmarks/nightly-annotation.md
+++ b/.buildkite/nightly-benchmarks/nightly-annotation.md
@ -1,4 +1,3 @@
-# Nightly benchmark annotation

 ## Description

@ -14,15 +13,15 @@ Please download the visualization scripts in the post

 - Find the docker we use in `benchmarking pipeline`
 - Deploy the docker, and inside the docker:
-    - Download `nightly-benchmarks.zip`.
-    - In the same folder, run the following code:
+  - Download `nightly-benchmarks.zip`.
+  - In the same folder, run the following code:

-    ```bash
-    export HF_TOKEN=<your HF token>
-    apt update
-    apt install -y git
-    unzip nightly-benchmarks.zip
-    VLLM_SOURCE_CODE_LOC=./ bash .buildkite/nightly-benchmarks/scripts/run-nightly-benchmarks.sh
-    ```
+  ```console
+  export HF_TOKEN=<your HF token>
+  apt update
+  apt install -y git
+  unzip nightly-benchmarks.zip
+  VLLM_SOURCE_CODE_LOC=./ bash .buildkite/nightly-benchmarks/scripts/run-nightly-benchmarks.sh
+  ```

 And the results will be inside `./benchmarks/results`.
--- a/.buildkite/nightly-benchmarks/nightly-descriptions.md
+++ b/.buildkite/nightly-benchmarks/nightly-descriptions.md
@ -13,25 +13,25 @@ Latest reproduction guilde: [github issue link](https://github.com/vllm-project/
 ## Setup

 - Docker images:
-    - vLLM: `vllm/vllm-openai:v0.6.2`
-    - SGLang: `lmsysorg/sglang:v0.3.2-cu121`
-    - LMDeploy: `openmmlab/lmdeploy:v0.6.1-cu12`
-    - TensorRT-LLM: `nvcr.io/nvidia/tritonserver:24.07-trtllm-python-py3`
-        - *NOTE: we use r24.07 as the current implementation only works for this version. We are going to bump this up.*
-    - Check [nightly-pipeline.yaml](nightly-pipeline.yaml) for the concrete docker images, specs and commands we use for the benchmark.
+  - vLLM: `vllm/vllm-openai:v0.6.2`
+  - SGLang: `lmsysorg/sglang:v0.3.2-cu121`
+  - LMDeploy: `openmmlab/lmdeploy:v0.6.1-cu12`
+  - TensorRT-LLM: `nvcr.io/nvidia/tritonserver:24.07-trtllm-python-py3`
+    - *NOTE: we uses r24.07 as the current implementation only works for this version. We are going to bump this up.*
+  - Check [nightly-pipeline.yaml](nightly-pipeline.yaml) for the concrete docker images, specs and commands we use for the benchmark.
 - Hardware
-    - 8x Nvidia A100 GPUs
+  - 8x Nvidia A100 GPUs
 - Workload:
-    - Dataset
-        - ShareGPT dataset
-        - Prefill-heavy dataset (in average 462 input tokens, 16 tokens as output)
-        - Decode-heavy dataset (in average 462 input tokens, 256 output tokens)
-        - Check [nightly-tests.json](tests/nightly-tests.json) for the concrete configuration of datasets we use.
-    - Models: llama-3 8B, llama-3 70B.
-        - We do not use llama 3.1 as it is incompatible with trt-llm r24.07. ([issue](https://github.com/NVIDIA/TensorRT-LLM/issues/2105)).
-    - Average QPS (query per second): 2, 4, 8, 16, 32 and inf.
-        - Queries are randomly sampled, and arrival patterns are determined via Poisson process, but all with fixed random seed.
-    - Evaluation metrics: Throughput (higher the better), TTFT (time to the first token, lower the better), ITL (inter-token latency, lower the better).
+  - Dataset
+    - ShareGPT dataset
+    - Prefill-heavy dataset (in average 462 input tokens, 16 tokens as output)
+    - Decode-heavy dataset (in average 462 input tokens, 256 output tokens)
+    - Check [nightly-tests.json](tests/nightly-tests.json) for the concrete configuration of datasets we use.
+  - Models: llama-3 8B, llama-3 70B.
+    - We do not use llama 3.1 as it is incompatible with trt-llm r24.07. ([issue](https://github.com/NVIDIA/TensorRT-LLM/issues/2105)).
+  - Average QPS (query per second): 2, 4, 8, 16, 32 and inf.
+    - Queries are randomly sampled, and arrival patterns are determined via Poisson process, but all with fixed random seed.
+  - Evaluation metrics: Throughput (higher the better), TTFT (time to the first token, lower the better), ITL (inter-token latency, lower the better).

 ## Known issues

--- a/.buildkite/nightly-benchmarks/performance-benchmarks-descriptions.md
+++ b/.buildkite/nightly-benchmarks/performance-benchmarks-descriptions.md
@ -1,12 +1,10 @@
-# Performance benchmarks descriptions

 ## Latency tests

 - Input length: 32 tokens.
 - Output length: 128 tokens.
 - Batch size: fixed (8).
- GPU Models: llama-3.1 8B, llama-3 70B, mixtral 8x7B.
- CPU Models: llama-3.1 8B.
+- Models: llama-3.1 8B, llama-3 70B, mixtral 8x7B.
 - Evaluation metrics: end-to-end latency (mean, median, p99).

 {latency_tests_markdown_table}
@ -16,8 +14,7 @@
 - Input length: randomly sample 200 prompts from ShareGPT dataset (with fixed random seed).
 - Output length: the corresponding output length of these 200 prompts.
 - Batch size: dynamically determined by vllm to achieve maximum throughput.
- GPU Models: llama-3.1 8B, llama-3 70B, mixtral 8x7B.
- CPU Models: llama-3.1 8B.
+- Models: llama-3.1 8B, llama-3 70B, mixtral 8x7B.
 - Evaluation metrics: throughput.

 {throughput_tests_markdown_table}
@ -28,18 +25,12 @@
 - Output length: the corresponding output length of these 200 prompts.
 - Batch size: dynamically determined by vllm and the arrival pattern of the requests.
 - **Average QPS (query per second)**: 1, 4, 16 and inf. QPS = inf means all requests come at once. For other QPS values, the arrival time of each query is determined using a random Poisson process (with fixed random seed).
- GPU Models: llama-3.1 8B, llama-3 70B, mixtral 8x7B.
- We also added a speculative decoding test for llama-3 70B on GPU, under QPS 2
- CPU Models: llama-3.1 8B.
+- Models: llama-3.1 8B, llama-3 70B, mixtral 8x7B.
+- We also added a speculative decoding test for llama-3 70B, under QPS 2
 - Evaluation metrics: throughput, TTFT (time to the first token, with mean, median and p99), ITL (inter-token latency, with mean, median and p99).
- For CPU, we added random dataset tests to benchmark fixed input/output length with 100 prompts.

 {serving_tests_markdown_table}

-## Platform Information
-
-{platform_markdown_table}
-
 ## json version of the benchmarking tables

 This section contains the data of the markdown tables above in JSON format.
--- a/.buildkite/nightly-benchmarks/scripts/compare-json-results.py
+++ b/.buildkite/nightly-benchmarks/scripts/compare-json-results.py
@ -1,307 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-import argparse
-import json
-import os
-from importlib import util
-
-import pandas as pd
-
-plotly_found = util.find_spec("plotly.express") is not None
-
-
-def compare_data_columns(
-    files, name_column, data_column, info_cols, drop_column, debug=False
-):
-    """
-    Align concatenation by keys derived from info_cols instead of row order.
-    - Pick one canonical key list: subset of info_cols present in ALL files.
-    - For each file: set index to those keys, aggregate duplicates
-    - (mean for metric, first for names).
-    - Concat along axis=1 (indexes align), then reset_index so callers can
-    - group by columns.
-    - If --debug, add a <file_label>_name column per file.
-    """
-    print("\ncompare_data_column:", data_column)
-
-    frames = []
-    raw_data_cols = []
-    compare_frames = []
-
-    # 1) choose a canonical key list from info_cols that exists in ALL files
-    cols_per_file = []
-    for f in files:
-        try:
-            df_tmp = pd.read_json(f, orient="records")
-        except Exception as err:
-            raise ValueError(f"Failed to read {f}") from err
-        cols_per_file.append(set(df_tmp.columns))
-
-    key_cols = [c for c in info_cols if all(c in cset for cset in cols_per_file)]
-    if not key_cols:
-        # soft fallback: use any info_cols present in the first file
-        key_cols = [c for c in info_cols if c in list(cols_per_file[0])]
-    if not key_cols:
-        raise ValueError(
-            "No common key columns found from info_cols across the input files."
-        )
-
-    # 2) build a single "meta" block (keys as columns) once, aligned by the key index
-    meta_added = False
-
-    for file in files:
-        df = pd.read_json(file, orient="records")
-
-        # Keep rows that actually have the compared metric (same as original behavior)
-        if drop_column in df.columns:
-            df = df.dropna(subset=[drop_column], ignore_index=True)
-
-        # Stabilize numeric key columns (harmless if missing)
-        for c in (
-            "Input Len",
-            "Output Len",
-            "TP Size",
-            "PP Size",
-            "# of max concurrency.",
-            "qps",
-        ):
-            if c in df.columns:
-                df[c] = pd.to_numeric(df[c], errors="coerce")
-
-        # Ensure all key columns exist
-        for c in key_cols:
-            if c not in df.columns:
-                df[c] = pd.NA
-
-        # Set index = key_cols and aggregate duplicates → unique MultiIndex
-        df_idx = df.set_index(key_cols, drop=False)
-
-        # meta (key columns), unique per key
-        meta = df_idx[key_cols]
-        if not meta.index.is_unique:
-            meta = meta.groupby(level=key_cols, dropna=False).first()
-
-        # metric series for this file, aggregated to one row per key
-        file_label = "/".join(file.split("/")[:-1]) or os.path.basename(file)
-        s = df_idx[data_column]
-        if not s.index.is_unique:
-            s = s.groupby(level=key_cols, dropna=False).mean()
-        s.name = file_label  # column label like original
-
-        # add meta once (from first file) so keys are the leftmost columns
-        if not meta_added:
-            frames.append(meta)
-            meta_added = True
-
-        # (NEW) debug: aligned test-name column per file
-        if debug and name_column in df_idx.columns:
-            name_s = df_idx[name_column]
-            if not name_s.index.is_unique:
-                name_s = name_s.groupby(level=key_cols, dropna=False).first()
-            name_s.name = f"{file_label}_name"
-            frames.append(name_s)
-
-        frames.append(s)
-        raw_data_cols.append(file_label)
-        compare_frames.append(s)
-
-        # Generalize ratio: for any file N>=2, add ratio (fileN / file1)
-        if len(compare_frames) >= 2:
-            base = compare_frames[0]
-            current = compare_frames[-1]
-            ratio = current / base
-            ratio = ratio.mask(base == 0)  # avoid inf when baseline is 0
-            ratio.name = f"Ratio 1 vs {len(compare_frames)}"
-            frames.append(ratio)
-
-    # 4) concat on columns with aligned MultiIndex;
-    # then reset_index to return keys as columns
-    concat_df = pd.concat(frames, axis=1)
-    concat_df = concat_df.reset_index(drop=True).reset_index()
-    if "index" in concat_df.columns:
-        concat_df = concat_df.drop(columns=["index"])
-
-    # Ensure key/info columns appear first (in your info_cols order)
-    front = [c for c in info_cols if c in concat_df.columns]
-    rest = [c for c in concat_df.columns if c not in front]
-    concat_df = concat_df[front + rest]
-
-    print(raw_data_cols)
-    return concat_df, raw_data_cols
-
-
-def split_json_by_tp_pp(
-    input_file: str = "benchmark_results.json", output_root: str = "."
-) -> list[str]:
-    """
-    Split a benchmark JSON into separate folders by (TP Size, PP Size).
-
-    Creates: <output_root>/tp{TP}_pp{PP}/benchmark_results.json
-    Returns: list of file paths written.
-    """
-    # Load JSON data into DataFrame
-    with open(input_file, encoding="utf-8") as f:
-        data = json.load(f)
-
-    # If the JSON is a dict with a list under common keys, use that list
-    if isinstance(data, dict):
-        for key in ("results", "serving_results", "benchmarks", "data"):
-            if isinstance(data.get(key), list):
-                data = data[key]
-                break
-
-    df = pd.DataFrame(data)
-
-    # Keep only "serving" tests
-    name_col = next(
-        (c for c in ["Test name", "test_name", "Test Name"] if c in df.columns), None
-    )
-    if name_col:
-        df = df[
-            df[name_col].astype(str).str.contains(r"serving", case=False, na=False)
-        ].copy()
-
-    # Handle alias column names
-    rename_map = {
-        "tp_size": "TP Size",
-        "tensor_parallel_size": "TP Size",
-        "pp_size": "PP Size",
-        "pipeline_parallel_size": "PP Size",
-    }
-    df.rename(
-        columns={k: v for k, v in rename_map.items() if k in df.columns}, inplace=True
-    )
-
-    # Ensure TP/PP columns exist (default to 1 if missing)
-    if "TP Size" not in df.columns:
-        df["TP Size"] = 1
-    if "PP Size" not in df.columns:
-        df["PP Size"] = 1
-
-    # make sure TP/PP are numeric ints with no NaN
-    df["TP Size"] = (
-        pd.to_numeric(df.get("TP Size", 1), errors="coerce").fillna(1).astype(int)
-    )
-    df["PP Size"] = (
-        pd.to_numeric(df.get("PP Size", 1), errors="coerce").fillna(1).astype(int)
-    )
-
-    # Split into separate folders
-    saved_paths: list[str] = []
-    for (tp, pp), group_df in df.groupby(["TP Size", "PP Size"], dropna=False):
-        folder_name = os.path.join(output_root, f"tp{int(tp)}_pp{int(pp)}")
-        os.makedirs(folder_name, exist_ok=True)
-        filepath = os.path.join(folder_name, "benchmark_results.json")
-        group_df.to_json(filepath, orient="records", indent=2, force_ascii=False)
-        print(f"Saved: {filepath}")
-        saved_paths.append(filepath)
-
-    return saved_paths
-
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser()
-    parser.add_argument(
-        "-f", "--file", action="append", type=str, help="input file name"
-    )
-    parser.add_argument(
-        "--debug", action="store_true", help="show all information for debugging"
-    )
-    parser.add_argument(
-        "--plot",
-        action=argparse.BooleanOptionalAction,
-        default=True,
-        help="plot perf diagrams or not --no-plot --plot",
-    )
-    parser.add_argument(
-        "-x",
-        "--xaxis",
-        type=str,
-        default="# of max concurrency.",
-        help="column name to use as X Axis in comparision graph",
-    )
-    args = parser.parse_args()
-
-    drop_column = "P99"
-    name_column = "Test name"
-    info_cols = [
-        "Model",
-        "Dataset Name",
-        "Input Len",
-        "Output Len",
-        "TP Size",
-        "PP Size",
-        "# of max concurrency.",
-        "qps",
-    ]
-    data_cols_to_compare = ["Output Tput (tok/s)", "Median TTFT (ms)", "Median"]
-    html_msgs_for_data_cols = [
-        "Compare Output Tokens /n",
-        "Median TTFT /n",
-        "Median TPOT /n",
-    ]
-
-    if len(args.file) == 1:
-        files = split_json_by_tp_pp(args.file[0], output_root="splits")
-        info_cols = [c for c in info_cols if c not in ("TP Size", "PP Size")]
-    else:
-        files = args.file
-    print("comparing : " + ", ".join(files))
-    debug = args.debug
-    plot = args.plot
-    # For Plot feature, assign y axis from one of info_cols
-    y_axis_index = info_cols.index(args.xaxis) if args.xaxis in info_cols else 6
-    with open("perf_comparison.html", "w") as text_file:
-        for i in range(len(data_cols_to_compare)):
-            output_df, raw_data_cols = compare_data_columns(
-                files,
-                name_column,
-                data_cols_to_compare[i],
-                info_cols,
-                drop_column,
-                debug=debug,
-            )
-
-            # For Plot feature, insert y axis from one of info_cols
-            raw_data_cols.insert(0, info_cols[y_axis_index])
-
-            filtered_info_cols = info_cols[:-2]
-            existing_group_cols = [
-                c for c in filtered_info_cols if c in output_df.columns
-            ]
-            if not existing_group_cols:
-                raise ValueError(
-                    f"No valid group-by columns  "
-                    f"Expected subset: {filtered_info_cols}, "
-                    f"but DataFrame has: {list(output_df.columns)}"
-                )
-            output_df_sorted = output_df.sort_values(by=existing_group_cols)
-            output_groups = output_df_sorted.groupby(existing_group_cols, dropna=False)
-            for name, group in output_groups:
-                html = group.to_html()
-                text_file.write(html_msgs_for_data_cols[i])
-                text_file.write(html)
-
-                if plot and plotly_found:
-                    import plotly.express as px
-
-                    df = group[raw_data_cols]
-                    df_sorted = df.sort_values(by=info_cols[y_axis_index])
-                    # Melt DataFrame for plotting
-                    df_melted = df_sorted.melt(
-                        id_vars=info_cols[y_axis_index],
-                        var_name="Configuration",
-                        value_name=data_cols_to_compare[i],
-                    )
-                    title = data_cols_to_compare[i] + " vs " + info_cols[y_axis_index]
-                    # Create Plotly line chart
-                    fig = px.line(
-                        df_melted,
-                        x=info_cols[y_axis_index],
-                        y=data_cols_to_compare[i],
-                        color="Configuration",
-                        title=title,
-                        markers=True,
-                    )
-                    # Export to HTML
-                    text_file.write(fig.to_html(full_html=True, include_plotlyjs="cdn"))
--- a/.buildkite/nightly-benchmarks/scripts/convert-results-json-to-markdown.py
+++ b/.buildkite/nightly-benchmarks/scripts/convert-results-json-to-markdown.py
@ -1,19 +1,15 @@
 # SPDX-License-Identifier: Apache-2.0
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project

-import argparse
 import json
 import os
-import shlex
-from importlib import util
 from pathlib import Path
-from typing import Any

 import pandas as pd
-import psutil
-import regex as re
 from tabulate import tabulate

+results_folder = Path("results/")
+
 # latency results and the keys that will be printed into markdown
 latency_results = []
 latency_column_mapping = {
@ -33,39 +29,28 @@ throughput_results = []
 throughput_results_column_mapping = {
    "test_name": "Test name",
    "gpu_type": "GPU",
-    "num_requests": "# of req.",
-    "total_num_tokens": "Total # of tokens",
-    "elapsed_time": "Elapsed time (s)",
+    # "num_requests": "# of req.",
+    # "total_num_tokens": "Total # of tokens",
+    # "elapsed_time": "Elapsed time (s)",
    "requests_per_second": "Tput (req/s)",
-    "tokens_per_second": "Tput (tok/s)",
+    # "tokens_per_second": "Tput (tok/s)",
 }

 # serving results and the keys that will be printed into markdown
 serving_results = []
 serving_column_mapping = {
    "test_name": "Test name",
-    "model_id": "Model",
-    "dataset_name": "Dataset Name",
-    "input_len": "Input Len",
-    "output_len": "Output Len",
-    "tp_size": "TP Size",
-    "pp_size": "PP Size",
-    "dtype": "dtype",
    "gpu_type": "GPU",
-    "completed": "# of req.",
-    "qps": "qps",
-    "max_concurrency": "# of max concurrency.",
+    # "completed": "# of req.",
    "request_throughput": "Tput (req/s)",
-    "total_token_throughput": "Total Token Tput (tok/s)",
-    "output_throughput": "Output Tput (tok/s)",
-    # "total_input_tokens": "Total input tokens",
-    # "total_output_tokens": "Total output tokens",
+    # "input_throughput": "Input Tput (tok/s)",
+    # "output_throughput": "Output Tput (tok/s)",
    "mean_ttft_ms": "Mean TTFT (ms)",
    "median_ttft_ms": "Median TTFT (ms)",
    "p99_ttft_ms": "P99 TTFT (ms)",
-    "mean_tpot_ms": "Mean TPOT (ms)",
-    "median_tpot_ms": "Median",
-    "p99_tpot_ms": "P99",
+    # "mean_tpot_ms": "Mean TPOT (ms)",
+    # "median_tpot_ms": "Median",
+    # "p99_tpot_ms": "P99",
    "mean_itl_ms": "Mean ITL (ms)",
    "median_itl_ms": "Median ITL (ms)",
    "p99_itl_ms": "P99 ITL (ms)",
@ -90,125 +75,15 @@ def results_to_json(latency, throughput, serving):
    )


-def get_size_with_unit(bytes, suffix="B"):
-    """
-    Scale bytes to its proper format
-    e.g:
-        1253656 => '1.20MB'
-        1253656678 => '1.17GB'
-    """
-    factor = 1024
-    for unit in ["", "K", "M", "G", "T", "P"]:
-        if bytes < factor:
-            return f"{bytes:.2f}{unit}{suffix}"
-        bytes /= factor
-
-
-def _coerce(val: str) -> Any:
-    """Best-effort type coercion from string to Python types."""
-    low = val.lower()
-    if low == "null":
-        return None
-    if low == "true":
-        return True
-    if low == "false":
-        return False
-    # integers
-    if re.fullmatch(r"[+-]?\d+", val):
-        try:
-            return int(val)
-        except ValueError:
-            pass
-    # floats (keep 'inf'/'-inf'/'nan' as strings)
-    if re.fullmatch(r"[+-]?\d*\.\d+", val):
-        try:
-            return float(val)
-        except ValueError:
-            pass
-    return val
-
-
-def parse_client_command(cmd: str) -> dict[str, Any]:
-    """Parse the client_command shell string into {executable, script, args}."""
-    toks = shlex.split(cmd)
-    if len(toks) < 2:
-        raise ValueError("client_command must include an executable and a script")
-    executable, script = toks[0], toks[1]
-    args: dict[str, Any] = {}
-
-    i = 2
-    while i < len(toks):
-        t = toks[i]
-        if t.startswith("--"):
-            # --key=value or --key (value) or boolean flag
-            if "=" in t:
-                key, val = t.split("=", 1)
-                if key == "--metadata":
-                    md = {}
-                    if val:
-                        if "=" in val:
-                            k, v = val.split("=", 1)
-                            md[k] = _coerce(v)
-                        else:
-                            md[val] = True
-                    args[key] = md
-                else:
-                    args[key] = _coerce(val)
-                i += 1
-                continue
-
-            key = t
-
-            # Special: consume metadata k=v pairs until next --flag
-            if key == "--metadata":
-                i += 1
-                md = {}
-                while i < len(toks) and not toks[i].startswith("--"):
-                    pair = toks[i]
-                    if "=" in pair:
-                        k, v = pair.split("=", 1)
-                        md[k] = _coerce(v)
-                    else:
-                        md[pair] = True
-                    i += 1
-                args[key] = md
-                continue
-
-            # Standard: check if next token is a value (not a flag)
-            if i + 1 < len(toks) and not toks[i + 1].startswith("--"):
-                args[key] = _coerce(toks[i + 1])
-                i += 2
-            else:
-                # lone flag -> True
-                args[key] = True
-                i += 1
-        else:
-            # unexpected positional; skip
-            i += 1
-
-    return {"executable": executable, "script": script, "args": args}
-
-
 if __name__ == "__main__":
-    parser = argparse.ArgumentParser()
-    parser.add_argument(
-        "-r",
-        "--result",
-        type=str,
-        default="results",
-        help="Folder name for benchmark output results.",
-    )
-    args = parser.parse_args()
-    results_folder = Path(args.result)
-    if not results_folder.exists():
-        raise FileNotFoundError(f"results folder does not exist: {results_folder}")
    # collect results
    for test_file in results_folder.glob("*.json"):
        with open(test_file) as f:
            raw_result = json.loads(f.read())

        if "serving" in str(test_file):
-            # this result is generated via `vllm bench serve` command
+            # this result is generated via `benchmark_serving.py`
+
            # attach the benchmarking command to raw_result
            try:
                with open(test_file.with_suffix(".commands")) as f:
@ -216,50 +91,18 @@ if __name__ == "__main__":
            except OSError as e:
                print(e)
                continue
-            # Parse Server Command Arg
-            out: dict[str, Any] = {
-                "server_command": parse_client_command(command["server_command"])
-            }
-            parse_args = [
-                "--tensor-parallel-size",
-                "--pipeline-parallel-size",
-                "--dtype",
-            ]
-            col_mapping = ["tp_size", "pp_size", "dtype"]
-            for index, arg in enumerate(parse_args):
-                if arg in out["server_command"]["args"]:
-                    raw_result.update(
-                        {col_mapping[index]: out["server_command"]["args"][arg]}
-                    )

-            # Parse Client Command Arg
-            out: dict[str, Any] = {
-                "client_command": parse_client_command(command["client_command"])
-            }
-            parse_args = [
-                "--dataset-name",
-                "--random-input-len",
-                "--random-output-len",
-                "--request-rate",
-            ]
-            col_mapping = ["dataset_name", "input_len", "output_len", "qps"]
-
-            for index, arg in enumerate(parse_args):
-                if arg in out["client_command"]["args"]:
-                    raw_result.update(
-                        {col_mapping[index]: out["client_command"]["args"][arg]}
-                    )
-            # Add Server, Client command
            raw_result.update(command)

            # update the test name of this result
            raw_result.update({"test_name": test_file.stem})
+
            # add the result to raw_result
            serving_results.append(raw_result)
            continue

        elif "latency" in f.name:
-            # this result is generated via `vllm bench latency` command
+            # this result is generated via `benchmark_latency.py`

            # attach the benchmarking command to raw_result
            try:
@ -287,7 +130,7 @@ if __name__ == "__main__":
            continue

        elif "throughput" in f.name:
-            # this result is generated via `vllm bench throughput` command
+            # this result is generated via `benchmark_throughput.py`

            # attach the benchmarking command to raw_result
            try:
@ -312,27 +155,6 @@ if __name__ == "__main__":
    serving_results = pd.DataFrame.from_dict(serving_results)
    throughput_results = pd.DataFrame.from_dict(throughput_results)

-    svmem = psutil.virtual_memory()
-    platform_data = {
-        "Physical cores": [psutil.cpu_count(logical=False)],
-        "Total cores": [psutil.cpu_count(logical=True)],
-        "Total Memory": [get_size_with_unit(svmem.total)],
-    }
-
-    if util.find_spec("numa") is not None:
-        from numa import info
-
-        platform_data["Total NUMA nodes"] = [info.get_num_configured_nodes()]
-
-    if util.find_spec("cpuinfo") is not None:
-        from cpuinfo import get_cpu_info
-
-        platform_data["CPU Brand"] = [get_cpu_info()["brand_raw"]]
-
-    platform_results = pd.DataFrame.from_dict(
-        platform_data, orient="index", columns=["Platform Info"]
-    )
-
    raw_results_json = results_to_json(
        latency_results, throughput_results, serving_results
    )
@ -343,10 +165,7 @@ if __name__ == "__main__":
            columns=latency_column_mapping
        )
    if not serving_results.empty:
-        valid_columns = [
-            col for col in serving_column_mapping if col in serving_results.columns
-        ]
-        serving_results = serving_results[valid_columns].rename(
+        serving_results = serving_results[list(serving_column_mapping.keys())].rename(
            columns=serving_column_mapping
        )
    if not throughput_results.empty:
@ -381,14 +200,9 @@ if __name__ == "__main__":
    throughput_md_table = tabulate(
        throughput_results, headers="keys", tablefmt="pipe", showindex=False
    )
-    platform_md_table = tabulate(
-        platform_results, headers="keys", tablefmt="pipe", showindex=True
-    )

    # document the result
-    md_file = "benchmark_results.md"
-    json_file = "benchmark_results.json"
-    with open(results_folder / md_file, "w") as f:
+    with open(results_folder / "benchmark_results.md", "w") as f:
        results = read_markdown(
            "../.buildkite/nightly-benchmarks/"
            + "performance-benchmarks-descriptions.md"
@ -397,13 +211,12 @@ if __name__ == "__main__":
            latency_tests_markdown_table=latency_md_table,
            throughput_tests_markdown_table=throughput_md_table,
            serving_tests_markdown_table=serving_md_table,
-            platform_markdown_table=platform_md_table,
            benchmarking_results_in_json_string=processed_results_json,
        )
        f.write(results)

    # document benchmarking results in json
-    with open(results_folder / json_file, "w") as f:
+    with open(results_folder / "benchmark_results.json", "w") as f:
        results = (
            latency_results.to_dict(orient="records")
            + throughput_results.to_dict(orient="records")
--- a/.buildkite/nightly-benchmarks/scripts/run-nightly-benchmarks.sh
+++ b/.buildkite/nightly-benchmarks/scripts/run-nightly-benchmarks.sh
@ -73,7 +73,7 @@ get_current_llm_serving_engine() {
    echo "Container: vllm"
    # move to a completely irrelevant directory, to avoid import vllm from current folder
    export CURRENT_LLM_SERVING_ENGINE=vllm
-
+    
    return
  fi
 }
@ -95,14 +95,12 @@ json2args() {
 }

 kill_gpu_processes() {
-  pkill -f '[p]ython'
-  pkill -f '[p]ython3'
-  pkill -f '[t]ritonserver'
-  pkill -f '[p]t_main_thread'
-  pkill -f '[t]ext-generation'
-  pkill -f '[l]mdeploy'
-  # vLLM now names the process with VLLM prefix after https://github.com/vllm-project/vllm/pull/21445
-  pkill -f '[V]LLM'
+  pkill -f python
+  pkill -f python3
+  pkill -f tritonserver
+  pkill -f pt_main_thread
+  pkill -f text-generation
+  pkill -f lmdeploy

  while [ "$(nvidia-smi --query-gpu=memory.used --format=csv,noheader,nounits | head -n 1)" -ge 1000 ]; do
    sleep 1
@ -127,7 +125,7 @@ ensure_installed() {
 }

 run_serving_tests() {
-  # run serving tests using `vllm bench serve` command
+  # run serving tests using `benchmark_serving.py`
  # $1: a json file specifying serving test cases

  local serving_test_file
@ -227,7 +225,7 @@ run_serving_tests() {

      if [[ "$dataset_name" = "sharegpt" ]]; then

-        client_command="vllm bench serve \
+        client_command="python3 benchmark_serving.py \
          --backend $backend \
          --tokenizer /tokenizer_cache \
          --model $model \
@ -248,7 +246,7 @@ run_serving_tests() {
        sonnet_output_len=$(echo "$common_params" | jq -r '.sonnet_output_len')
        sonnet_prefix_len=$(echo "$common_params" | jq -r '.sonnet_prefix_len')

-        client_command="vllm bench serve \
+        client_command="python3 benchmark_serving.py \
          --backend $backend \
          --tokenizer /tokenizer_cache \
          --model $model \
@ -267,13 +265,13 @@ run_serving_tests() {
          $client_args"

      else
-
+  
        echo "The dataset name must be either 'sharegpt' or 'sonnet'. Got $dataset_name."
        exit 1

      fi

-
+        

      echo "Running test case $test_name with qps $qps"
      echo "Client command: $client_command"
@ -304,7 +302,7 @@ run_serving_tests() {
 }

 run_genai_perf_tests() {
-  # run genai-perf tests
+  # run genai-perf tests 

  # $1: a json file specifying genai-perf test cases
  local genai_perf_test_file
@ -313,14 +311,14 @@ run_genai_perf_tests() {
  # Iterate over genai-perf tests
  jq -c '.[]' "$genai_perf_test_file" | while read -r params; do
    # get the test name, and append the GPU type back to it.
-    test_name=$(echo "$params" | jq -r '.test_name')
-
+    test_name=$(echo "$params" | jq -r '.test_name')    
+    
    # if TEST_SELECTOR is set, only run the test cases that match the selector
    if [[ -n "$TEST_SELECTOR" ]] && [[ ! "$test_name" =~ $TEST_SELECTOR ]]; then
      echo "Skip test case $test_name."
      continue
    fi
-
+    
    # prepend the current serving engine to the test name
    test_name=${CURRENT_LLM_SERVING_ENGINE}_${test_name}

@ -371,10 +369,10 @@ run_genai_perf_tests() {
        qps=$num_prompts
        echo "now qps is $qps"
      fi
-
+    
      new_test_name=$test_name"_qps_"$qps
      backend=$CURRENT_LLM_SERVING_ENGINE
-
+      
      if [[ "$backend" == *"vllm"* ]]; then
        backend="vllm"
      fi
@ -382,7 +380,7 @@ run_genai_perf_tests() {
      client_command="genai-perf profile \
        -m $model \
        --service-kind openai \
-        --backend "$backend" \
+        --backend vllm \
        --endpoint-type chat \
        --streaming \
        --url localhost:$port \
@ -415,7 +413,7 @@ prepare_dataset() {
  do
    cat sonnet.txt >> sonnet_4x.txt
  done
-
+  
 }

 main() {
--- a/.buildkite/nightly-benchmarks/scripts/run-performance-benchmarks.sh
+++ b/.buildkite/nightly-benchmarks/scripts/run-performance-benchmarks.sh
@ -31,20 +31,6 @@ check_gpus() {
  echo "GPU type is $gpu_type"
 }

-check_cpus() {
-  # check the number of CPUs and NUMA Node and GPU type.
-  declare -g numa_count=$(lscpu | grep "NUMA node(s):" | awk '{print $3}')
-  if [[ $numa_count -gt 0 ]]; then
-    echo "NUMA found."
-    echo $numa_count
-  else
-    echo "Need at least 1 NUMA to run benchmarking."
-    exit 1
-  fi
-  declare -g gpu_type="cpu"
-  echo "GPU type is $gpu_type"
-}
-
 check_hf_token() {
  # check if HF_TOKEN is available and valid
  if [[ -z "$HF_TOKEN" ]]; then
@ -83,22 +69,6 @@ json2args() {
  echo "$args"
 }

-json2envs() {
-  # transforms the JSON string to environment variables.
-  # example:
-  # input: { "VLLM_CPU_KVCACHE_SPACE": 5 }
-  # output: VLLM_CPU_KVCACHE_SPACE=5
-  local json_string=$1
-  local args=$(
-    echo "$json_string" | jq -r '
-      to_entries |
-      map((.key ) + "=" + (.value | tostring)) |
-      join(" ")
-    '
-  )
-  echo "$args"
-}
-
 wait_for_server() {
  # wait for vllm server to start
  # return 1 if vllm server crashes
@ -126,8 +96,7 @@ kill_gpu_processes() {
  ps -aux
  lsof -t -i:8000 | xargs -r kill -9
  pgrep python3 | xargs -r kill -9
-  # vLLM now names the process with VLLM prefix after https://github.com/vllm-project/vllm/pull/21445
-  pgrep VLLM | xargs -r kill -9
+

  # wait until GPU memory usage smaller than 1GB
  if command -v nvidia-smi; then
@ -165,7 +134,7 @@ upload_to_buildkite() {
 }

 run_latency_tests() {
-  # run latency tests using `vllm bench latency` command
+  # run latency tests using `benchmark_latency.py`
  # $1: a json file specifying latency test cases

  local latency_test_file
@ -189,26 +158,15 @@ run_latency_tests() {
    # get arguments
    latency_params=$(echo "$params" | jq -r '.parameters')
    latency_args=$(json2args "$latency_params")
-    latency_environment_variables=$(echo "$params" | jq -r '.environment_variables')
-    latency_envs=$(json2envs "$latency_environment_variables")

    # check if there is enough GPU to run the test
    tp=$(echo "$latency_params" | jq -r '.tensor_parallel_size')
-    if [ "$ON_CPU" == "1" ]; then
-      pp=$(echo "$latency_params" | jq -r '.pipeline_parallel_size')
-      world_size=$(($tp*$pp))
-      if [[ $numa_count -lt $world_size  && -z "${REMOTE_HOST}" ]]; then
-        echo "Required world-size $world_size but only $numa_count NUMA nodes found. Skip testcase $test_name."
-        continue
-      fi
-    else
-      if [[ $gpu_count -lt $tp ]]; then
-        echo "Required tensor-parallel-size $tp but only $gpu_count GPU found. Skip testcase $test_name."
-        continue
-      fi
+    if [[ $gpu_count -lt $tp ]]; then
+      echo "Required tensor-parallel-size $tp but only $gpu_count GPU found. Skip testcase $test_name."
+      continue
    fi

-    latency_command=" $latency_envs vllm bench latency \
+    latency_command="python3 benchmark_latency.py \
      --output-json $RESULTS_FOLDER/${test_name}.json \
      $latency_args"

@ -234,7 +192,7 @@ run_latency_tests() {
 }

 run_throughput_tests() {
-  # run throughput tests using `vllm bench throughput`
+  # run throughput tests using `benchmark_throughput.py`
  # $1: a json file specifying throughput test cases

  local throughput_test_file
@ -258,26 +216,15 @@ run_throughput_tests() {
    # get arguments
    throughput_params=$(echo "$params" | jq -r '.parameters')
    throughput_args=$(json2args "$throughput_params")
-    throughput_environment_variables=$(echo "$params" | jq -r '.environment_variables')
-    throughput_envs=$(json2envs "$throughput_environment_variables")

    # check if there is enough GPU to run the test
    tp=$(echo "$throughput_params" | jq -r '.tensor_parallel_size')
-    if [ "$ON_CPU" == "1" ]; then
-      pp=$(echo "$throughput_params" | jq -r '.pipeline_parallel_size')
-      world_size=$(($tp*$pp))
-      if [[ $numa_count -lt $world_size  && -z "${REMOTE_HOST}" ]]; then
-        echo "Required world-size $world_size but only $numa_count NUMA nodes found. Skip testcase $test_name."
-        continue
-      fi
-    else
-      if [[ $gpu_count -lt $tp ]]; then
-        echo "Required tensor-parallel-size $tp but only $gpu_count GPU found. Skip testcase $test_name."
-        continue
-      fi
+    if [[ $gpu_count -lt $tp ]]; then
+      echo "Required tensor-parallel-size $tp but only $gpu_count GPU found. Skip testcase $test_name."
+      continue
    fi

-    throughput_command=" $throughput_envs vllm bench throughput \
+    throughput_command="python3 benchmark_throughput.py \
      --output-json $RESULTS_FOLDER/${test_name}.json \
      $throughput_args"

@ -302,7 +249,7 @@ run_throughput_tests() {
 }

 run_serving_tests() {
-  # run serving tests using `vllm bench serve` command
+  # run serving tests using `benchmark_serving.py`
  # $1: a json file specifying serving test cases

  local serving_test_file
@ -325,36 +272,18 @@ run_serving_tests() {

    # get client and server arguments
    server_params=$(echo "$params" | jq -r '.server_parameters')
-    server_envs=$(echo "$params" | jq -r '.server_environment_variables')
    client_params=$(echo "$params" | jq -r '.client_parameters')
    server_args=$(json2args "$server_params")
-    server_envs=$(json2envs "$server_envs")
    client_args=$(json2args "$client_params")
    qps_list=$(echo "$params" | jq -r '.qps_list')
    qps_list=$(echo "$qps_list" | jq -r '.[] | @sh')
    echo "Running over qps list $qps_list"
-    max_concurrency_list=$(echo "$params" | jq -r '.max_concurrency_list')
-    if [[ -z "$max_concurrency_list" || "$max_concurrency_list" == "null" ]]; then
-        num_prompts=$(echo "$client_params" | jq -r '.num_prompts')
-        max_concurrency_list="[$num_prompts]"
-    fi
-    max_concurrency_list=$(echo "$max_concurrency_list" | jq -r '.[] | @sh')
-    echo "Running over max concurrency list $max_concurrency_list"

-    # check if there is enough resources to run the test
+    # check if there is enough GPU to run the test
    tp=$(echo "$server_params" | jq -r '.tensor_parallel_size')
-    if [ "$ON_CPU" == "1" ]; then
-      pp=$(echo "$server_params" | jq -r '.pipeline_parallel_size')
-      world_size=$(($tp*$pp))
-      if [[ $numa_count -lt $world_size  && -z "${REMOTE_HOST}" ]]; then
-        echo "Required world-size $world_size but only $numa_count NUMA nodes found. Skip testcase $test_name."
-        continue
-      fi
-    else
-      if [[ $gpu_count -lt $tp ]]; then
-        echo "Required tensor-parallel-size $tp but only $gpu_count GPU found. Skip testcase $test_name."
-        continue
-      fi
+    if [[ $gpu_count -lt $tp ]]; then
+      echo "Required tensor-parallel-size $tp but only $gpu_count GPU found. Skip testcase $test_name."
+      continue
    fi

    # check if server model and client model is aligned
@ -365,33 +294,23 @@ run_serving_tests() {
      continue
    fi

-    server_command="$server_envs python3 \
+    server_command="python3 \
      -m vllm.entrypoints.openai.api_server \
      $server_args"

    # run the server
    echo "Running test case $test_name"
    echo "Server command: $server_command"
-    # support remote vllm server
-    client_remote_args=""
-    if [[ -z "${REMOTE_HOST}" ]]; then
-      bash -c "$server_command" &
-      server_pid=$!
-      # wait until the server is alive
-      if wait_for_server; then
-        echo ""
-        echo "vLLM server is up and running."
-      else
-        echo ""
-        echo "vLLM failed to start within the timeout period."
-      fi
+    bash -c "$server_command" &
+    server_pid=$!
+
+    # wait until the server is alive
+    if wait_for_server; then
+      echo ""
+      echo "vllm server is up and running."
    else
-      server_command="Using Remote Server $REMOTE_HOST $REMOTE_PORT"
-      if [[ ${REMOTE_PORT} ]]; then
-        client_remote_args=" --host=$REMOTE_HOST --port=$REMOTE_PORT "
-      else
-        client_remote_args=" --host=$REMOTE_HOST "
-      fi
+      echo ""
+      echo "vllm failed to start within the timeout period."
    fi

    # iterate over different QPS
@ -403,39 +322,35 @@ run_serving_tests() {
        echo "now qps is $qps"
      fi

-      # iterate over different max_concurrency
-      for max_concurrency in $max_concurrency_list; do
-        new_test_name=$test_name"_qps_"$qps"_concurrency_"$max_concurrency
-        echo " new test name $new_test_name"
-        # pass the tensor parallel size to the client so that it can be displayed
-        # on the benchmark dashboard
-        client_command="vllm bench serve \
-          --save-result \
-          --result-dir $RESULTS_FOLDER \
-          --result-filename ${new_test_name}.json \
-          --request-rate $qps \
-          --max-concurrency $max_concurrency \
-          --metadata "tensor_parallel_size=$tp" \
-          $client_args $client_remote_args "
+      new_test_name=$test_name"_qps_"$qps

-        echo "Running test case $test_name with qps $qps"
-        echo "Client command: $client_command"
+      # pass the tensor parallel size to the client so that it can be displayed
+      # on the benchmark dashboard
+      client_command="python3 benchmark_serving.py \
+        --save-result \
+        --result-dir $RESULTS_FOLDER \
+        --result-filename ${new_test_name}.json \
+        --request-rate $qps \
+        --metadata "tensor_parallel_size=$tp" \
+        $client_args"

-        bash -c "$client_command"
+      echo "Running test case $test_name with qps $qps"
+      echo "Client command: $client_command"

-        # record the benchmarking commands
-        jq_output=$(jq -n \
-          --arg server "$server_command" \
-          --arg client "$client_command" \
-          --arg gpu "$gpu_type" \
-          '{
-            server_command: $server,
-            client_command: $client,
-            gpu_type: $gpu
-          }')
-        echo "$jq_output" >"$RESULTS_FOLDER/${new_test_name}.commands"
+      bash -c "$client_command"
+
+      # record the benchmarking commands
+      jq_output=$(jq -n \
+        --arg server "$server_command" \
+        --arg client "$client_command" \
+        --arg gpu "$gpu_type" \
+        '{
+          server_command: $server,
+          client_command: $client,
+          gpu_type: $gpu
+        }')
+      echo "$jq_output" >"$RESULTS_FOLDER/${new_test_name}.commands"

-      done
    done

    # clean up
@ -445,14 +360,7 @@ run_serving_tests() {
 }

 main() {
-  local ARCH
-  ARCH=''
-  if [ "$ON_CPU" == "1" ];then
-     check_cpus
-     ARCH='-cpu'
-  else
-     check_gpus
-  fi
+  check_gpus
  check_hf_token

  # Set to v1 to run v1 benchmark
@ -465,7 +373,7 @@ main() {
  (which jq) || (apt-get update && apt-get -y install jq)
  (which lsof) || (apt-get update && apt-get install -y lsof)

-  # get the current IP address, required by `vllm bench serve` command
+  # get the current IP address, required by benchmark_serving.py
  export VLLM_HOST_IP=$(hostname -I | awk '{print $1}')
  # turn of the reporting of the status of each request, to clean up the terminal output
  export VLLM_LOGGING_LEVEL="WARNING"
@ -478,9 +386,9 @@ main() {
  QUICK_BENCHMARK_ROOT=../.buildkite/nightly-benchmarks/

  # benchmarking
-  run_serving_tests $QUICK_BENCHMARK_ROOT/tests/"${SERVING_JSON:-serving-tests$ARCH.json}"
-  run_latency_tests $QUICK_BENCHMARK_ROOT/tests/"${LATENCY_JSON:-latency-tests$ARCH.json}"
-  run_throughput_tests $QUICK_BENCHMARK_ROOT/tests/"${THROUGHPUT_JSON:-throughput-tests$ARCH.json}"
+  run_serving_tests $QUICK_BENCHMARK_ROOT/tests/serving-tests.json
+  run_latency_tests $QUICK_BENCHMARK_ROOT/tests/latency-tests.json
+  run_throughput_tests $QUICK_BENCHMARK_ROOT/tests/throughput-tests.json

  # postprocess benchmarking results
  pip install tabulate pandas
--- a/.buildkite/nightly-benchmarks/tests/genai-perf-tests.json
+++ b/.buildkite/nightly-benchmarks/tests/genai-perf-tests.json
@ -11,7 +11,9 @@
        },
        "vllm_server_parameters": {
            "disable_log_stats": "",
+            "disable_log_requests": "",
            "gpu_memory_utilization": 0.9,
+            "num_scheduler_steps": 10,
            "max_num_seqs": 512,
            "dtype": "bfloat16"
        },
--- a/.buildkite/nightly-benchmarks/tests/latency-tests-cpu.json
+++ b/.buildkite/nightly-benchmarks/tests/latency-tests-cpu.json
@ -1,30 +0,0 @@
-[
-    {
-        "test_name": "latency_llama8B_tp1",
-        "environment_variables": {
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 1,
-            "load_format": "dummy",
-            "num_iters_warmup": 5,
-            "num_iters": 15
-        }
-    },
-    {
-        "test_name": "latency_llama8B_tp4",
-        "environment_variables": {
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 4,
-            "load_format": "dummy",
-            "num_iters_warmup": 5,
-            "num_iters": 15
-        }
-    }
-]
--- a/.buildkite/nightly-benchmarks/tests/nightly-tests.json
+++ b/.buildkite/nightly-benchmarks/tests/nightly-tests.json
@ -35,7 +35,9 @@
        }, 
        "vllm_server_parameters": {
            "disable_log_stats": "",
+            "disable_log_requests": "",
            "gpu_memory_utilization": 0.9,
+            "num_scheduler_steps": 10,
            "max_num_seqs": 512,
            "dtype": "bfloat16"
        },
@ -88,7 +90,9 @@
        }, 
        "vllm_server_parameters": {
            "disable_log_stats": "",
+            "disable_log_requests": "",
            "gpu_memory_utilization": 0.9,
+            "num_scheduler_steps": 10,
            "max_num_seqs": 512,
            "dtype": "bfloat16"
        },
@ -141,7 +145,9 @@
        }, 
        "vllm_server_parameters": {
            "disable_log_stats": "",
+            "disable_log_requests": "",
            "gpu_memory_utilization": 0.9,
+            "num_scheduler_steps": 10,
            "max_num_seqs": 512,
            "dtype": "bfloat16"
        },
@ -191,7 +197,9 @@
        }, 
        "vllm_server_parameters": {
            "disable_log_stats": "",
+            "disable_log_requests": "",
            "gpu_memory_utilization": 0.9,
+            "num_scheduler_steps": 10,
            "max_num_seqs": 512,
            "dtype": "bfloat16"
        },
@ -243,7 +251,9 @@
        }, 
        "vllm_server_parameters": {
            "disable_log_stats": "",
+            "disable_log_requests": "",
            "gpu_memory_utilization": 0.9,
+            "num_scheduler_steps": 10,
            "max_num_seqs": 512,
            "dtype": "bfloat16"
        },
@ -295,7 +305,9 @@
        }, 
        "vllm_server_parameters": {
            "disable_log_stats": "",
+            "disable_log_requests": "",
            "gpu_memory_utilization": 0.9,
+            "num_scheduler_steps": 10,
            "max_num_seqs": 512,
            "dtype": "bfloat16"
        },
--- a/.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc2.json
+++ b/.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc2.json
@ -1,202 +0,0 @@
-[
-    {
-        "test_name": "serving_llama8B_tp1_sharegpt",
-        "qps_list": ["inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 1,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200
-        }
-    },
-    {
-        "test_name": "serving_llama8B_tp2_sharegpt",
-        "qps_list": ["inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 2,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200
-        }
-    },
-    {
-        "test_name": "serving_llama8B_tp4_sharegpt",
-        "qps_list": ["inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 4,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200
-        }
-    },
-    {
-        "test_name": "serving_llama8B_tp1_random_128_128",
-        "qps_list": ["inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 1,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-	    "enable_chunked_prefill": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "random",
-	    "random-input-len": 128,
-	    "random-output-len": 128,
-	    "ignore-eos": "",
-            "num_prompts": 1000
-        }
-    },
-    {
-        "test_name": "serving_llama8B_tp2_random_128_128",
-        "qps_list": ["inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 2,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-	    "enable_chunked_prefill": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "random",
-	    "random-input-len": 128,
-	    "random-output-len": 128,
-	    "ignore-eos": "",
-            "num_prompts": 1000
-        }
-    },
-    {
-        "test_name": "serving_llama8B_tp4_random_128_128",
-        "qps_list": ["inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 4,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-	    "enable_chunked_prefill": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "random",
-	    "random-input-len": 128,
-	    "random-output-len": 128,
-            "num_prompts": 1000
-        }
-    }
-]
--- a/.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc3.json
+++ b/.buildkite/nightly-benchmarks/tests/serving-tests-cpu-snc3.json
@ -1,205 +0,0 @@
-[
-    {
-        "test_name": "serving_llama8B_pp1_sharegpt",
-        "qps_list": ["inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "pipeline_parallel_size": 1,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200
-        }
-    },
-    {
-        "test_name": "serving_llama8B_pp3_sharegpt",
-        "qps_list": ["inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "pipeline_parallel_size": 3,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200
-        }
-    },
-    {
-        "test_name": "serving_llama8B_tp2pp3_sharegpt",
-        "qps_list": ["inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 2,
-            "pipeline_parallel_size": 3,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200
-        }
-    },
-    {
-        "test_name": "serving_llama8B_pp1_random_128_128",
-        "qps_list": ["inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "pipeline_parallel_size": 1,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-	    "enable_chunked_prefill": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "random",
-	    "random-input-len": 128,
-	    "random-output-len": 128,
-	    "ignore-eos": "",
-            "num_prompts": 1000
-        }
-    },
-    {
-        "test_name": "serving_llama8B_pp3_random_128_128",
-        "qps_list": ["inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "pipeline_parallel_size": 3,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-	    "enable_chunked_prefill": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "random",
-	    "random-input-len": 128,
-	    "random-output-len": 128,
-	    "ignore-eos": "",
-            "num_prompts": 1000
-        }
-    },
-    {
-        "test_name": "serving_llama8B_tp2pp3_random_128_128",
-        "qps_list": ["inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200, 1000],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 2,
-            "pipeline_parallel_size": 3,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-	    "enable_chunked_prefill": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "random",
-	    "random-input-len": 128,
-	    "random-output-len": 128,
-	    "ignore-eos": "",
-            "num_prompts": 1000
-        }
-    }
-]
--- a/.buildkite/nightly-benchmarks/tests/serving-tests-cpu.json
+++ b/.buildkite/nightly-benchmarks/tests/serving-tests-cpu.json
@ -1,168 +0,0 @@
-[
-    {
-        "test_name": "serving_llama8B_tp1_sharegpt",
-        "qps_list": [1, 4, 16, "inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 1,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200
-        }
-    },
-    {
-        "test_name": "serving_llama8B_tp2_sharegpt",
-        "qps_list": [1, 4, 16, "inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 2,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200
-        }
-    },
-    {
-        "test_name": "serving_llama8B_tp4_sharegpt",
-        "qps_list": [1, 4, 16, "inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 4,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "sharegpt",
-            "dataset_path": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200
-        }
-    },
-    {
-        "test_name": "serving_llama8B_tp4_random_1024_128",
-        "qps_list": [1, 4, 16, "inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 4,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-	    "enable_chunked_prefill": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "random",
-	    "random-input-len": 1024,
-	    "random-output-len": 128,
-	    "ignore-eos": "",
-            "num_prompts": 100
-        }
-    },
-    {
-        "test_name": "serving_llama8B_pp6_random_1024_128",
-        "qps_list": [1, 4, 16, "inf"],
-        "max_concurrency_list": [12, 16, 24, 32, 64, 128, 200],
-        "server_environment_variables": {
-            "VLLM_RPC_TIMEOUT": 100000,
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_ENGINE_ITERATION_TIMEOUT_S": 120,
-	    "VLLM_CPU_SGL_KERNEL": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "server_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "pipeline_parallel_size": 6,
-	    "dtype": "bfloat16",
-	    "distributed_executor_backend": "mp",
-	    "block_size": 128,
-	    "trust_remote_code": "",
-	    "enable_chunked_prefill": "",
-            "disable_log_stats": "",
-	    "enforce_eager": "",
-	    "max_num_batched_tokens": 2048,
-	    "max_num_seqs": 256,
-            "load_format": "dummy"
-        },
-        "client_parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "backend": "vllm",
-            "dataset_name": "random",
-	    "random-input-len": 1024,
-	    "random-output-len": 128,
-	    "ignore-eos": "",
-            "num_prompts": 100
-        }
-    }
-]
--- a/.buildkite/nightly-benchmarks/tests/serving-tests.json
+++ b/.buildkite/nightly-benchmarks/tests/serving-tests.json
@ -7,6 +7,7 @@
            "tensor_parallel_size": 1,
            "swap_space": 16,
            "disable_log_stats": "",
+            "disable_log_requests": "",
            "load_format": "dummy"
        },
        "client_parameters": {
@ -25,6 +26,7 @@
            "tensor_parallel_size": 4,
            "swap_space": 16,
            "disable_log_stats": "",
+            "disable_log_requests": "",
            "load_format": "dummy"
        },
        "client_parameters": {
@ -43,6 +45,7 @@
            "tensor_parallel_size": 2,
            "swap_space": 16,
            "disable_log_stats": "",
+            "disable_log_requests": "",
            "load_format": "dummy"
        },
        "client_parameters": {
@ -57,7 +60,8 @@
        "test_name": "serving_llama70B_tp4_sharegpt_specdecode",
        "qps_list": [2],
        "server_parameters": {
-            "model": "meta-llama/Meta-Llama-3.1-70B-Instruct", 
+            "model": "meta-llama/Meta-Llama-3.1-70B-Instruct",
+            "disable_log_requests": "", 
            "tensor_parallel_size": 4,
            "swap_space": 16,
            "speculative_config": {
--- a/.buildkite/nightly-benchmarks/tests/throughput-tests-cpu.json
+++ b/.buildkite/nightly-benchmarks/tests/throughput-tests-cpu.json
@ -1,32 +0,0 @@
-[
-    {
-        "test_name": "throughput_llama8B_tp1",
-        "environment_variables": {
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 1,
-            "load_format": "dummy",
-            "dataset": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200,
-            "backend": "vllm"
-        }
-    },
-    {
-        "test_name": "throughput_llama8B_tp4",
-        "environment_variables": {
-	    "VLLM_ALLOW_LONG_MAX_MODEL_LEN": 1,
-	    "VLLM_CPU_KVCACHE_SPACE": 40
-        },
-        "parameters": {
-            "model": "meta-llama/Llama-3.1-8B-Instruct",
-            "tensor_parallel_size": 4,
-            "load_format": "dummy",
-            "dataset": "./ShareGPT_V3_unfiltered_cleaned_split.json",
-            "num_prompts": 200,
-            "backend": "vllm"
-        }
-    }
-]
--- a/.buildkite/release-pipeline.yaml
+++ b/.buildkite/release-pipeline.yaml
@ -1,22 +1,5 @@
 steps:
-  # aarch64 + CUDA builds
-  - label: "Build arm64 wheel - CUDA 12.8"
-    id: build-wheel-arm64-cuda-12-8
-    agents:
-      queue: arm64_cpu_queue_postmerge
-    commands:
-      # #NOTE: torch_cuda_arch_list is derived from upstream PyTorch build files here:
-      # https://github.com/pytorch/pytorch/blob/main/.ci/aarch64_linux/aarch64_ci_build.sh#L7
-      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg GIT_REPO_CHECK=1 --build-arg CUDA_VERSION=12.8.1 --build-arg torch_cuda_arch_list='8.7 9.0 10.0+PTX 12.0' --tag vllm-ci:build-image --target build --progress plain -f docker/Dockerfile ."
-      - "mkdir artifacts"
-      - "docker run --rm -v $(pwd)/artifacts:/artifacts_host vllm-ci:build-image bash -c 'cp -r dist /artifacts_host && chmod -R a+rw /artifacts_host'"
-      - "bash .buildkite/scripts/upload-wheels.sh"
-    env:
-      DOCKER_BUILDKIT: "1"
-
-  # x86 + CUDA builds
  - label: "Build wheel - CUDA 12.8"
-    id: build-wheel-cuda-12-8
    agents:
      queue: cpu_queue_postmerge
    commands:
@ -27,13 +10,7 @@ steps:
    env:
      DOCKER_BUILDKIT: "1"

-  - block: "Build CUDA 12.6 wheel"
-    key: block-build-cu126-wheel
-    depends_on: ~
-
  - label: "Build wheel - CUDA 12.6"
-    depends_on: block-build-cu126-wheel
-    id: build-wheel-cuda-12-6
    agents:
      queue: cpu_queue_postmerge
    commands:
@ -51,7 +28,6 @@ steps:

  - label: "Build wheel - CUDA 11.8"
    # depends_on: block-build-cu118-wheel
-    id: build-wheel-cuda-11-8
    agents:
      queue: cpu_queue_postmerge
    commands:
@ -62,58 +38,19 @@ steps:
    env:
      DOCKER_BUILDKIT: "1"

-  - block: "Build release image (x86)"
+  - block: "Build release image"
    depends_on: ~
    key: block-release-image-build

-  - label: "Build release image (x86)"
+  - label: "Build release image"
    depends_on: block-release-image-build
-    id: build-release-image-x86
    agents:
      queue: cpu_queue_postmerge
    commands:
      - "aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7"
-      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg GIT_REPO_CHECK=1 --build-arg CUDA_VERSION=12.8.1 --build-arg FLASHINFER_AOT_COMPILE=true --build-arg INSTALL_KV_CONNECTORS=true --tag public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT-$(uname -m) --target vllm-openai --progress plain -f docker/Dockerfile ."
-      - "docker push public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT-$(uname -m)"
-      # re-tag to default image tag and push, just in case arm64 build fails
-      - "docker tag public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT-$(uname -m) public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT"
+      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg GIT_REPO_CHECK=1 --build-arg CUDA_VERSION=12.8.1 --tag public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT --target vllm-openai --progress plain -f docker/Dockerfile ."
      - "docker push public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT"

-  - label: "Build release image (arm64)"
-    depends_on: block-release-image-build
-    id: build-release-image-arm64
-    agents:
-      queue: arm64_cpu_queue_postmerge
-    commands:
-      - "aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7"
-      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg USE_SCCACHE=1 --build-arg GIT_REPO_CHECK=1 --build-arg CUDA_VERSION=12.8.1 --build-arg torch_cuda_arch_list='8.7 9.0 10.0+PTX 12.0' --build-arg INSTALL_KV_CONNECTORS=true --tag public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT-$(uname -m) --target vllm-openai --progress plain -f docker/Dockerfile ."
-      - "docker push public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT-$(uname -m)"
-
-  # Add job to create multi-arch manifest
-  - label: "Create multi-arch manifest"
-    depends_on:
-      - build-release-image-x86
-      - build-release-image-arm64
-    id: create-multi-arch-manifest
-    agents:
-      queue: cpu_queue_postmerge
-    commands:
-      - "aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7"
-      - "docker manifest create public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT-x86_64 public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT-aarch64 --amend"
-      - "docker manifest push public.ecr.aws/q9t5s3a7/vllm-release-repo:$BUILDKITE_COMMIT"
-
-  - label: "Annotate release workflow"
-    depends_on:
-      - create-multi-arch-manifest
-      - build-wheel-cuda-12-8
-      - build-wheel-cuda-12-6
-      - build-wheel-cuda-11-8
-    id: annotate-release-workflow
-    agents:
-      queue: cpu_queue_postmerge
-    commands:
-      - "bash .buildkite/scripts/annotate-release.sh"
-
  - label: "Build and publish TPU release image"
    depends_on: ~
    if: build.env("NIGHTLY") == "1"
@ -133,10 +70,9 @@ steps:
      DOCKER_BUILDKIT: "1"

  - input: "Provide Release version here"
-    id: input-release-version
    fields:
      - text: "What is the release version?"
-        key: release-version
+        key: "release-version"

  - block: "Build CPU release image"
    key: block-cpu-release-image-build
@ -148,8 +84,7 @@ steps:
      queue: cpu_queue_postmerge
    commands:
      - "aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7"
-      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg GIT_REPO_CHECK=1 --build-arg VLLM_CPU_AVX512BF16=true --build-arg VLLM_CPU_AVX512VNNI=true --tag public.ecr.aws/q9t5s3a7/vllm-cpu-release-repo:$(buildkite-agent meta-data get release-version) --tag public.ecr.aws/q9t5s3a7/vllm-cpu-release-repo:latest --progress plain --target vllm-openai -f docker/Dockerfile.cpu ."
-      - "docker push public.ecr.aws/q9t5s3a7/vllm-cpu-release-repo:latest"
+      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg GIT_REPO_CHECK=1 --tag public.ecr.aws/q9t5s3a7/vllm-cpu-release-repo:$(buildkite-agent meta-data get release-version) --tag public.ecr.aws/q9t5s3a7/vllm-cpu-release-repo:latest --progress plain --target vllm-openai -f docker/Dockerfile.cpu ."
      - "docker push public.ecr.aws/q9t5s3a7/vllm-cpu-release-repo:$(buildkite-agent meta-data get release-version)"
    env:
      DOCKER_BUILDKIT: "1"
@ -165,7 +100,6 @@ steps:
    commands:
      - "aws ecr-public get-login-password --region us-east-1 | docker login --username AWS --password-stdin public.ecr.aws/q9t5s3a7"
      - "DOCKER_BUILDKIT=1 docker build --build-arg max_jobs=16 --build-arg GIT_REPO_CHECK=1 --tag public.ecr.aws/q9t5s3a7/vllm-neuron-release-repo:$(buildkite-agent meta-data get release-version) --tag public.ecr.aws/q9t5s3a7/vllm-neuron-release-repo:latest --progress plain -f docker/Dockerfile.neuron ."
-      - "docker push public.ecr.aws/q9t5s3a7/vllm-neuron-release-repo:latest"
      - "docker push public.ecr.aws/q9t5s3a7/vllm-neuron-release-repo:$(buildkite-agent meta-data get release-version)"
    env:
      DOCKER_BUILDKIT: "1"
--- a/.buildkite/scripts/annotate-release.sh
+++ b/.buildkite/scripts/annotate-release.sh
@ -1,31 +0,0 @@
-#!/bin/bash
-
-set -ex
-
-# Get release version and strip leading 'v' if present
-RELEASE_VERSION=$(buildkite-agent meta-data get release-version | sed 's/^v//')
-
-if [ -z "$RELEASE_VERSION" ]; then
-  echo "Error: RELEASE_VERSION is empty. 'release-version' metadata might not be set or is invalid."
-  exit 1
-fi
-
-buildkite-agent annotate --style 'info' --context 'release-workflow' << EOF
-To download the wheel:
-\`\`\`
-aws s3 cp s3://vllm-wheels/${RELEASE_VERSION}/vllm-${RELEASE_VERSION}-cp38-abi3-manylinux1_x86_64.whl .
-aws s3 cp s3://vllm-wheels/${RELEASE_VERSION}+cu126/vllm-${RELEASE_VERSION}+cu126-cp38-abi3-manylinux1_x86_64.whl .
-aws s3 cp s3://vllm-wheels/${RELEASE_VERSION}+cu118/vllm-${RELEASE_VERSION}+cu118-cp38-abi3-manylinux1_x86_64.whl . 
-\`\`\`
-
-To download and upload the image:
-
-\`\`\`
-docker pull public.ecr.aws/q9t5s3a7/vllm-release-repo:${BUILDKITE_COMMIT}
-docker tag public.ecr.aws/q9t5s3a7/vllm-release-repo:${BUILDKITE_COMMIT} vllm/vllm-openai
-docker tag vllm/vllm-openai vllm/vllm-openai:latest
-docker tag vllm/vllm-openai vllm/vllm-openai:v${RELEASE_VERSION}
-docker push vllm/vllm-openai:latest
-docker push vllm/vllm-openai:v${RELEASE_VERSION}
-\`\`\`
-EOF 
--- a/.buildkite/scripts/ci-clean-log.sh
+++ b/.buildkite/scripts/ci-clean-log.sh
@ -1,17 +0,0 @@
-#!/bin/bash
-# Usage: ./ci_clean_log.sh ci.log
-# This script strips timestamps and color codes from CI log files.
-
-# Check if argument is given
-if [ $# -lt 1 ]; then
-    echo "Usage: $0 ci.log"
-    exit 1
-fi
-
-INPUT_FILE="$1"
-
-# Strip timestamps
-sed -i 's/^\[[0-9]\{4\}-[0-9]\{2\}-[0-9]\{2\}T[0-9]\{2\}:[0-9]\{2\}:[0-9]\{2\}Z\] //' "$INPUT_FILE"
-
-# Strip colorization
-sed -i -r 's/\x1B\[[0-9;]*[mK]//g' "$INPUT_FILE"
--- a/.buildkite/scripts/hardware_ci/run-amd-test.sh
+++ b/.buildkite/scripts/hardware_ci/run-amd-test.sh
@ -107,8 +107,10 @@ fi

 if [[ $commands == *" kernels/attention"* ]]; then
  commands="${commands} \
-  --ignore=kernels/attention/test_attention_selector.py \
+  --ignore=kernels/attention/stest_attention_selector.py \
+  --ignore=kernels/attention/test_blocksparse_attention.py \
  --ignore=kernels/attention/test_encoder_decoder_attn.py \
+  --ignore=kernels/attention/test_attention_selector.py \
  --ignore=kernels/attention/test_flash_attn.py \
  --ignore=kernels/attention/test_flashinfer.py \
  --ignore=kernels/attention/test_prefix_prefill.py \
@ -121,6 +123,7 @@ fi
 if [[ $commands == *" kernels/quantization"* ]]; then
  commands="${commands} \
  --ignore=kernels/quantization/test_int8_quant.py \
+  --ignore=kernels/quantization/test_aqlm.py \
  --ignore=kernels/quantization/test_machete_mm.py \
  --ignore=kernels/quantization/test_block_fp8.py \
  --ignore=kernels/quantization/test_block_int8.py \
--- a/.buildkite/scripts/hardware_ci/run-cpu-test-ppc64le.sh
+++ b/.buildkite/scripts/hardware_ci/run-cpu-test-ppc64le.sh
@ -7,7 +7,6 @@ set -ex
 # Setup cleanup
 remove_docker_container() {
  if [[ -n "$container_id" ]]; then
-      podman stop --all -t0
      podman rm -f "$container_id" || true
  fi
  podman system prune -f
@ -38,7 +37,7 @@ function cpu_tests() {
    pytest -v -s tests/models/language/generation/test_common.py::test_models[False-5-32-facebook/opt-125m]
    pytest -v -s tests/models/language/generation/test_common.py::test_models[False-5-32-google/gemma-1.1-2b-it]
    pytest -v -s tests/models/language/pooling/test_classification.py::test_models[float-jason9693/Qwen2.5-1.5B-apeach]
-    pytest -v -s tests/models/language/pooling/test_embedding.py -m cpu_model"
+    pytest -v -s tests/models/language/pooling/test_embedding.py::test_models[half-BAAI/bge-base-en-v1.5]"
 }

 # All of CPU tests are expected to be finished less than 40 mins.
--- a/.buildkite/scripts/hardware_ci/run-cpu-test.sh
+++ b/.buildkite/scripts/hardware_ci/run-cpu-test.sh
@ -6,16 +6,15 @@ set -ex

 # allow to bind to different cores
 CORE_RANGE=${CORE_RANGE:-48-95}
-# used for TP/PP E2E test
 OMP_CORE_RANGE=${OMP_CORE_RANGE:-48-95}
 NUMA_NODE=${NUMA_NODE:-1}

 export CMAKE_BUILD_PARALLEL_LEVEL=32

 # Setup cleanup
-remove_docker_container() {
-    set -e;
-    docker rm -f cpu-test-"$NUMA_NODE" cpu-test-"$NUMA_NODE"-avx2 || true;
+remove_docker_container() { 
+    set -e; 
+    docker rm -f cpu-test-"$NUMA_NODE" cpu-test-"$NUMA_NODE"-avx2 || true; 
 }
 trap remove_docker_container EXIT
 remove_docker_container
@ -25,83 +24,66 @@ numactl -C "$CORE_RANGE" -N "$NUMA_NODE" docker build --tag cpu-test-"$NUMA_NODE
 numactl -C "$CORE_RANGE" -N "$NUMA_NODE" docker build --build-arg VLLM_CPU_DISABLE_AVX512="true" --tag cpu-test-"$NUMA_NODE"-avx2 --target vllm-test -f docker/Dockerfile.cpu .

 # Run the image, setting --shm-size=4g for tensor parallel.
-docker run -itd --cpuset-cpus="$CORE_RANGE" --cpuset-mems="$NUMA_NODE" --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --privileged=true -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=4 --env VLLM_CPU_CI_ENV=1 -e E2E_OMP_THREADS="$OMP_CORE_RANGE" --shm-size=4g --name cpu-test-"$NUMA_NODE" cpu-test-"$NUMA_NODE"
-docker run -itd --cpuset-cpus="$CORE_RANGE" --cpuset-mems="$NUMA_NODE" --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --privileged=true -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=4 --env VLLM_CPU_CI_ENV=1 -e E2E_OMP_THREADS="$OMP_CORE_RANGE" --shm-size=4g --name cpu-test-"$NUMA_NODE"-avx2 cpu-test-"$NUMA_NODE"-avx2
+docker run -itd --cpuset-cpus="$CORE_RANGE" --cpuset-mems="$NUMA_NODE" --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --privileged=true -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=4 --env VLLM_CPU_OMP_THREADS_BIND="$OMP_CORE_RANGE" --shm-size=4g --name cpu-test-"$NUMA_NODE" cpu-test-"$NUMA_NODE"
+docker run -itd --cpuset-cpus="$CORE_RANGE" --cpuset-mems="$NUMA_NODE" --entrypoint /bin/bash -v ~/.cache/huggingface:/root/.cache/huggingface --privileged=true -e HF_TOKEN --env VLLM_CPU_KVCACHE_SPACE=4 --env VLLM_CPU_OMP_THREADS_BIND="$OMP_CORE_RANGE" --shm-size=4g --name cpu-test-"$NUMA_NODE"-avx2 cpu-test-"$NUMA_NODE"-avx2

 function cpu_tests() {
  set -e
  export NUMA_NODE=$2

-  # list packages
-  docker exec cpu-test-"$NUMA_NODE"-avx2 bash -c "
-    set -e
-    pip list"
-
-  docker exec cpu-test-"$NUMA_NODE" bash -c "
-    set -e
-    pip list"
-
  # offline inference
  docker exec cpu-test-"$NUMA_NODE"-avx2 bash -c "
    set -e
    python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m"

-  # Run kernel tests
-  docker exec cpu-test-"$NUMA_NODE" bash -c "
-    set -e
-    pytest -v -s tests/kernels/test_onednn.py"
-
  # Run basic model test
  docker exec cpu-test-"$NUMA_NODE" bash -c "
    set -e
-    # Note: disable until supports V1
-    # pytest -v -s tests/kernels/attention/test_cache.py -m cpu_model
-    # pytest -v -s tests/kernels/attention/test_mla_decode_cpu.py -m cpu_model
-
-    # Note: disable Bart until supports V1
-    pytest -v -s tests/models/language/generation -m cpu_model \
-                --ignore=tests/models/language/generation/test_bart.py
-    VLLM_CPU_SGL_KERNEL=1 pytest -v -s tests/models/language/generation -m cpu_model \
-                --ignore=tests/models/language/generation/test_bart.py
-
+    pytest -v -s tests/kernels/attention/test_cache.py -m cpu_model
+    pytest -v -s tests/kernels/attention/test_mla_decode_cpu.py -m cpu_model
+    pytest -v -s tests/models/language/generation -m cpu_model
    pytest -v -s tests/models/language/pooling -m cpu_model
-    pytest -v -s tests/models/multimodal/generation \
-                --ignore=tests/models/multimodal/generation/test_mllama.py \
-                --ignore=tests/models/multimodal/generation/test_pixtral.py \
-                -m cpu_model"
+    pytest -v -s tests/models/multimodal/generation --ignore=tests/models/multimodal/generation/test_mllama.py -m cpu_model"

  # Run compressed-tensor test
  docker exec cpu-test-"$NUMA_NODE" bash -c "
    set -e
    pytest -s -v \
-    tests/quantization/test_compressed_tensors.py::test_compressed_tensors_w8a8_logprobs[False-10-32-neuralmagic/Llama-3.2-1B-quantized.w8a8]"
+    tests/quantization/test_compressed_tensors.py::test_compressed_tensors_w8a8_static_setup \
+    tests/quantization/test_compressed_tensors.py::test_compressed_tensors_w8a8_dynamic_per_token"

-  # Note: disable it until supports V1
  # Run AWQ test
-  # docker exec cpu-test-"$NUMA_NODE" bash -c "
-  #   set -e
-  #   VLLM_USE_V1=0 pytest -s -v \
-  #   tests/quantization/test_ipex_quant.py"
+  docker exec cpu-test-"$NUMA_NODE" bash -c "
+    set -e
+    VLLM_USE_V1=0 pytest -s -v \
+    tests/quantization/test_ipex_quant.py"
+
+  # Run chunked-prefill and prefix-cache test
+  docker exec cpu-test-"$NUMA_NODE" bash -c "
+    set -e
+    pytest -s -v -k cpu_model \
+    tests/basic_correctness/test_chunked_prefill.py"  
+
+  # online serving
+  docker exec cpu-test-"$NUMA_NODE" bash -c "
+    set -e
+    python3 -m vllm.entrypoints.openai.api_server --model facebook/opt-125m --dtype half & 
+    timeout 600 bash -c 'until curl localhost:8000/v1/models; do sleep 1; done' || exit 1
+    python3 benchmarks/benchmark_serving.py \
+      --backend vllm \
+      --dataset-name random \
+      --model facebook/opt-125m \
+      --num-prompts 20 \
+      --endpoint /v1/completions \
+      --tokenizer facebook/opt-125m"

  # Run multi-lora tests
  docker exec cpu-test-"$NUMA_NODE" bash -c "
    set -e
    pytest -s -v \
    tests/lora/test_qwen2vl.py"
-
-  # online serving
-  docker exec cpu-test-"$NUMA_NODE" bash -c '
-    set -e
-    VLLM_CPU_OMP_THREADS_BIND=$E2E_OMP_THREADS VLLM_CPU_SGL_KERNEL=1 vllm serve meta-llama/Llama-3.2-3B-Instruct -tp=2 -pp=2 &
-    timeout 600 bash -c "until curl localhost:8000/v1/models; do sleep 1; done" || exit 1
-    vllm bench serve \
-      --backend vllm \
-      --dataset-name random \
-      --model meta-llama/Llama-3.2-3B-Instruct \
-      --num-prompts 20 \
-      --endpoint /v1/completions'
 }

 # All of CPU tests are expected to be finished less than 40 mins.
 export -f cpu_tests
-timeout 2h bash -c "cpu_tests $CORE_RANGE $NUMA_NODE"
+timeout 1h bash -c "cpu_tests $CORE_RANGE $NUMA_NODE"
--- a/.buildkite/scripts/hardware_ci/run-gh200-test.sh
+++ b/.buildkite/scripts/hardware_ci/run-gh200-test.sh
@ -16,7 +16,8 @@ DOCKER_BUILDKIT=1 docker build . \
  --build-arg max_jobs=66 \
  --build-arg nvcc_threads=2 \
  --build-arg RUN_WHEEL_CHECK=false \
-  --build-arg torch_cuda_arch_list="9.0+PTX"
+  --build-arg torch_cuda_arch_list="9.0+PTX" \
+  --build-arg vllm_fa_cmake_gpu_arches="90-real"

 # Setup cleanup
 remove_docker_container() { docker rm -f gh200-test || true; }
--- a/.buildkite/scripts/hardware_ci/run-hpu-test.sh
+++ b/.buildkite/scripts/hardware_ci/run-hpu-test.sh
@ -2,32 +2,10 @@

 # This script build the CPU docker image and run the offline inference inside the container.
 # It serves a sanity check for compilation and basic model usage.
-set -exuo pipefail
+set -ex

 # Try building the docker image
-cat <<EOF | docker build -t hpu-plugin-v1-test-env -f - .
-FROM gaudi-base-image:latest
-
-COPY ./ /workspace/vllm
-
-WORKDIR /workspace/vllm
-
-ENV no_proxy=localhost,127.0.0.1
-ENV PT_HPU_ENABLE_LAZY_COLLECTIVES=true
-
-RUN VLLM_TARGET_DEVICE=empty pip install .
-RUN pip install git+https://github.com/vllm-project/vllm-gaudi.git
-
-# install development dependencies (for testing)
-RUN python3 -m pip install -e tests/vllm_test_utils
-
-WORKDIR /workspace/
-
-RUN git clone https://github.com/vllm-project/vllm-gaudi.git
-
-RUN ln -s /workspace/vllm/tests && ln -s /workspace/vllm/examples && ln -s /workspace/vllm/benchmarks
-
-EOF
+docker build -t hpu-test-env -f docker/Dockerfile.hpu .

 # Setup cleanup
 # certain versions of HPU software stack have a bug that can
@ -36,21 +14,13 @@ EOF
 # functions, while other platforms only need one remove_docker_container
 # function.
 EXITCODE=1
-remove_docker_containers() { docker rm -f hpu-plugin-v1-test || true; }
-trap 'remove_docker_containers; exit $EXITCODE;' EXIT
+remove_docker_containers() { docker rm -f hpu-test || true; docker rm -f hpu-test-tp2 || true; }
+remove_docker_containers_and_exit() { remove_docker_containers; exit $EXITCODE; }
+trap remove_docker_containers_and_exit EXIT
 remove_docker_containers

-echo "Running HPU plugin v1 test"
-docker run --rm --runtime=habana --name=hpu-plugin-v1-test --network=host \
-  -e HABANA_VISIBLE_DEVICES=all \
-  hpu-plugin-v1-test-env \
-  /bin/bash "/workspace/vllm-gaudi/tests/upstream_tests/ci_tests.sh"
+# Run the image and launch offline inference
+docker run --runtime=habana --name=hpu-test --network=host -e HABANA_VISIBLE_DEVICES=all -e VLLM_SKIP_WARMUP=true --entrypoint="" hpu-test-env python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m
+docker run --runtime=habana --name=hpu-test-tp2 --network=host -e HABANA_VISIBLE_DEVICES=all -e VLLM_SKIP_WARMUP=true --entrypoint="" hpu-test-env python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m --tensor-parallel-size 2

 EXITCODE=$?
-if [ $EXITCODE -eq 0 ]; then
-  echo "Test with basic model passed"
-else
-  echo "Test with basic model FAILED with exit code: $EXITCODE" >&2
-fi
-
-# The trap will handle the container removal and final exit.
--- a/.buildkite/scripts/hardware_ci/run-neuron-test.sh
+++ b/.buildkite/scripts/hardware_ci/run-neuron-test.sh
@ -54,11 +54,10 @@ docker run --rm -it --device=/dev/neuron0 --network bridge \
       --name "${container_name}" \
       ${image_name} \
       /bin/bash -c "
-            set -e; # Exit on first error
            python3 /workspace/vllm/examples/offline_inference/neuron.py;
            python3 -m pytest /workspace/vllm/tests/neuron/1_core/ -v --capture=tee-sys;
            for f in /workspace/vllm/tests/neuron/2_core/*.py; do
-                echo \"Running test file: \$f\";
+                echo 'Running test file: '$f;
                python3 -m pytest \$f -v --capture=tee-sys;
            done
       "
--- a/.buildkite/scripts/hardware_ci/run-tpu-v1-test-part2.sh
+++ b/.buildkite/scripts/hardware_ci/run-tpu-v1-test-part2.sh
@ -1,167 +0,0 @@
-#!/bin/bash
-
-set -xu
-
-
-remove_docker_container() { 
-    docker rm -f tpu-test || true;
-}
-
-trap remove_docker_container EXIT
-
-# Remove the container that might not be cleaned up in the previous run.
-remove_docker_container
-
-# Build the docker image.
-docker build -f docker/Dockerfile.tpu -t vllm-tpu .
-
-# Set up cleanup.
-cleanup_docker() {
-  # Get Docker's root directory
-  docker_root=$(docker info -f '{{.DockerRootDir}}')
-  if [ -z "$docker_root" ]; then
-    echo "Failed to determine Docker root directory."
-    exit 1
-  fi
-  echo "Docker root directory: $docker_root"
-  # Check disk usage of the filesystem where Docker's root directory is located
-  disk_usage=$(df "$docker_root" | tail -1 | awk '{print $5}' | sed 's/%//')
-  # Define the threshold
-  threshold=70
-  if [ "$disk_usage" -gt "$threshold" ]; then
-    echo "Disk usage is above $threshold%. Cleaning up Docker images and volumes..."
-    # Remove dangling images (those that are not tagged and not used by any container)
-    docker image prune -f
-    # Remove unused volumes / force the system prune for old images as well.
-    docker volume prune -f && docker system prune --force --filter "until=72h" --all
-    echo "Docker images and volumes cleanup completed."
-  else
-    echo "Disk usage is below $threshold%. No cleanup needed."
-  fi
-}
-cleanup_docker
-
-# For HF_TOKEN.
-source /etc/environment
-
-docker run --privileged --net host --shm-size=16G -it \
-    -e "HF_TOKEN=$HF_TOKEN" --name tpu-test \
-    vllm-tpu /bin/bash -c '
-set -e # Exit immediately if a command exits with a non-zero status.
-set -u # Treat unset variables as an error.
-
-echo "--- Starting script inside Docker container ---"
-
-# Create results directory
-RESULTS_DIR=$(mktemp -d)
-# If mktemp fails, set -e will cause the script to exit.
-echo "Results will be stored in: $RESULTS_DIR"
-
-# Install dependencies
-echo "--- Installing Python dependencies ---"
-python3 -m pip install --progress-bar off git+https://github.com/thuml/depyf.git \
-    && python3 -m pip install --progress-bar off pytest pytest-asyncio tpu-info \
-    && python3 -m pip install --progress-bar off "lm-eval @ git+https://github.com/EleutherAI/lm-evaluation-harness.git@206b7722158f58c35b7ffcd53b035fdbdda5126d" \
-    && python3 -m pip install --progress-bar off hf-transfer
-echo "--- Python dependencies installed ---"
-export VLLM_USE_V1=1
-export VLLM_XLA_CHECK_RECOMPILATION=1
-export VLLM_XLA_CACHE_PATH=
-echo "Using VLLM V1"
-
-echo "--- Hardware Information ---"
-# tpu-info
-echo "--- Starting Tests ---"
-set +e
-overall_script_exit_code=0
-
-# --- Test Definitions ---
-# If a test fails, this function will print logs and will not cause the main script to exit.
-run_test() {
-    local test_num=$1
-    local test_name=$2
-    local test_command=$3
-    local log_file="$RESULTS_DIR/test_${test_num}.log"
-    local actual_exit_code
-
-    echo "--- TEST_$test_num: Running $test_name ---"
-    
-    # Execute the test command.
-    eval "$test_command" > >(tee -a "$log_file") 2> >(tee -a "$log_file" >&2)
-    actual_exit_code=$?
-
-    echo "TEST_${test_num}_COMMAND_EXIT_CODE: $actual_exit_code" # This goes to main log
-    echo "TEST_${test_num}_COMMAND_EXIT_CODE: $actual_exit_code" >> "$log_file" # Also to per-test log
-
-    if [ "$actual_exit_code" -ne 0 ]; then
-        echo "TEST_$test_num ($test_name) FAILED with exit code $actual_exit_code." >&2
-        echo "--- Log for failed TEST_$test_num ($test_name) ---" >&2
-        if [ -f "$log_file" ]; then
-            cat "$log_file" >&2
-        else
-            echo "Log file $log_file not found for TEST_$test_num ($test_name)." >&2
-        fi
-        echo "--- End of log for TEST_$test_num ($test_name) ---" >&2
-        return "$actual_exit_code" # Return the failure code
-    else
-        echo "TEST_$test_num ($test_name) PASSED."
-        return 0 # Return success
-    fi
-}
-
-# Helper function to call run_test and update the overall script exit code
-run_and_track_test() {
-    local test_num_arg="$1"
-    local test_name_arg="$2"
-    local test_command_arg="$3"
-
-    # Run the test
-    run_test "$test_num_arg" "$test_name_arg" "$test_command_arg"
-    local test_specific_exit_code=$?
-
-    # If the test failed, set the overall script exit code to 1
-    if [ "$test_specific_exit_code" -ne 0 ]; then
-        # No need for extra echo here, run_test already logged the failure.
-        overall_script_exit_code=1
-    fi
-}
-
-# --- Actual Test Execution ---
-run_and_track_test 1 "test_struct_output_generate.py" \
-    "python3 -m pytest -s -v /workspace/vllm/tests/v1/entrypoints/llm/test_struct_output_generate.py -k \"not test_structured_output_with_reasoning_matrices\""
-run_and_track_test 2 "test_moe_pallas.py" \
-    "python3 -m pytest -s -v /workspace/vllm/tests/tpu/test_moe_pallas.py"
-run_and_track_test 3 "test_lora.py" \
-    "VLLM_XLA_CHECK_RECOMPILATION=0 python3 -m pytest -s -v /workspace/vllm/tests/tpu/lora/test_lora.py"
-run_and_track_test 4 "test_tpu_qkv_linear.py" \
-    "python3 -m pytest -s -v /workspace/vllm/tests/v1/tpu/test_tpu_qkv_linear.py"
-run_and_track_test 5 "test_spmd_model_weight_loading.py" \
-    "python3 -m pytest -s -v /workspace/vllm/tests/v1/tpu/test_spmd_model_weight_loading.py"
-run_and_track_test 6 "test_kv_cache_update_kernel.py" \
-    "python3 -m pytest -s -v /workspace/vllm/tests/v1/tpu/test_kv_cache_update_kernel.py"
-run_and_track_test 7 "test_tpu_int8.py" \
-    "python3 -m pytest -s -v /workspace/vllm/tests/v1/tpu/test_tpu_int8.py"
-
-# After all tests have been attempted, exit with the overall status.
-if [ "$overall_script_exit_code" -ne 0 ]; then
-    echo "--- One or more tests FAILED. Overall script exiting with failure code 1. ---"
-else
-    echo "--- All tests have completed and PASSED. Overall script exiting with success code 0. ---"
-fi
-exit "$overall_script_exit_code"
-' # IMPORTANT: This is the closing single quote for the bash -c "..." command. Ensure it is present and correct.
-
-# Capture the exit code of the docker run command
-DOCKER_RUN_EXIT_CODE=$?
-
-# The trap will run for cleanup.
-# Exit the main script with the Docker run command's exit code.
-if [ "$DOCKER_RUN_EXIT_CODE" -ne 0 ]; then
-    echo "Docker run command failed with exit code $DOCKER_RUN_EXIT_CODE."
-    exit "$DOCKER_RUN_EXIT_CODE"
-else
-    echo "Docker run command completed successfully."
-    exit 0
-fi
-# TODO: This test fails because it uses RANDOM_SEED sampling
-# pytest -v -s /workspace/vllm/tests/tpu/test_custom_dispatcher.py \
--- a/.buildkite/scripts/hardware_ci/run-tpu-v1-test.sh
+++ b/.buildkite/scripts/hardware_ci/run-tpu-v1-test.sh
@ -5,6 +5,7 @@ set -xu

 remove_docker_container() { 
    docker rm -f tpu-test || true; 
+    docker rm -f vllm-tpu || true;
 }

 trap remove_docker_container EXIT
@ -61,8 +62,7 @@ echo "Results will be stored in: $RESULTS_DIR"
 echo "--- Installing Python dependencies ---"
 python3 -m pip install --progress-bar off git+https://github.com/thuml/depyf.git \
    && python3 -m pip install --progress-bar off pytest pytest-asyncio tpu-info \
-    && python3 -m pip install --progress-bar off "lm-eval @ git+https://github.com/EleutherAI/lm-evaluation-harness.git@206b7722158f58c35b7ffcd53b035fdbdda5126d" \
-    && python3 -m pip install --progress-bar off hf-transfer
+    && python3 -m pip install --progress-bar off lm_eval[api]==0.4.4
 echo "--- Python dependencies installed ---"
 export VLLM_USE_V1=1
 export VLLM_XLA_CHECK_RECOMPILATION=1
@ -70,7 +70,7 @@ export VLLM_XLA_CACHE_PATH=
 echo "Using VLLM V1"

 echo "--- Hardware Information ---"
-# tpu-info
+tpu-info
 echo "--- Starting Tests ---"
 set +e
 overall_script_exit_code=0
@ -149,6 +149,16 @@ run_and_track_test 9 "test_multimodal.py" \
    "python3 -m pytest -s -v /workspace/vllm/tests/v1/tpu/test_multimodal.py"
 run_and_track_test 10 "test_pallas.py" \
    "python3 -m pytest -s -v /workspace/vllm/tests/v1/tpu/test_pallas.py"
+run_and_track_test 11 "test_struct_output_generate.py" \
+    "python3 -m pytest -s -v /workspace/vllm/tests/v1/entrypoints/llm/test_struct_output_generate.py -k 'not test_structured_output_with_reasoning_matrices'"
+run_and_track_test 12 "test_moe_pallas.py" \
+    "python3 -m pytest -s -v /workspace/vllm/tests/tpu/test_moe_pallas.py"
+run_and_track_test 13 "test_lora.py" \
+    "VLLM_XLA_CHECK_RECOMPILATION=0 python3 -m pytest -s -v /workspace/vllm/tests/tpu/lora/test_lora.py"
+run_and_track_test 14 "test_tpu_qkv_linear.py" \
+    "python3 -m pytest -s -v /workspace/vllm/tests/v1/tpu/test_tpu_qkv_linear.py"
+run_and_track_test 15 "test_spmd_model_weight_loading.py" \
+    "python3 -m pytest -s -v /workspace/vllm/tests/v1/tpu/test_spmd_model_weight_loading.py"

 # After all tests have been attempted, exit with the overall status.
 if [ "$overall_script_exit_code" -ne 0 ]; then
--- a/.buildkite/scripts/hardware_ci/run-xpu-test.sh
+++ b/.buildkite/scripts/hardware_ci/run-xpu-test.sh
@ -11,8 +11,8 @@ container_name="xpu_${BUILDKITE_COMMIT}_$(tr -dc A-Za-z0-9 < /dev/urandom | head
 docker build -t ${image_name} -f docker/Dockerfile.xpu .

 # Setup cleanup
-remove_docker_container() {
-  docker rm -f "${container_name}" || true;
+remove_docker_container() { 
+  docker rm -f "${container_name}" || true; 
  docker image rm -f "${image_name}" || true;
  docker system prune -f || true;
 }
@ -23,26 +23,9 @@ docker run \
    --device /dev/dri \
    -v /dev/dri/by-path:/dev/dri/by-path \
    --entrypoint="" \
-    -e "HF_TOKEN=${HF_TOKEN}" \
-    -e "ZE_AFFINITY_MASK=${ZE_AFFINITY_MASK}" \
    --name "${container_name}" \
    "${image_name}" \
-    bash -c '
-    set -e
-    echo $ZE_AFFINITY_MASK
-    VLLM_USE_V1=1 python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m --block-size 64 --enforce-eager
-    VLLM_USE_V1=1 python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m --block-size 64 -O3 -O.cudagraph_mode=NONE
-    VLLM_USE_V1=1 python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m --block-size 64 --enforce-eager -tp 2 --distributed-executor-backend ray
-    VLLM_USE_V1=1 python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m --block-size 64 --enforce-eager -tp 2 --distributed-executor-backend mp
-    cd tests
-    pytest -v -s v1/core
-    pytest -v -s v1/engine
-    pytest -v -s v1/sample --ignore=v1/sample/test_logprobs.py --ignore=v1/sample/test_logprobs_e2e.py
-    pytest -v -s v1/worker --ignore=v1/worker/test_gpu_model_runner.py
-    pytest -v -s v1/structured_output
-    pytest -v -s v1/spec_decode --ignore=v1/spec_decode/test_max_len.py --ignore=v1/spec_decode/test_eagle.py --ignore=v1/spec_decode/test_tree_attention.py
-    pytest -v -s v1/kv_connector/unit --ignore=v1/kv_connector/unit/test_multi_connector.py --ignore=v1/kv_connector/unit/test_nixl_connector.py --ignore=v1/kv_connector/unit/test_shared_storage_connector.py
-    pytest -v -s v1/test_serial_utils.py
-    pytest -v -s v1/test_utils.py
-    pytest -v -s v1/test_metrics_reader.py
+    sh -c '
+    VLLM_USE_V1=0 python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m
+    VLLM_USE_V1=0 python3 examples/offline_inference/basic/generate.py --model facebook/opt-125m -tp 2
 '
--- a/.buildkite/scripts/rerun-test.sh
+++ b/.buildkite/scripts/rerun-test.sh
@ -1,18 +0,0 @@
-#!/bin/bash
-
-# Usage: ./rerun_test.sh path/to/test.py::test_name
-
-# Check if argument is given
-if [ $# -lt 1 ]; then
-    echo "Usage: $0 path/to/test.py::test_name"
-    echo "Example: $0 tests/v1/engine/test_engine_core_client.py::test_kv_cache_events[True-tcp]"
-    exit 1
-fi
-
-TEST=$1
-COUNT=1
-
-while pytest -sv "$TEST"; do
-    COUNT=$((COUNT + 1))
-    echo "RUN NUMBER ${COUNT}"
-done
--- a/.buildkite/scripts/run-benchmarks.sh
+++ b/.buildkite/scripts/run-benchmarks.sh
@ -11,10 +11,10 @@ cd "$(dirname "${BASH_SOURCE[0]}")/../.."
 (which wget && which curl) || (apt-get update && apt-get install -y wget curl)

 # run python-based benchmarks and upload the result to buildkite
-vllm bench latency --output-json latency_results.json 2>&1 | tee benchmark_latency.txt
+python3 benchmarks/benchmark_latency.py --output-json latency_results.json 2>&1 | tee benchmark_latency.txt
 bench_latency_exit_code=$?

-vllm bench throughput --input-len 256 --output-len 256 --output-json throughput_results.json 2>&1 | tee benchmark_throughput.txt
+python3 benchmarks/benchmark_throughput.py --input-len 256 --output-len 256 --output-json throughput_results.json 2>&1 | tee benchmark_throughput.txt
 bench_throughput_exit_code=$?

 # run server-based benchmarks and upload the result to buildkite
@ -24,7 +24,7 @@ wget https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/r

 # wait for server to start, timeout after 600 seconds
 timeout 600 bash -c 'until curl localhost:8000/v1/models; do sleep 1; done' || exit 1
-vllm bench serve \
+python3 benchmarks/benchmark_serving.py \
    --backend vllm \
    --dataset-name sharegpt \
    --dataset-path ./ShareGPT_V3_unfiltered_cleaned_split.json \
--- a/.buildkite/scripts/tpu/cleanup_docker.sh
+++ b/.buildkite/scripts/tpu/cleanup_docker.sh
@ -1,24 +0,0 @@
-#!/bin/bash
-
-set -euo pipefail
-
-docker_root=$(docker info -f '{{.DockerRootDir}}')
-if [ -z "$docker_root" ]; then
-  echo "Failed to determine Docker root directory."
-  exit 1
-fi
-echo "Docker root directory: $docker_root"
-# Check disk usage of the filesystem where Docker's root directory is located
-disk_usage=$(df "$docker_root" | tail -1 | awk '{print $5}' | sed 's/%//')
-# Define the threshold
-threshold=70
-if [ "$disk_usage" -gt "$threshold" ]; then
-  echo "Disk usage is above $threshold%. Cleaning up Docker images and volumes..."
-  # Remove dangling images (those that are not tagged and not used by any container)
-  docker image prune -f
-  # Remove unused volumes / force the system prune for old images as well.
-  docker volume prune -f && docker system prune --force --filter "until=24h" --all
-  echo "Docker images and volumes cleanup completed."
-else
-  echo "Disk usage is below $threshold%. No cleanup needed."
-fi
--- a/.buildkite/scripts/tpu/config_v6e_1.env
+++ b/.buildkite/scripts/tpu/config_v6e_1.env
@ -1,14 +0,0 @@
-# Environment config
-TEST_NAME=llama8b
-CONTAINER_NAME=tpu-test
-
-# vllm config
-MODEL=meta-llama/Llama-3.1-8B-Instruct
-MAX_NUM_SEQS=256
-MAX_NUM_BATCHED_TOKENS=1024
-TENSOR_PARALLEL_SIZE=1
-MAX_MODEL_LEN=2048
-DOWNLOAD_DIR=/mnt/disks/persist
-EXPECTED_THROUGHPUT=8.0
-INPUT_LEN=1800
-OUTPUT_LEN=128
--- a/.buildkite/scripts/tpu/docker_run_bm.sh
+++ b/.buildkite/scripts/tpu/docker_run_bm.sh
@ -1,90 +0,0 @@
-#!/bin/bash
-
-if [ ! -f "$1" ]; then
-  echo "Error: The env file '$1' does not exist."
-  exit 1  # Exit the script with a non-zero status to indicate an error
-fi
-
-ENV_FILE=$1
-
-# For testing on local vm, use `set -a` to export all variables
-source /etc/environment
-source $ENV_FILE
-
-remove_docker_container() { 
-    docker rm -f $CONTAINER_NAME || true;
-}
-
-trap remove_docker_container EXIT
-
-# Remove the container that might not be cleaned up in the previous run.
-remove_docker_container
-
-LOG_ROOT=$(mktemp -d)
-# If mktemp fails, set -e will cause the script to exit.
-echo "Results will be stored in: $LOG_ROOT"
-
-if [ -z "$HF_TOKEN" ]; then
-  echo "Error: HF_TOKEN is not set or is empty."  
-  exit 1
-fi
-
-# Make sure mounted disk or dir exists
-if [ ! -d "$DOWNLOAD_DIR" ]; then
-    echo "Error: Folder $DOWNLOAD_DIR does not exist. This is useually a mounted drive. If no mounted drive, just create a folder."
-    exit 1
-fi
-
-echo "Run model $MODEL"
-echo
-
-echo "starting docker...$CONTAINER_NAME"
-echo    
-docker run \
- -v $DOWNLOAD_DIR:$DOWNLOAD_DIR \
- --env-file $ENV_FILE \
- -e HF_TOKEN="$HF_TOKEN" \
- -e TARGET_COMMIT=$BUILDKITE_COMMIT \
- -e MODEL=$MODEL \
- -e WORKSPACE=/workspace \
- --name $CONTAINER_NAME \
- -d \
- --privileged \
- --network host \
- -v /dev/shm:/dev/shm \
- vllm/vllm-tpu-bm tail -f /dev/null
-
-echo "run script..."
-echo
-docker exec "$CONTAINER_NAME" /bin/bash -c ".buildkite/scripts/tpu/run_bm.sh"
-
-echo "copy result back..."
-VLLM_LOG="$LOG_ROOT/$TEST_NAME"_vllm_log.txt
-BM_LOG="$LOG_ROOT/$TEST_NAME"_bm_log.txt
-docker cp "$CONTAINER_NAME:/workspace/vllm_log.txt" "$VLLM_LOG" 
-docker cp "$CONTAINER_NAME:/workspace/bm_log.txt" "$BM_LOG"
-
-throughput=$(grep "Request throughput (req/s):" "$BM_LOG" | sed 's/[^0-9.]//g')
-echo "throughput for $TEST_NAME at $BUILDKITE_COMMIT: $throughput"
-
-if [ "$BUILDKITE" = "true" ]; then
-  echo "Running inside Buildkite"
-  buildkite-agent artifact upload "$VLLM_LOG" 
-  buildkite-agent artifact upload "$BM_LOG"
-else
-  echo "Not running inside Buildkite"
-fi
-
-#
-# compare the throughput with EXPECTED_THROUGHPUT 
-# and assert meeting the expectation
-# 
-if [[ -z "$throughput" || ! "$throughput" =~ ^[0-9]+([.][0-9]+)?$ ]]; then
-  echo "Failed to get the throughput"
-  exit 1
-fi
-
-if (( $(echo "$throughput < $EXPECTED_THROUGHPUT" | bc -l) )); then
-  echo "Error: throughput($throughput) is less than expected($EXPECTED_THROUGHPUT)"
-  exit 1
-fi
--- a/.buildkite/scripts/tpu/quantized_v6e_1.env
+++ b/.buildkite/scripts/tpu/quantized_v6e_1.env
@ -1,14 +0,0 @@
-# Environment config
-TEST_NAME=llama8bw8a8
-CONTAINER_NAME=tpu-test
-
-# vllm config
-MODEL=RedHatAI/Meta-Llama-3.1-8B-Instruct-quantized.w8a8
-MAX_NUM_SEQS=128
-MAX_NUM_BATCHED_TOKENS=1024
-TENSOR_PARALLEL_SIZE=1
-MAX_MODEL_LEN=2048
-DOWNLOAD_DIR=/mnt/disks/persist
-EXPECTED_THROUGHPUT=10.0
-INPUT_LEN=1800
-OUTPUT_LEN=128
--- a/.buildkite/scripts/tpu/run_bm.sh
+++ b/.buildkite/scripts/tpu/run_bm.sh
@ -1,93 +0,0 @@
-#!/bin/bash
-
-set -euo pipefail
-
-VLLM_LOG="$WORKSPACE/vllm_log.txt"
-BM_LOG="$WORKSPACE/bm_log.txt"
-
-if [ -n "$TARGET_COMMIT" ]; then
-  head_hash=$(git rev-parse HEAD)
-  if [ "$TARGET_COMMIT" != "$head_hash" ]; then
-    echo "Error: target commit $TARGET_COMMIT does not match HEAD: $head_hash"
-    exit 1
-  fi
-fi
-
-echo "model: $MODEL"
-echo
-
-#
-# create a log folder
-#
-mkdir "$WORKSPACE/log"
-
-# TODO: Move to image building.
-pip install pandas
-pip install datasets
-
-#
-# create sonnet_4x
-#
-echo "Create sonnet_4x.txt"
-echo "" > benchmarks/sonnet_4x.txt
-for _ in {1..4}
- do
-  cat benchmarks/sonnet.txt >> benchmarks/sonnet_4x.txt
-done
-
-#
-# start vllm service in backend
-#
-echo "lanching vllm..."
-echo "logging to $VLLM_LOG"
-echo
-
-VLLM_USE_V1=1 vllm serve $MODEL \
- --seed 42 \
- --max-num-seqs $MAX_NUM_SEQS \
- --max-num-batched-tokens $MAX_NUM_BATCHED_TOKENS \
- --tensor-parallel-size $TENSOR_PARALLEL_SIZE \
- --no-enable-prefix-caching \
- --download_dir $DOWNLOAD_DIR \
- --max-model-len $MAX_MODEL_LEN > "$VLLM_LOG" 2>&1 &
-
-
-echo "wait for 20 minutes.."
-echo
-# sleep 1200
-# wait for 10 minutes...
-for i in {1..120}; do
-    # TODO: detect other type of errors.
-    if grep -Fq "raise RuntimeError" "$VLLM_LOG"; then
-        echo "Detected RuntimeError, exiting."
-        exit 1
-    elif grep -Fq "Application startup complete" "$VLLM_LOG"; then
-        echo "Application started"
-        break
-    else
-        echo "wait for 10 seconds..."
-        sleep 10
-    fi
-done
-
-#
-# run test
-#
-echo "run benchmark test..."
-echo "logging to $BM_LOG"
-echo
-vllm bench serve \
-    --backend vllm \
-    --model $MODEL  \
-    --dataset-name sonnet \
-    --dataset-path benchmarks/sonnet_4x.txt \
-    --sonnet-input-len $INPUT_LEN \
-    --sonnet-output-len $OUTPUT_LEN \
-    --ignore-eos > "$BM_LOG"
-
-echo "completed..."
-echo
-
-throughput=$(grep "Request throughput (req/s):" "$BM_LOG" | sed 's/[^0-9.]//g')
-echo "throughput: $throughput"
-echo
--- a/.buildkite/scripts/upload-wheels.sh
+++ b/.buildkite/scripts/upload-wheels.sh
@ -14,19 +14,8 @@ fi
 # Get the single wheel file
 wheel="${wheel_files[0]}"

-# Detect architecture and rename 'linux' to appropriate manylinux version
-arch=$(uname -m)
-if [[ $arch == "x86_64" ]]; then
-    manylinux_version="manylinux1"
-elif [[ $arch == "aarch64" ]]; then
-    manylinux_version="manylinux2014"
-else
-    echo "Warning: Unknown architecture $arch, using manylinux1 as default"
-    manylinux_version="manylinux1"
-fi
-
-# Rename 'linux' to the appropriate manylinux version in the wheel filename
-new_wheel="${wheel/linux/$manylinux_version}"
+# Rename 'linux' to 'manylinux1' in the wheel filename
+new_wheel="${wheel/linux/manylinux1}"
 mv -- "$wheel" "$new_wheel"
 wheel="$new_wheel"

--- a/.buildkite/test-pipeline.yaml
+++ b/.buildkite/test-pipeline.yaml
@ -31,15 +31,15 @@
 steps:
 ##### fast check tests  #####

- label: Pytorch Nightly Dependency Override Check # 2min
-  # if this test fails, it means the nightly torch version is not compatible with some
-  # of the dependencies. Please check the error message and add the package to whitelist
-  # in /vllm/tools/generate_nightly_torch_test.py
-  soft_fail: true
-  source_file_dependencies:
-  - requirements/nightly_torch_test.txt
+- label: Documentation Build # 2min
+  mirror_hardwares: [amdexperimental]
+  working_dir: "/vllm-workspace/test_docs"
+  fast_check: true
+  no_gpu: True
  commands:
-  - bash standalone_tests/pytorch_nightly_dependency.sh
+  - pip install -r ../requirements/docs.txt
+  # TODO: add `--strict` once warnings in docstrings are fixed
+  - mkdocs build

 - label: Async Engine, Inputs, Utils, Worker Test # 24min
  mirror_hardwares: [amdexperimental]
@ -47,20 +47,20 @@ steps:
  - vllm/
  - tests/mq_llm_engine
  - tests/async_engine
-  - tests/test_inputs.py
-  - tests/test_outputs.py
+  - tests/test_inputs
  - tests/multimodal
-  - tests/utils_
+  - tests/test_utils
  - tests/worker
  - tests/standalone_tests/lazy_imports.py
  commands:
  - python3 standalone_tests/lazy_imports.py
  - pytest -v -s mq_llm_engine # MQLLMEngine
  - pytest -v -s async_engine # AsyncLLMEngine
+  - NUM_SCHEDULER_STEPS=4 pytest -v -s async_engine/test_async_llm_engine.py
  - pytest -v -s test_inputs.py
  - pytest -v -s test_outputs.py
  - pytest -v -s multimodal
-  - pytest -v -s utils_ # Utils
+  - pytest -v -s test_utils.py # Utils
  - pytest -v -s worker # Worker

 - label: Python-only Installation Test
@ -72,7 +72,7 @@ steps:
  - bash standalone_tests/python_only_compile.sh

 - label: Basic Correctness Test # 30min
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  fast_check: true
  torch_nightly: true
  source_file_dependencies:
@ -88,8 +88,17 @@ steps:
  - pytest -v -s basic_correctness/test_cpu_offload.py
  - VLLM_TEST_ENABLE_ARTIFICIAL_PREEMPT=1 pytest -v -s basic_correctness/test_preemption.py

- label: Core Test # 10min
+- label: Chunked Prefill Test
  mirror_hardwares: [amdexperimental]
+  source_file_dependencies:
+  - vllm/
+  - tests/basic_correctness/test_chunked_prefill
+  commands:
+  - VLLM_ATTENTION_BACKEND=XFORMERS pytest -v -s basic_correctness/test_chunked_prefill.py
+  - VLLM_ATTENTION_BACKEND=FLASH_ATTN pytest -v -s basic_correctness/test_chunked_prefill.py
+
+- label: Core Test # 10min
+  mirror_hardwares: [amdexperimental, amdproduction]
  fast_check: true
  source_file_dependencies:
  - vllm/core
@ -98,7 +107,7 @@ steps:
  commands:
  - pytest -v -s core

- label: Entrypoints Test (LLM) # 40min
+- label: Entrypoints Test # 40min
  mirror_hardwares: [amdexperimental]
  working_dir: "/vllm-workspace/tests"
  fast_check: true
@ -106,29 +115,19 @@ steps:
  source_file_dependencies:
  - vllm/
  - tests/entrypoints/llm
+  - tests/entrypoints/openai
+  - tests/entrypoints/test_chat_utils
  - tests/entrypoints/offline_mode
  commands:
  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-  - pytest -v -s entrypoints/llm --ignore=entrypoints/llm/test_lazy_outlines.py --ignore=entrypoints/llm/test_generate.py --ignore=entrypoints/llm/test_generate_multiple_loras.py --ignore=entrypoints/llm/test_collective_rpc.py
+  - pytest -v -s entrypoints/llm --ignore=entrypoints/llm/test_lazy_outlines.py --ignore=entrypoints/llm/test_generate.py --ignore=entrypoints/llm/test_generate_multiple_loras.py --ignore=entrypoints/llm/test_guided_generate.py --ignore=entrypoints/llm/test_collective_rpc.py
  - pytest -v -s entrypoints/llm/test_lazy_outlines.py # it needs a clean process
  - pytest -v -s entrypoints/llm/test_generate.py # it needs a clean process
  - pytest -v -s entrypoints/llm/test_generate_multiple_loras.py # it needs a clean process
-  - VLLM_USE_V1=0 pytest -v -s entrypoints/offline_mode # Needs to avoid interference with other tests
-
- label: Entrypoints Test (API Server) # 40min
-  mirror_hardwares: [amdexperimental]
-  working_dir: "/vllm-workspace/tests"
-  fast_check: true
-  torch_nightly: true
-  source_file_dependencies:
-  - vllm/
-  - tests/entrypoints/openai
-  - tests/entrypoints/test_chat_utils
-  commands:
-  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
-  - PYTHONPATH=/vllm-workspace pytest -v -s entrypoints/openai/test_collective_rpc.py # PYTHONPATH is needed to import custom Worker extension
-  - pytest -v -s entrypoints/openai --ignore=entrypoints/openai/test_chat_with_tool_reasoning.py --ignore=entrypoints/openai/test_oot_registration.py --ignore=entrypoints/openai/test_tensorizer_entrypoint.py --ignore=entrypoints/openai/correctness/ --ignore=entrypoints/openai/test_collective_rpc.py
+  - VLLM_USE_V1=0 pytest -v -s entrypoints/llm/test_guided_generate.py # it needs a clean process
+  - pytest -v -s entrypoints/openai --ignore=entrypoints/openai/test_chat_with_tool_reasoning.py --ignore=entrypoints/openai/test_oot_registration.py --ignore=entrypoints/openai/test_tensorizer_entrypoint.py --ignore=entrypoints/openai/correctness/
  - pytest -v -s entrypoints/test_chat_utils.py
+  - VLLM_USE_V1=0 pytest -v -s entrypoints/offline_mode # Needs to avoid interference with other tests

 - label: Distributed Tests (4 GPUs) # 10min
  mirror_hardwares: [amdexperimental]
@ -140,14 +139,12 @@ steps:
  - tests/distributed/test_utils
  - tests/distributed/test_pynccl
  - tests/distributed/test_events
+  - tests/spec_decode/e2e/test_integration_dist_tp4
  - tests/compile/test_basic_correctness
  - examples/offline_inference/rlhf.py
  - examples/offline_inference/rlhf_colocate.py
  - tests/examples/offline_inference/data_parallel.py
  - tests/v1/test_async_llm_dp.py
-  - tests/v1/test_external_lb_dp.py
-  - tests/v1/test_internal_lb_dp.py
-  - tests/v1/test_hybrid_lb_dp.py
  - tests/v1/engine/test_engine_core_client.py
  commands:
  # test with tp=2 and external_dp=2
@ -156,16 +153,14 @@ steps:
  # test with tp=2 and pp=2
  - PP_SIZE=2 torchrun --nproc-per-node=4 distributed/test_torchrun_example.py
  # test with internal dp
-  - python3 ../examples/offline_inference/data_parallel.py --enforce-eager
+  - python3 ../examples/offline_inference/data_parallel.py
  - TP_SIZE=2 DP_SIZE=2 pytest -v -s v1/test_async_llm_dp.py
-  - TP_SIZE=2 DP_SIZE=2 pytest -v -s v1/test_external_lb_dp.py
-  - TP_SIZE=1 DP_SIZE=4 pytest -v -s v1/test_internal_lb_dp.py
-  - TP_SIZE=1 DP_SIZE=4 pytest -v -s v1/test_hybrid_lb_dp.py
  - pytest -v -s v1/engine/test_engine_core_client.py::test_kv_cache_events_dp
  - pytest -v -s distributed/test_utils.py
  - pytest -v -s compile/test_basic_correctness.py
  - pytest -v -s distributed/test_pynccl.py
  - pytest -v -s distributed/test_events.py
+  - pytest -v -s spec_decode/e2e/test_integration_dist_tp4.py
  # TODO: create a dedicated test section for multi-GPU example tests
  # when we have multiple distributed example tests
  - pushd ../examples/offline_inference
@ -173,25 +168,8 @@ steps:
  - VLLM_ALLOW_INSECURE_SERIALIZATION=1 RAY_DEDUP_LOGS=0 python3 rlhf_colocate.py
  - popd

- label: EPLB Algorithm Test
-  working_dir: "/vllm-workspace/tests"
-  source_file_dependencies:
-  - vllm/distributed/eplb
-  - tests/distributed/test_eplb_algo.py
-  commands:
-  - pytest -v -s distributed/test_eplb_algo.py
-
- label: EPLB Execution Test # 5min
-  working_dir: "/vllm-workspace/tests"
-  num_gpus: 4
-  source_file_dependencies:
-  - vllm/distributed/eplb
-  - tests/distributed/test_eplb_execute.py
-  commands:
-  - pytest -v -s distributed/test_eplb_execute.py
-
 - label: Metrics, Tracing Test # 10min
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  num_gpus: 2
  source_file_dependencies:
  - vllm/
@ -199,18 +177,13 @@ steps:
  - tests/tracing
  commands:
  - pytest -v -s metrics
-  - "pip install \
-      'opentelemetry-sdk>=1.26.0' \
-      'opentelemetry-api>=1.26.0' \
-      'opentelemetry-exporter-otlp>=1.26.0' \
-      'opentelemetry-semantic-conventions-ai>=0.4.1'"
  - pytest -v -s tracing

 ##### fast check tests  #####
 #####  1 GPU test  #####

 - label: Regression Test # 5min
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  source_file_dependencies:
  - vllm/
  - tests/test_regression
@ -220,7 +193,7 @@ steps:
  working_dir: "/vllm-workspace/tests" # optional

 - label: Engine Test # 10min
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  source_file_dependencies:
  - vllm/
  - tests/engine
@ -244,14 +217,11 @@ steps:
    - pytest -v -s v1/core
    - pytest -v -s v1/engine
    - pytest -v -s v1/entrypoints
-    - pytest -v -s v1/executor
    - pytest -v -s v1/sample
-    - pytest -v -s v1/logits_processors
    - pytest -v -s v1/worker
    - pytest -v -s v1/structured_output
    - pytest -v -s v1/spec_decode
    - pytest -v -s v1/kv_connector/unit
-    - pytest -v -s v1/metrics
    - pytest -v -s v1/test_serial_utils.py
    - pytest -v -s v1/test_utils.py
    - pytest -v -s v1/test_oracle.py
@ -260,7 +230,7 @@ steps:
    # VLLM_USE_FLASHINFER_SAMPLER or not on H100.
    - pytest -v -s v1/e2e
    # Integration test for streaming correctness (requires special branch).
-    - pip install -U git+https://github.com/robertgshaw2-redhat/lm-evaluation-harness.git@streaming-api
+    - pip install -U git+https://github.com/robertgshaw2-neuralmagic/lm-evaluation-harness.git@streaming-api
    - pytest -v -s entrypoints/openai/correctness/test_lmeval.py::test_lm_eval_accuracy_v1_engine

 - label: Examples Test # 25min
@ -278,7 +248,7 @@ steps:
    - python3 offline_inference/llm_engine_example.py
    - python3 offline_inference/audio_language.py --seed 0
    - python3 offline_inference/vision_language.py --seed 0
-    - python3 offline_inference/vision_language_pooling.py --seed 0
+    - python3 offline_inference/vision_language_embedding.py --seed 0
    - python3 offline_inference/vision_language_multi_image.py --seed 0
    - VLLM_USE_V1=0 python3 others/tensorize_vllm_model.py --model facebook/opt-125m serialize --serialized-directory /tmp/ --suffix v1 && python3 others/tensorize_vllm_model.py --model facebook/opt-125m deserialize --path-to-tensors /tmp/vllm/facebook/opt-125m/v1/model.tensors
    - python3 offline_inference/encoder_decoder.py
@ -288,13 +258,13 @@ steps:
    - python3 offline_inference/basic/score.py
    - VLLM_USE_V1=0 python3 offline_inference/profiling.py --model facebook/opt-125m run_num_steps --num-steps 2

- label: Platform Tests (CUDA)
-  mirror_hardwares: [amdexperimental]
+- label: Prefix Caching Test # 9min
+  mirror_hardwares: [amdexperimental, amdproduction]
  source_file_dependencies:
  - vllm/
-  - tests/cuda
+  - tests/prefix_caching
  commands:
-    - pytest -v -s cuda/test_cuda_context.py
+    - pytest -v -s prefix_caching

 - label: Samplers Test # 36min
  mirror_hardwares: [amdexperimental]
@ -307,16 +277,27 @@ steps:
    - pytest -v -s samplers
    - VLLM_USE_FLASHINFER_SAMPLER=1 pytest -v -s samplers

- label: LoRA Test %N # 15min each
+- label: Speculative decoding tests # 40min
  mirror_hardwares: [amdexperimental]
  source_file_dependencies:
+  - vllm/spec_decode
+  - tests/spec_decode
+  - vllm/model_executor/models/eagle.py
+  commands:
+    - pytest -v -s spec_decode/e2e/test_multistep_correctness.py
+    - VLLM_ATTENTION_BACKEND=FLASH_ATTN pytest -v -s spec_decode --ignore=spec_decode/e2e/test_multistep_correctness.py --ignore=spec_decode/e2e/test_mtp_correctness.py
+    - pytest -v -s spec_decode/e2e/test_eagle_correctness.py
+
+- label: LoRA Test %N # 15min each
+  mirror_hardwares: [amdexperimental, amdproduction]
+  source_file_dependencies:
  - vllm/lora
  - tests/lora
  command: pytest -v -s lora --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT --ignore=lora/test_chatglm3_tp.py --ignore=lora/test_llama_tp.py
  parallelism: 4

 - label: PyTorch Compilation Unit Tests
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  torch_nightly: true
  source_file_dependencies:
    - vllm/
@ -324,15 +305,12 @@ steps:
  commands:
    - pytest -v -s compile/test_pass_manager.py
    - pytest -v -s compile/test_fusion.py
-    - pytest -v -s compile/test_fusion_attn.py
    - pytest -v -s compile/test_silu_mul_quant_fusion.py
    - pytest -v -s compile/test_sequence_parallelism.py
    - pytest -v -s compile/test_async_tp.py
-    - pytest -v -s compile/test_fusion_all_reduce.py
-    - pytest -v -s compile/test_decorator.py

 - label: PyTorch Fullgraph Smoke Test # 9min
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  torch_nightly: true
  source_file_dependencies:
  - vllm/
@ -343,10 +321,9 @@ steps:
  - pytest -v -s compile/piecewise/test_simple.py
  - pytest -v -s compile/piecewise/test_toy_llama.py
  - pytest -v -s compile/piecewise/test_full_cudagraph.py
-  - pytest -v -s compile/piecewise/test_multiple_graphs.py

 - label: PyTorch Fullgraph Test # 18min
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  torch_nightly: true
  source_file_dependencies:
  - vllm/
@ -355,7 +332,7 @@ steps:
  - pytest -v -s compile/test_full_graph.py

 - label: Kernels Core Operation Test
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  source_file_dependencies:
  - csrc/
  - tests/kernels/core
@ -363,7 +340,7 @@ steps:
    - pytest -v -s kernels/core

 - label: Kernels Attention Test %N
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  source_file_dependencies:
  - csrc/attention/
  - vllm/attention
@ -374,26 +351,23 @@ steps:
  parallelism: 2

 - label: Kernels Quantization Test %N
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  source_file_dependencies:
  - csrc/quantization/
  - vllm/model_executor/layers/quantization
  - tests/kernels/quantization
  commands:
-    - pytest -v -s kernels/quantization --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
+    - pytest -v -s kernels/quantization  --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
  parallelism: 2

- label: Kernels MoE Test %N
+- label: Kernels MoE Test
  mirror_hardwares: [amdexperimental]
  source_file_dependencies:
-  - csrc/quantization/cutlass_w8a8/moe/
  - csrc/moe/
  - tests/kernels/moe
  - vllm/model_executor/layers/fused_moe/
-  - vllm/distributed/device_communicators/
  commands:
-    - pytest -v -s kernels/moe --shard-id=$$BUILDKITE_PARALLEL_JOB --num-shards=$$BUILDKITE_PARALLEL_JOB_COUNT
-  parallelism: 2
+    - pytest -v -s kernels/moe

 - label: Kernels Mamba Test
  mirror_hardwares: [amdexperimental]
@ -404,7 +378,8 @@ steps:
    - pytest -v -s kernels/mamba

 - label: Tensorizer Test # 11min
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
+  soft_fail: true
  source_file_dependencies:
  - vllm/model_executor/model_loader
  - tests/tensorizer_loader
@ -416,7 +391,8 @@ steps:
    - pytest -v -s entrypoints/openai/test_tensorizer_entrypoint.py

 - label: Model Executor Test
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
+  soft_fail: true
  source_file_dependencies:
  - vllm/model_executor
  - tests/model_executor
@ -426,7 +402,7 @@ steps:
    - pytest -v -s model_executor

 - label: Benchmarks # 9min
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  working_dir: "/vllm-workspace/.buildkite"
  source_file_dependencies:
  - benchmarks/
@ -434,7 +410,7 @@ steps:
  - bash scripts/run-benchmarks.sh

 - label: Benchmarks CLI Test # 10min
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  source_file_dependencies:
  - vllm/
  - tests/benchmarks/
@ -448,18 +424,17 @@ steps:
  - vllm/model_executor/layers/quantization
  - tests/quantization
  commands:
-  # temporary install here since we need nightly, will move to requirements/test.in
-  # after torchao 0.12 release
-  - pip install --pre torchao --index-url https://download.pytorch.org/whl/nightly/cu126
  - VLLM_TEST_FORCE_LOAD_FORMAT=auto pytest -v -s quantization

 - label: LM Eval Small Models # 53min
  mirror_hardwares: [amdexperimental]
+  working_dir: "/vllm-workspace/.buildkite/lm-eval-harness"
  source_file_dependencies:
  - csrc/
  - vllm/model_executor/layers/quantization
  commands:
-  - pytest -s -v evals/gsm8k/test_gsm8k_correctness.py --config-list-file=configs/models-small.txt --tp-size=1
+  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
+  - pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-small.txt --tp-size=1

 - label: OpenAI API correctness
  mirror_hardwares: [amdexperimental]
@ -492,7 +467,7 @@ steps:
 #####  models test  #####

 - label: Basic Models Test # 24min
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  torch_nightly: true
  source_file_dependencies:
  - vllm/
@ -511,22 +486,11 @@ steps:
  - vllm/
  - tests/models/language
  commands:
+    # Install causal-conv1d for plamo2 models here, as it is not compatible with pip-compile.
+    - pip install 'git+https://github.com/Dao-AILab/causal-conv1d@v1.5.0.post8'
    - pip freeze | grep -E 'torch'
    - pytest -v -s models/language -m core_model

- label: Language Models Test (Hybrid) # 35 min
-  mirror_hardwares: [amdexperimental]
-  torch_nightly: true
-  source_file_dependencies:
-  - vllm/
-  - tests/models/language/generation
-  commands:
-    # Install fast path packages for testing against transformers
-    # Note: also needed to run plamo2 model in vLLM
-    - uv pip install --system --no-build-isolation 'git+https://github.com/state-spaces/mamba@v2.2.5'
-    - uv pip install --system --no-build-isolation 'git+https://github.com/Dao-AILab/causal-conv1d@v1.5.2'
-    - pytest -v -s models/language/generation -m hybrid_model
-
 - label: Language Models Test (Extended Generation) # 1hr20min
  mirror_hardwares: [amdexperimental]
  optional: true
@ -536,7 +500,7 @@ steps:
  commands:
    # Install causal-conv1d for plamo2 models here, as it is not compatible with pip-compile.
    - pip install 'git+https://github.com/Dao-AILab/causal-conv1d@v1.5.0.post8'
-    - pytest -v -s models/language/generation -m '(not core_model) and (not hybrid_model)'
+    - pytest -v -s models/language/generation -m 'not core_model'

 - label: Language Models Test (Extended Pooling)  # 36min
  mirror_hardwares: [amdexperimental]
@ -547,15 +511,6 @@ steps:
  commands:
    - pytest -v -s models/language/pooling -m 'not core_model'

- label: Multi-Modal Processor Test
-  source_file_dependencies:
-  - vllm/
-  - tests/models/multimodal
-  commands:
-    - pip install git+https://github.com/TIGER-AI-Lab/Mantis.git
-    - pytest -v -s models/multimodal/processing --ignore models/multimodal/processing/test_tensor_schema.py
-    - pytest -v -s models/multimodal/processing/test_tensor_schema.py
-
 - label: Multi-Modal Models Test (Standard)
  mirror_hardwares: [amdexperimental]
  torch_nightly: true
@ -565,7 +520,8 @@ steps:
  commands:
    - pip install git+https://github.com/TIGER-AI-Lab/Mantis.git
    - pip freeze | grep -E 'torch'
-    - pytest -v -s models/multimodal -m core_model --ignore models/multimodal/generation/test_whisper.py --ignore models/multimodal/processing
+    - pytest -v -s models/multimodal/processing
+    - pytest -v -s --ignore models/multimodal/generation/test_whisper.py models/multimodal -m core_model
    - cd .. && pytest -v -s tests/models/multimodal/generation/test_whisper.py -m core_model  # Otherwise, mp_method="spawn" doesn't work

 - label: Multi-Modal Models Test (Extended) 1
@ -576,7 +532,7 @@ steps:
  - tests/models/multimodal
  commands:
    - pip install git+https://github.com/TIGER-AI-Lab/Mantis.git
-    - pytest -v -s models/multimodal -m 'not core_model' --ignore models/multimodal/generation/test_common.py --ignore models/multimodal/processing
+    - pytest -v -s --ignore models/multimodal/generation/test_common.py --ignore models/multimodal/processing models/multimodal -m 'not core_model'

 - label: Multi-Modal Models Test (Extended) 2
  mirror_hardwares: [amdexperimental]
@ -589,7 +545,7 @@ steps:
    - pytest -v -s models/multimodal/generation/test_common.py -m 'split(group=0) and not core_model'

 - label: Multi-Modal Models Test (Extended) 3
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  optional: true
  source_file_dependencies:
  - vllm/
@ -599,7 +555,7 @@ steps:
    - pytest -v -s models/multimodal/generation/test_common.py -m 'split(group=1) and not core_model'

 - label: Quantized Models Test
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  source_file_dependencies:
  - vllm/model_executor/layers/quantization
  - tests/models/quantization
@ -608,7 +564,7 @@ steps:

 # This test is used only in PR development phase to test individual models and should never run on main
 - label: Custom Models Test
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  optional: true
  commands:
    - echo 'Testing custom models...'
@ -616,59 +572,11 @@ steps:
    # e.g. pytest -v -s models/encoder_decoder/vision_language/test_mllama.py
    # *To avoid merge conflicts, remember to REMOVE (not just comment out) them before merging the PR*

- label: Transformers Nightly Models Test
-  working_dir: "/vllm-workspace/"
-  optional: true
-  commands:
-    - pip install --upgrade git+https://github.com/huggingface/transformers
-    - pytest -v -s tests/models/test_initialization.py
-    - pytest -v -s tests/models/multimodal/processing/
-    - pytest -v -s tests/models/multimodal/test_mapping.py
-    - python3 examples/offline_inference/basic/chat.py
-    - python3 examples/offline_inference/audio_language.py --model-type whisper
-    - python3 examples/offline_inference/vision_language.py --model-type qwen2_5_vl
-
- label: Blackwell Test
-  working_dir: "/vllm-workspace/"
-  gpu: b200
-  # optional: true
-  source_file_dependencies:
-  - csrc/quantization/fp4/
-  - csrc/attention/mla/
-  - csrc/quantization/cutlass_w8a8/moe/
-  - vllm/model_executor/layers/fused_moe/cutlass_moe.py
-  - vllm/model_executor/layers/fused_moe/flashinfer_cutlass_moe.py
-  - vllm/model_executor/layers/fused_moe/flashinfer_cutlass_prepare_finalize.py
-  - vllm/model_executor/layers/quantization/utils/flashinfer_utils.py
-  - vllm/v1/attention/backends/flashinfer.py
-  - vllm/compilation/fusion.py
-  - vllm/compilation/fusion_attn.py
-  commands:
-    - nvidia-smi
-    - python3 examples/offline_inference/basic/chat.py
-    # Attention
-    # num_heads2 broken by https://github.com/flashinfer-ai/flashinfer/issues/1353
-    - pytest -v -s tests/kernels/attention/test_flashinfer.py -k 'not num_heads2'
-    - pytest -v -s tests/kernels/attention/test_flashinfer_trtllm_attention.py
-    - pytest -v -s tests/kernels/test_cutlass_mla_decode.py
-    # Quantization
-    - pytest -v -s tests/kernels/quantization/test_cutlass_scaled_mm.py -k 'fp8'
-    - pytest -v -s tests/kernels/quantization/test_nvfp4_quant.py
-    - pytest -v -s tests/kernels/quantization/test_nvfp4_scaled_mm.py
-    - pytest -v -s tests/kernels/quantization/test_flashinfer_scaled_mm.py
-    - pytest -v -s tests/kernels/quantization/test_flashinfer_nvfp4_scaled_mm.py
-    - pytest -v -s tests/kernels/moe/test_nvfp4_moe.py
-    - pytest -v -s tests/kernels/moe/test_mxfp4_moe.py
-    # Fusion
-    - pytest -v -s tests/compile/test_fusion_all_reduce.py
-    - pytest -v -s tests/compile/test_fusion_attn.py::test_attention_quant_pattern
-    - pytest -v -s tests/kernels/moe/test_flashinfer.py
-
 #####  1 GPU test  #####
 #####  multi gpus test  #####

 - label: Distributed Comm Ops Test # 7min
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  working_dir: "/vllm-workspace/tests"
  num_gpus: 2
  source_file_dependencies:
@ -689,18 +597,13 @@ steps:
  - vllm/executor/
  - vllm/model_executor/models/
  - tests/distributed/
-  - tests/examples/offline_inference/data_parallel.py
  commands:
  - # the following commands are for the first node, with ip 192.168.10.10 (ray environment already set up)
    - VLLM_TEST_SAME_HOST=0 torchrun --nnodes 2 --nproc-per-node=2 --rdzv_backend=c10d --rdzv_endpoint=192.168.10.10 distributed/test_same_node.py | grep 'Same node test passed'
-    - NUM_NODES=2 torchrun --nnodes 2 --nproc-per-node=2 --rdzv_backend=c10d --rdzv_endpoint=192.168.10.10 distributed/test_node_count.py | grep 'Node count test passed'
-    - python3 ../examples/offline_inference/data_parallel.py --dp-size=2 --tp-size=1 --node-size=2 --node-rank=0 --master-addr=192.168.10.10 --master-port=12345 --enforce-eager --trust-remote-code
    - VLLM_MULTI_NODE=1 pytest -v -s distributed/test_multi_node_assignment.py
    - VLLM_MULTI_NODE=1 pytest -v -s distributed/test_pipeline_parallel.py
  - # the following commands are for the second node, with ip 192.168.10.11 (ray environment already set up)
    - VLLM_TEST_SAME_HOST=0 torchrun --nnodes 2 --nproc-per-node=2 --rdzv_backend=c10d --rdzv_endpoint=192.168.10.10 distributed/test_same_node.py | grep 'Same node test passed'
-    - NUM_NODES=2 torchrun --nnodes 2 --nproc-per-node=2 --rdzv_backend=c10d --rdzv_endpoint=192.168.10.10 distributed/test_node_count.py | grep 'Node count test passed'
-    - python3 ../examples/offline_inference/data_parallel.py --dp-size=2 --tp-size=1 --node-size=2 --node-rank=1 --master-addr=192.168.10.10 --master-port=12345 --enforce-eager --trust-remote-code

 - label: Distributed Tests (2 GPUs) # 40min
  mirror_hardwares: [amdexperimental]
@ -718,12 +621,10 @@ steps:
  - vllm/worker/model_runner.py
  - entrypoints/llm/test_collective_rpc.py
  - tests/v1/test_async_llm_dp.py
-  - tests/v1/test_external_lb_dp.py
  - tests/v1/entrypoints/openai/test_multi_api_servers.py
  - vllm/v1/engine/
  commands:
  - TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/test_async_llm_dp.py
-  - TP_SIZE=1 DP_SIZE=2 pytest -v -s v1/test_external_lb_dp.py
  - DP_SIZE=2 pytest -v -s v1/entrypoints/openai/test_multi_api_servers.py
  - pytest -v -s entrypoints/llm/test_collective_rpc.py
  - pytest -v -s ./compile/test_basic_correctness.py
@ -738,9 +639,10 @@ steps:
  - pytest -v -s distributed/test_sequence_parallel.py
  # this test fails consistently.
  # TODO: investigate and fix
+  # - pytest -v -s spec_decode/e2e/test_integration_dist_tp2.py
  - VLLM_USE_V1=0 CUDA_VISIBLE_DEVICES=0,1 pytest -v -s test_sharded_state_loader.py
+  - VLLM_USE_V1=0 CUDA_VISIBLE_DEVICES=0,1 pytest -v -s kv_transfer/test_disagg.py
  - CUDA_VISIBLE_DEVICES=0,1 pytest -v -s v1/shutdown
-  - pytest -v -s models/multimodal/generation/test_maverick.py

 - label: Plugin Tests (2 GPUs) # 40min
  mirror_hardwares: [amdexperimental]
@ -763,11 +665,32 @@ steps:
  - pytest -v -s models/test_oot_registration.py # it needs a clean process
  - pytest -v -s plugins/lora_resolvers # unit tests for in-tree lora resolver plugins

- label: Pipeline Parallelism Test # 45min
+- label: Multi-step Tests (4 GPUs) # 36min
  mirror_hardwares: [amdexperimental]
  working_dir: "/vllm-workspace/tests"
  num_gpus: 4
  source_file_dependencies:
+  - vllm/model_executor/layers/sampler.py
+  - vllm/sequence.py
+  - vllm/worker/worker_base.py
+  - vllm/worker/worker.py
+  - vllm/worker/multi_step_worker.py
+  - vllm/worker/model_runner_base.py
+  - vllm/worker/model_runner.py
+  - vllm/worker/multi_step_model_runner.py
+  - vllm/engine
+  - tests/multi_step
+  commands:
+  # this test is quite flaky
+  # TODO: investigate and fix.
+  # - pytest -v -s multi_step/test_correctness_async_llm.py
+  - pytest -v -s multi_step/test_correctness_llm.py
+
+- label: Pipeline Parallelism Test # 45min
+  mirror_hardwares: [amdexperimental, amdproduction]
+  working_dir: "/vllm-workspace/tests"
+  num_gpus: 4
+  source_file_dependencies:
  - vllm/distributed/
  - vllm/engine/
  - vllm/executor/
@ -778,7 +701,7 @@ steps:
  - pytest -v -s distributed/test_pipeline_parallel.py

 - label: LoRA TP Test (Distributed)
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental, amdproduction]
  num_gpus: 4
  source_file_dependencies:
  - vllm/lora
@ -791,7 +714,6 @@ steps:
    # requires multi-GPU testing for validation.
    - pytest -v -s -x lora/test_chatglm3_tp.py
    - pytest -v -s -x lora/test_llama_tp.py
-    - pytest -v -s -x lora/test_multi_loras_with_tp.py


 - label: Weight Loading Multiple GPU Test  # 33min
@ -805,7 +727,7 @@ steps:
    - bash weight_loading/run_model_weight_loading_test.sh -c weight_loading/models.txt

 - label: Weight Loading Multiple GPU Test - Large Models # optional
-  mirror_hardwares: [amdexperimental]
+  mirror_hardwares: [amdexperimental] 
  working_dir: "/vllm-workspace/tests"
  num_gpus: 2
  gpu: a100
@ -845,10 +767,3 @@ steps:
  commands:
  - export VLLM_WORKER_MULTIPROC_METHOD=spawn
  - pytest -s -v test_lm_eval_correctness.py --config-list-file=configs/models-large.txt --tp-size=4
-
- label: Qwen MoE EP Test # optional
-  gpu: h200
-  optional: true
-  num_gpus: 2
-  commands:
-    - CUDA_VISIBLE_DEVICES=1,2 VLLM_ALL2ALL_BACKEND=deepep_high_throughput VLLM_USE_DEEP_GEMM=1 VLLM_LOGGING_LEVEL=DEBUG python3 /vllm-workspace/examples/offline_inference/data_parallel.py --model Qwen/Qwen1.5-MoE-A2.7B --tp-size=1  --dp-size=2 --max-model-len 2048
--- a/.gemini/config.yaml
+++ b/.gemini/config.yaml
@ -1,6 +0,0 @@
-# https://developers.google.com/gemini-code-assist/docs/customize-gemini-behavior-github
-have_fun: false  # Just review the code
-code_review:
-  comment_severity_threshold: HIGH  # Reduce quantity of comments
-  pull_request_opened:
-    summary: false  # Don't summarize the PR in a separate comment
--- a/.github/CODEOWNERS
+++ b/.github/CODEOWNERS
@ -9,80 +9,40 @@
 /vllm/worker/worker_base.py @zhuohan123 @youkaichao @alexm-redhat @comaniac @njhill
 /vllm/worker/worker.py @zhuohan123 @youkaichao @alexm-redhat @comaniac @njhill
 /vllm/model_executor/layers/sampler.py @zhuohan123 @youkaichao @alexm-redhat @comaniac @njhill
-/vllm/model_executor/layers/quantization @mgoin @robertgshaw2-redhat @tlrmchlsmth @yewentao256
-/vllm/model_executor/layers/mamba @tdoublep
+/vllm/model_executor/layers/quantization @mgoin @robertgshaw2-redhat @tlrmchlsmth
+/vllm/model_executor/guided_decoding @mgoin @russellb
 /vllm/multimodal @DarkLight1337 @ywang96
 /vllm/vllm_flash_attn @LucasWilkinson
 /vllm/lora @jeejeelee
-/vllm/reasoning @aarnphm
-/vllm/entrypoints @aarnphm
-/vllm/compilation @zou3519 @youkaichao @ProExpertProg
-CMakeLists.txt @tlrmchlsmth @LucasWilkinson
-
-# Any change to the VllmConfig changes can have a large user-facing impact,
-# so spam a lot of people
-/vllm/config @simon-mo @WoosukKwon @youkaichao @robertgshaw2-redhat @mgoin @tlrmchlsmth @houseroad @hmellor @yewentao256 @ProExpertProg
+CMakeLists.txt @tlrmchlsmth

 # vLLM V1
 /vllm/v1 @WoosukKwon @robertgshaw2-redhat @njhill @ywang96 @comaniac @alexm-redhat
-/vllm/v1/structured_output @mgoin @russellb @aarnphm
-/vllm/v1/attention/backends/triton_attn.py @tdoublep
+/vllm/v1/structured_output @mgoin @russellb

 # Test ownership
 /.buildkite/lm-eval-harness @mgoin @simon-mo
 /tests/async_engine @njhill @robertgshaw2-redhat @simon-mo
+/tests/basic_correctness/test_chunked_prefill @rkooo567 @comaniac
 /tests/distributed/test_multi_node_assignment.py @youkaichao
 /tests/distributed/test_pipeline_parallel.py @youkaichao
 /tests/distributed/test_same_node.py @youkaichao
-/tests/entrypoints @DarkLight1337 @robertgshaw2-redhat @simon-mo @aarnphm
-/tests/kernels @tlrmchlsmth @WoosukKwon @yewentao256
+/tests/entrypoints @DarkLight1337 @robertgshaw2-redhat @simon-mo
+/tests/entrypoints/llm/test_guided_generate.py @mgoin @russellb
+/tests/kernels @tlrmchlsmth @WoosukKwon
+/tests/model_executor/test_guided_processors.py @mgoin @russellb
 /tests/models @DarkLight1337 @ywang96
+/tests/multi_step @alexm-redhat @comaniac
 /tests/multimodal @DarkLight1337 @ywang96
 /tests/prefix_caching @comaniac @KuntaiDu
-/tests/quantization @mgoin @robertgshaw2-redhat @yewentao256
+/tests/quantization @mgoin @robertgshaw2-redhat
+/tests/spec_decode @njhill @LiuXiaoxuanPKU
 /tests/test_inputs.py @DarkLight1337 @ywang96
-/tests/v1/entrypoints/llm/test_struct_output_generate.py @mgoin @russellb @aarnphm
-/tests/v1/structured_output @mgoin @russellb @aarnphm
-/tests/weight_loading @mgoin @youkaichao @yewentao256
+/tests/v1/entrypoints/llm/test_struct_output_generate.py @mgoin @russellb
+/tests/v1/structured_output @mgoin @russellb
+/tests/weight_loading @mgoin @youkaichao
 /tests/lora @jeejeelee
-/tests/models/language/generation/test_hybrid.py @tdoublep

 # Docs
 /docs @hmellor
-mkdocs.yaml @hmellor
-
-# CPU
-/vllm/v1/worker/^cpu @bigPYJ1151
-/csrc/cpu @bigPYJ1151
-/vllm/platforms/cpu.py @bigPYJ1151
-/cmake/cpu_extension.cmake @bigPYJ1151
-/docker/Dockerfile.cpu @bigPYJ1151
-
-# Intel GPU
-/vllm/v1/worker/^xpu @jikunshang
-/vllm/platforms/xpu.py @jikunshang
-/docker/Dockerfile.xpu @jikunshang
-
-# Qwen-specific files
-/vllm/attention/backends/dual_chunk_flash_attn.py @sighingnow
-/vllm/model_executor/models/qwen* @sighingnow
-
-# Mistral-specific files
-/vllm/model_executor/models/mistral*.py @patrickvonplaten
-/vllm/model_executor/models/mixtral*.py @patrickvonplaten
-/vllm/model_executor/models/voxtral*.py @patrickvonplaten
-/vllm/model_executor/models/pixtral*.py @patrickvonplaten
-/vllm/transformers_utils/configs/mistral.py @patrickvonplaten
-/vllm/transformers_utils/tokenizers/mistral.py @patrickvonplaten
-
-# Kernels
-/vllm/attention/ops/chunked_prefill_paged_decode.py @tdoublep
-/vllm/attention/ops/triton_unified_attention.py @tdoublep
-
-# ROCm related: specify owner with write access to notify AMD folks for careful code review
-/docker/Dockerfile.rocm* @gshtras
-/vllm/v1/attention/backends/rocm*.py @gshtras
-/vllm/v1/attention/backends/mla/rocm*.py @gshtras
-/vllm/attention/ops/rocm*.py @gshtras
-/vllm/model_executor/layers/fused_moe/rocm*.py @gshtras
-
+mkdocs.yaml @hmellor
--- a/.github/ISSUE_TEMPLATE/400-bug-report.yml
+++ b/.github/ISSUE_TEMPLATE/400-bug-report.yml
@ -8,16 +8,6 @@ body:
  attributes:
    value: >
      #### Before submitting an issue, please make sure the issue hasn't been already addressed by searching through [the existing and past issues](https://github.com/vllm-project/vllm/issues?q=is%3Aissue+sort%3Acreated-desc+).
- type: markdown
-  attributes:
-    value: |
-      ⚠️ **SECURITY WARNING:** Please review any text you paste to ensure it does not contain sensitive information such as:
-      - API tokens or keys (e.g., Hugging Face tokens, OpenAI API keys)
-      - Passwords or authentication credentials
-      - Private URLs or endpoints
-      - Personal or confidential data
-      
-      Consider redacting or replacing sensitive values with placeholders like `<YOUR_TOKEN_HERE>` when sharing configuration or code examples.
 - type: textarea
  attributes:
    label: Your current environment
--- a/.github/ISSUE_TEMPLATE/750-RFC.yml
+++ b/.github/ISSUE_TEMPLATE/750-RFC.yml
@ -46,7 +46,7 @@ body:
 - type: markdown
  attributes:
    value: >
-      Thanks for contributing 🎉! The vLLM core team hosts a biweekly RFC review session at 9:30AM Pacific Time, while most RFCs can be discussed online, you can optionally sign up for a slot to discuss your RFC online [here](https://docs.google.com/document/d/1CiLVBZeIVfR7_PNAKVSusxpceywkoOOB78qoWqHvSZc/edit).
+      Thanks for contributing 🎉!
 - type: checkboxes
  id: askllm
  attributes:
--- a/.github/PULL_REQUEST_TEMPLATE.md
+++ b/.github/PULL_REQUEST_TEMPLATE.md
@ -1,5 +1,9 @@
-<!-- markdownlint-disable -->
-PLEASE FILL IN THE PR DESCRIPTION HERE ENSURING ALL CHECKLIST ITEMS (AT THE BOTTOM) HAVE BEEN CONSIDERED.
+## Essential Elements of an Effective PR Description Checklist
+- [ ] The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
+- [ ] The test plan, such as providing test command.
+- [ ] The test results, such as pasting the results comparison before and after, or e2e results
+
+PLEASE FILL IN THE PR DESCRIPTION HERE ENSURING ALL CHECKLIST ITEMS ABOVE HAVE BEEN CONSIDERED.

 ## Purpose

@ -7,15 +11,5 @@ PLEASE FILL IN THE PR DESCRIPTION HERE ENSURING ALL CHECKLIST ITEMS (AT THE BOTT

 ## Test Result

---
-<details>
-<summary> Essential Elements of an Effective PR Description Checklist </summary>
-
- [ ] The purpose of the PR, such as "Fix some issue (link existing issues this PR will resolve)".
- [ ] The test plan, such as providing test command.
- [ ] The test results, such as pasting the results comparison before and after, or e2e results
- [ ] (Optional) The necessary documentation update, such as updating `supported_models.md` and `examples` for a new model.
- [ ] (Optional) Release notes update. If your change is user facing, please update the release notes draft in the [Google Doc](https://docs.google.com/document/d/1YyVqrgX4gHTtrstbq8oWUImOyPCKSGnJ7xtTpmXzlRs/edit?tab=t.0).
-</details>
-
+<!--- pyml disable-next-line no-emphasis-as-heading -->
 **BEFORE SUBMITTING, PLEASE READ <https://docs.vllm.ai/en/latest/contributing>** (anything written below this line will be removed by GitHub Actions)
--- a/.github/mergify.yml
+++ b/.github/mergify.yml
@ -27,22 +27,6 @@ pull_request_rules:
      add:
        - ci/build

- name: label-deepseek
-  description: Automatically apply deepseek label
-  conditions:
-    - or:
-      - files~=^examples/.*deepseek.*\.py
-      - files~=^tests/.*deepseek.*\.py
-      - files~=^vllm/entrypoints/openai/tool_parsers/.*deepseek.*\.py
-      - files~=^vllm/model_executor/models/.*deepseek.*\.py
-      - files~=^vllm/reasoning/.*deepseek.*\.py
-      - files~=^vllm/transformers_utils/.*deepseek.*\.py
-      - title~=(?i)DeepSeek
-  actions:
-    label:
-      add:
-        - deepseek
-
 - name: label-frontend
  description: Automatically apply frontend label
  conditions:
@ -52,21 +36,6 @@ pull_request_rules:
      add:
        - frontend

- name: label-llama
-  description: Automatically apply llama label
-  conditions:
-    - or:
-      - files~=^examples/.*llama.*\.py
-      - files~=^tests/.*llama.*\.py
-      - files~=^vllm/entrypoints/openai/tool_parsers/llama.*\.py
-      - files~=^vllm/model_executor/models/.*llama.*\.py
-      - files~=^vllm/transformers_utils/configs/.*llama.*\.py
-      - title~=(?i)llama
-  actions:
-    label:
-      add:
-        - llama
-
 - name: label-multi-modality
  description: Automatically apply multi-modality label
  conditions:
@ -74,84 +43,14 @@ pull_request_rules:
      - files~=^vllm/multimodal/
      - files~=^tests/multimodal/
      - files~=^tests/models/multimodal/
+      - files~=^tests/models/*/audio_language/
+      - files~=^tests/models/*/vision_language/
      - files=tests/models/test_vision.py
  actions:
    label:
      add:
        - multi-modality

- name: label-new-model
-  description: Automatically apply new-model label
-  conditions:
-    - and:
-      - files~=^vllm/model_executor/models/
-      - files=vllm/model_executor/models/registry.py
-  actions:
-    label:
-      add:
-        - new-model
-
- name: label-performance
-  description: Automatically apply performance label
-  conditions:
-    - or:
-      - files~=^benchmarks/
-      - files~=^vllm/benchmarks/
-      - files~=^tests/benchmarks/
-      - files~=^\.buildkite/nightly-benchmarks/
-  actions:
-    label:
-      add:
-        - performance
-
- name: label-qwen
-  description: Automatically apply qwen label
-  conditions:
-    - or:
-      - files~=^examples/.*qwen.*\.py
-      - files~=^tests/.*qwen.*\.py
-      - files~=^vllm/model_executor/models/.*qwen.*\.py
-      - files~=^vllm/reasoning/.*qwen.*\.py
-      - title~=(?i)Qwen
-  actions:
-    label:
-      add:
-        - qwen
-
- name: label-gpt-oss
-  description: Automatically apply gpt-oss label
-  conditions:
-    - or:
-      - files~=^examples/.*gpt[-_]?oss.*\.py
-      - files~=^tests/.*gpt[-_]?oss.*\.py
-      - files~=^vllm/model_executor/models/.*gpt[-_]?oss.*\.py
-      - files~=^vllm/model_executor/layers/.*gpt[-_]?oss.*\.py
-      - title~=(?i)gpt[-_]?oss
-  actions:
-    label:
-      add:
-        - gpt-oss
-
- name: label-rocm
-  description: Automatically apply rocm label
-  conditions:
-    - or:
-      - files~=^csrc/rocm/
-      - files~=^docker/Dockerfile.rocm
-      - files~=^requirements/rocm.*\.txt
-      - files~=^vllm/attention/backends/rocm.*\.py
-      - files~=^vllm/attention/ops/rocm.*\.py
-      - files~=^vllm/model_executor/layers/fused_moe/rocm.*\.py
-      - files~=^vllm/v1/attention/backends/mla/rocm.*\.py
-      - files~=^tests/kernels/.*_rocm.*\.py
-      - files=vllm/platforms/rocm.py
-      - title~=(?i)AMD
-      - title~=(?i)ROCm
-  actions:
-    label:
-      add:
-        - rocm
-
 - name: label-structured-output
  description: Automatically apply structured-output label
  conditions:
@ -163,6 +62,9 @@ pull_request_rules:
      - files=examples/offline_inference/structured_outputs.py
      - files=examples/online_serving/openai_chat_completion_structured_outputs.py
      - files=examples/online_serving/openai_chat_completion_structured_outputs_with_reasoning.py
+      - files~=^vllm/model_executor/guided_decoding/
+      - files=tests/model_executor/test_guided_processors.py
+      - files=tests/entrypoints/llm/test_guided_generate.py
      - files~=^tests/v1/structured_output/
      - files=tests/v1/entrypoints/llm/test_guided_generate.py
      - files~=^vllm/v1/structured_output/
@ -175,12 +77,9 @@ pull_request_rules:
  description: Automatically apply speculative-decoding label
  conditions:
    - or:
-      - files~=^vllm/v1/spec_decode/
-      - files~=^tests/v1/spec_decode/
-      - files~=^examples/.*(spec_decode|mlpspeculator|eagle|speculation).*\.py
-      - files~=^vllm/model_executor/models/.*eagle.*\.py
-      - files=vllm/model_executor/models/mlp_speculator.py
-      - files~=^vllm/transformers_utils/configs/(eagle|medusa|mlp_speculator)\.py
+      - files~=^vllm/spec_decode/
+      - files=vllm/model_executor/layers/spec_decode_base_sampler.py
+      - files~=^tests/spec_decode/
  actions:
    label:
      add:
--- a/.github/scripts/cleanup_pr_body.sh
+++ b/.github/scripts/cleanup_pr_body.sh
@ -15,11 +15,11 @@ NEW=/tmp/new_pr_body.txt
 gh pr view --json body --template "{{.body}}" "${PR_NUMBER}" > "${OLD}"
 cp "${OLD}" "${NEW}"

-# Remove markdown comments (like the <!-- markdownlint-disable --> at the start)
-sed -i '/<!--.*-->$/d' "${NEW}"
+# Remove "FIX #xxxx (*link existing issues this PR will resolve*)"
+sed -i '/FIX #xxxx.*$/d' "${NEW}"

-# Remove "PLEASE FILL IN THE PR DESCRIPTION HERE ENSURING ALL CHECKLIST ITEMS (AT THE BOTTOM) HAVE BEEN CONSIDERED."
-sed -i '/PLEASE FILL IN THE PR DESCRIPTION HERE.*$/d' "${NEW}"
+# Remove "FILL IN THE PR DESCRIPTION HERE"
+sed -i '/FILL IN THE PR DESCRIPTION HERE/d' "${NEW}"

 # Remove all lines after and including "**BEFORE SUBMITTING, PLEASE READ THE CHECKLIST BELOW AND FILL IN THE DESCRIPTION ABOVE**"
 sed -i '/\*\*BEFORE SUBMITTING, PLEASE READ.*\*\*/,$d' "${NEW}"
--- a/.github/workflows/issue_autolabel.yml
+++ b/.github/workflows/issue_autolabel.yml
@ -1,305 +0,0 @@
-name: Label issues based on keywords
-on:
-  issues:
-    types: [opened, edited, reopened]
-permissions:
-  issues: write          # needed so the workflow can add labels
-  contents: read
-concurrency:
-  group: issue-labeler-${{ github.event.issue.number }}
-  cancel-in-progress: true
-jobs:
-  add-labels:
-    runs-on: ubuntu-latest
-    steps:
-      - name: Label issues based on keywords
-        uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea  # v7.0.1
-        with:
-          script: |
-            // Configuration: Add new labels and keywords here
-            const labelConfig = {
-              rocm: {
-                // Keyword search - matches whole words only (with word boundaries)
-                keywords: [
-                  {
-                    term: "composable kernel",
-                    searchIn: "both"
-                  },
-                  {
-                    term: "rccl",
-                    searchIn: "body"  // only search in body
-                  },
-                  {
-                    term: "migraphx",
-                    searchIn: "title"  // only search in title
-                  },
-                  {
-                    term: "hipgraph",
-                    searchIn: "both"
-                  },
-                  {
-                    term: "ROCm System Management Interface",
-                    searchIn: "body"
-                  },
-                ],
-                
-                // Substring search - matches anywhere in text (partial matches)
-                substrings: [
-                  {
-                    term: "VLLM_ROCM_",
-                    searchIn: "both"
-                  },
-                  {
-                    term: "rocm",
-                    searchIn: "title"
-                  },
-                  {
-                    term: "amd",
-                    searchIn: "title"
-                  },
-                  {
-                    term: "hip-",
-                    searchIn: "both"
-                  },
-                  {
-                    term: "gfx",
-                    searchIn: "both"
-                  },
-                  {
-                    term: "cdna",
-                    searchIn: "both"
-                  },
-                  {
-                    term: "rdna",
-                    searchIn: "both"
-                  },
-                  {
-                    term: "torch_hip",
-                    searchIn: "body"  // only in body
-                  },
-                  {
-                    term: "_hip",
-                    searchIn: "both"
-                  },
-                  {
-                    term: "hip_",
-                    searchIn: "both"
-                  },
-                  
-                  // ROCm tools and libraries
-                  {
-                    term: "hipify",
-                    searchIn: "both"
-                  },
-                ],
-                
-                // Regex patterns - for complex pattern matching
-                regexPatterns: [
-                  {
-                    pattern: "\\bmi\\d{3}[a-z]*\\b",
-                    description: "AMD GPU names (mi + 3 digits + optional letters)",
-                    flags: "gi",
-                    searchIn: "both"  // "title", "body", or "both"
-                  }
-                ],
-              },
-            };
-            
-            // Helper function to create regex based on search type
-            function createSearchRegex(term, type) {
-              // Escape special regex characters in the term
-              const escapedTerm = term.replace(/[.*+?^${}()|[\]\\]/g, '\\$&');
-              
-              switch (type) {
-                case 'keyword':
-                  // Word boundary search - matches whole words only
-                  return new RegExp(`\\b${escapedTerm}\\b`, "gi");
-                case 'substring':
-                  // Substring search - matches anywhere in the text
-                  return new RegExp(escapedTerm, "gi");
-                default:
-                  throw new Error(`Unknown search type: ${type}`);
-              }
-            }
-            
-            // Helper function to find matching terms in text with line information
-            function findMatchingTermsWithLines(text, searchTerms = [], searchType = 'keyword', searchLocation = '') {
-              const matches = [];
-              const lines = text.split('\n');
-              
-              for (const termConfig of searchTerms) {
-                let regex;
-                let term, searchIn, pattern, description, flags;
-                
-                // Handle different input formats (string or object)
-                if (typeof termConfig === 'string') {
-                  term = termConfig;
-                  searchIn = 'both'; // default
-                } else {
-                  term = termConfig.term;
-                  searchIn = termConfig.searchIn || 'both';
-                  pattern = termConfig.pattern;
-                  description = termConfig.description;
-                  flags = termConfig.flags;
-                }
-                
-                // Skip if this term shouldn't be searched in the current location
-                if (searchIn !== 'both' && searchIn !== searchLocation) {
-                  continue;
-                }
-                
-                // Create appropriate regex
-                if (searchType === 'regex') {
-                  regex = new RegExp(pattern, flags || "gi");
-                } else {
-                  regex = createSearchRegex(term, searchType);
-                }
-                
-                const termMatches = [];
-                
-                // Check each line for matches
-                lines.forEach((line, lineIndex) => {
-                  const lineMatches = line.match(regex);
-                  if (lineMatches) {
-                    lineMatches.forEach(match => {
-                      termMatches.push({
-                        match: match,
-                        lineNumber: lineIndex + 1,
-                        lineContent: line.trim(),
-                        searchType: searchType,
-                        searchLocation: searchLocation,
-                        originalTerm: term || pattern,
-                        description: description,
-                        // Show context around the match in the line
-                        context: line.length > 100 ? 
-                          line.substring(Math.max(0, line.toLowerCase().indexOf(match.toLowerCase()) - 30), 
-                                       line.toLowerCase().indexOf(match.toLowerCase()) + match.length + 30) + '...' 
-                          : line.trim()
-                      });
-                    });
-                  }
-                });
-                
-                if (termMatches.length > 0) {
-                  matches.push({
-                    term: term || (description || pattern),
-                    searchType: searchType,
-                    searchLocation: searchLocation,
-                    searchIn: searchIn,
-                    pattern: pattern,
-                    matches: termMatches,
-                    count: termMatches.length
-                  });
-                }
-              }
-              
-              return matches;
-            }
-            
-            // Helper function to check if label should be added
-            async function processLabel(labelName, config) {
-              const body = context.payload.issue.body || "";
-              const title = context.payload.issue.title || "";
-              
-              core.notice(`Processing label: ${labelName}`);
-              core.notice(`Issue Title: "${title}"`);
-              core.notice(`Issue Body length: ${body.length} characters`);
-              
-              let shouldAddLabel = false;
-              let allMatches = [];
-              let reason = '';
-              
-              const keywords = config.keywords || [];
-              const substrings = config.substrings || [];
-              const regexPatterns = config.regexPatterns || [];
-              
-              core.notice(`Searching with ${keywords.length} keywords, ${substrings.length} substrings, and ${regexPatterns.length} regex patterns`);
-              
-              // Search in title
-              if (title.trim()) {
-                core.notice(`Searching in title: "${title}"`);
-                
-                const titleKeywordMatches = findMatchingTermsWithLines(title, keywords, 'keyword', 'title');
-                const titleSubstringMatches = findMatchingTermsWithLines(title, substrings, 'substring', 'title');
-                const titleRegexMatches = findMatchingTermsWithLines(title, regexPatterns, 'regex', 'title');
-                
-                allMatches.push(...titleKeywordMatches, ...titleSubstringMatches, ...titleRegexMatches);
-              }
-              
-              // Search in body
-              if (body.trim()) {
-                core.notice(`Searching in body (${body.length} characters)`);
-                
-                const bodyKeywordMatches = findMatchingTermsWithLines(body, keywords, 'keyword', 'body');
-                const bodySubstringMatches = findMatchingTermsWithLines(body, substrings, 'substring', 'body');
-                const bodyRegexMatches = findMatchingTermsWithLines(body, regexPatterns, 'regex', 'body');
-                
-                allMatches.push(...bodyKeywordMatches, ...bodySubstringMatches, ...bodyRegexMatches);
-              }
-              
-              if (allMatches.length > 0) {
-                core.notice(`Found ${allMatches.length} matching term(s):`);
-                
-                for (const termMatch of allMatches) {
-                  const locationText = termMatch.searchLocation === 'title' ? 'title' : 'body';
-                  const searchInText = termMatch.searchIn === 'both' ? 'both' : termMatch.searchIn;
-                  
-                  if (termMatch.searchType === 'regex') {
-                    core.notice(`  📍 Regex: "${termMatch.term}" (pattern: ${termMatch.pattern}) found ${termMatch.count} time(s) in ${locationText} (configured to search in: ${searchInText}):`);
-                  } else {
-                    core.notice(`  📍 Term: "${termMatch.term}" (${termMatch.searchType} search) found ${termMatch.count} time(s) in ${locationText} (configured to search in: ${searchInText}):`);
-                  }
-                  
-                  // Show details for each match
-                  termMatch.matches.forEach((match, index) => {
-                    core.notice(`    ${index + 1}. Line ${match.lineNumber} in ${match.searchLocation}: "${match.match}" [${match.searchType}]`);
-                    if (match.description) {
-                      core.notice(`       Description: ${match.description}`);
-                    }
-                    core.notice(`       Context: ${match.context}`);
-                    if (match.lineContent !== match.context) {
-                      core.notice(`       Full line: ${match.lineContent}`);
-                    }
-                  });
-                }
-                
-                shouldAddLabel = true;
-                const totalMatches = allMatches.reduce((sum, t) => sum + t.count, 0);
-                const titleMatches = allMatches.filter(t => t.searchLocation === 'title').reduce((sum, t) => sum + t.count, 0);
-                const bodyMatches = allMatches.filter(t => t.searchLocation === 'body').reduce((sum, t) => sum + t.count, 0);
-                const keywordMatches = allMatches.filter(t => t.searchType === 'keyword').reduce((sum, t) => sum + t.count, 0);
-                const substringMatches = allMatches.filter(t => t.searchType === 'substring').reduce((sum, t) => sum + t.count, 0);
-                const regexMatches = allMatches.filter(t => t.searchType === 'regex').reduce((sum, t) => sum + t.count, 0);
-                
-                reason = `Found ${totalMatches} total matches (${titleMatches} in title, ${bodyMatches} in body) - ${keywordMatches} keyword matches, ${substringMatches} substring matches, ${regexMatches} regex matches`;
-              }
-              
-              core.notice(`Final decision: ${shouldAddLabel ? 'ADD LABEL' : 'DO NOT ADD LABEL'}`);
-              core.notice(`Reason: ${reason || 'No matching terms found'}`);
-              
-              if (shouldAddLabel) {
-                const existingLabels = context.payload.issue.labels.map(l => l.name);
-                if (!existingLabels.includes(labelName)) {
-                  await github.rest.issues.addLabels({
-                    owner: context.repo.owner,
-                    repo: context.repo.repo,
-                    issue_number: context.issue.number,
-                    labels: [labelName],
-                  });
-                  core.notice(`Label "${labelName}" added. ${reason}`);
-                  return true;
-                }
-                core.notice(`Label "${labelName}" already present.`);
-                return false;
-              }
-              
-              core.notice(`No matching terms found for label "${labelName}".`);
-              return false;
-            }
-            
-            // Process all configured labels
-            const processLabels = Object.entries(labelConfig)
-              .map(([labelName, config]) => processLabel(labelName, config));
-            const labelsAdded = await Promise.all(processLabels);
-            const numLabelsAdded = labelsAdded.reduce((x, y) => x + y, 0);
-            core.notice(`Processing complete. ${numLabelsAdded} label(s) added.`);
--- a/.github/workflows/lint-and-deploy.yaml
+++ b/.github/workflows/lint-and-deploy.yaml
@ -0,0 +1,85 @@
+name: Lint and Deploy Charts
+
+on: pull_request
+
+permissions:
+  contents: read
+
+jobs:
+  lint-and-deploy:
+    runs-on: ubuntu-latest
+    steps:
+      - name: Checkout
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+        with:
+          fetch-depth: 0
+
+      - name: Set up Helm
+        uses: azure/setup-helm@b9e51907a09c216f16ebe8536097933489208112 # v4.3.0
+        with:
+          version: v3.14.4
+
+       #Python is required because ct lint runs Yamale and yamllint which require Python.
+      - uses: actions/setup-python@42375524e23c412d93fb67b49958b491fce71c38 # v5.4.0
+        with:
+          python-version: '3.13'
+
+      - name: Set up chart-testing
+        uses: helm/chart-testing-action@0d28d3144d3a25ea2cc349d6e59901c4ff469b3b # v2.7.0
+        with:
+          version: v3.10.1
+
+      - name: Run chart-testing (lint)
+        run: ct lint --target-branch ${{ github.event.repository.default_branch }} --chart-dirs examples/online_serving/chart-helm --charts examples/online_serving/chart-helm
+
+      - name: Setup minio
+        run: |
+          docker network create vllm-net
+          docker run -d -p 9000:9000 --name minio --net vllm-net \
+                     -e "MINIO_ACCESS_KEY=minioadmin" \
+                     -e "MINIO_SECRET_KEY=minioadmin" \
+                     -v /tmp/data:/data \
+                     -v /tmp/config:/root/.minio \
+                     minio/minio server /data
+          export AWS_ACCESS_KEY_ID=minioadmin
+          export AWS_SECRET_ACCESS_KEY=minioadmin
+          export AWS_EC2_METADATA_DISABLED=true
+          mkdir opt-125m
+          cd opt-125m && curl -O -Ls "https://huggingface.co/facebook/opt-125m/resolve/main/{pytorch_model.bin,config.json,generation_config.json,merges.txt,special_tokens_map.json,tokenizer_config.json,vocab.json}" && cd ..
+          aws --endpoint-url http://127.0.0.1:9000/ s3 mb s3://testbucket
+          aws --endpoint-url http://127.0.0.1:9000/ s3 cp opt-125m/ s3://testbucket/opt-125m --recursive
+
+      - name: Create kind cluster
+        uses: helm/kind-action@a1b0e391336a6ee6713a0583f8c6240d70863de3 # v1.12.0
+
+      - name: Build the Docker image vllm cpu
+        run: docker buildx build -f docker/Dockerfile.cpu -t vllm-cpu-env .
+
+      - name: Configuration of docker images, network and namespace for the kind cluster
+        run: |
+          docker pull amazon/aws-cli:2.6.4
+          kind load docker-image  amazon/aws-cli:2.6.4 --name chart-testing
+          kind load docker-image vllm-cpu-env:latest --name chart-testing
+          docker network connect vllm-net "$(docker ps -aqf "name=chart-testing-control-plane")"
+          kubectl create ns ns-vllm
+
+      - name: Run chart-testing (install)
+        run: |
+          export AWS_ACCESS_KEY_ID=minioadmin
+          export AWS_SECRET_ACCESS_KEY=minioadmin
+          sleep 30 && kubectl -n ns-vllm logs -f "$(kubectl -n ns-vllm get pods | awk '/deployment/ {print $1;exit}')" &
+          helm install --wait --wait-for-jobs --timeout 5m0s --debug --create-namespace --namespace=ns-vllm test-vllm examples/online_serving/chart-helm -f examples/online_serving/chart-helm/values.yaml --set secrets.s3endpoint=http://minio:9000 --set secrets.s3bucketname=testbucket --set secrets.s3accesskeyid=$AWS_ACCESS_KEY_ID --set secrets.s3accesskey=$AWS_SECRET_ACCESS_KEY --set resources.requests.cpu=1 --set resources.requests.memory=4Gi --set resources.limits.cpu=2 --set resources.limits.memory=5Gi --set image.env[0].name=VLLM_CPU_KVCACHE_SPACE --set image.env[1].name=VLLM_LOGGING_LEVEL --set-string image.env[0].value="1" --set-string image.env[1].value="DEBUG" --set-string extraInit.s3modelpath="opt-125m/" --set-string 'resources.limits.nvidia\.com/gpu=0' --set-string 'resources.requests.nvidia\.com/gpu=0' --set-string image.repository="vllm-cpu-env"
+
+      - name: curl test
+        run: |
+          kubectl -n ns-vllm port-forward service/test-vllm-service 8001:80 &
+          sleep 10
+          CODE="$(curl -v -f --location http://localhost:8001/v1/completions \
+                  --header "Content-Type: application/json" \
+                  --data '{
+                          "model": "opt-125m",
+                          "prompt": "San Francisco is a",
+                          "max_tokens": 7,
+                          "temperature": 0
+                  }'):$CODE"
+          echo "$CODE"
--- a/.github/workflows/matchers/markdownlint.json
+++ b/.github/workflows/matchers/markdownlint.json
@ -1,17 +0,0 @@
-{
-  "problemMatcher": [
-    {
-      "owner": "markdownlint",
-      "pattern": [
-        {
-          "regexp": "^([^:]*):(\\d+):?(\\d+)?\\s([\\w-\\/]*)\\s(.*)$",
-          "file": 1,
-          "line": 2,
-          "column": 3,
-          "code": 4,
-          "message": 5
-        }
-      ]
-    }
-  ]
-}
--- a/.github/workflows/pre-commit.yml
+++ b/.github/workflows/pre-commit.yml
@ -5,10 +5,6 @@ on:
  push:
    branches: [main]

-concurrency:
-  group: ${{ github.workflow }}-${{ github.ref }}
-  cancel-in-progress: ${{ github.event_name == 'pull_request' }}
-
 permissions:
  contents: read

@ -21,7 +17,6 @@ jobs:
      with:
        python-version: "3.12"
    - run: echo "::add-matcher::.github/workflows/matchers/actionlint.json"
-    - run: echo "::add-matcher::.github/workflows/matchers/markdownlint.json"
    - run: echo "::add-matcher::.github/workflows/matchers/mypy.json"
    - uses: pre-commit/action@2c7b3805fd2a0fd8c1884dcaebf91fc102a13ecd # v3.0.1
      with:
--- a/.github/workflows/publish.yml
+++ b/.github/workflows/publish.yml
@ -0,0 +1,111 @@
+# This workflow will upload a Python Package to Release asset
+# For more information see: https://help.github.com/en/actions/language-and-framework-guides/using-python-with-github-actions
+
+name: Create Release
+
+on:
+  push:
+    tags:
+      - v*
+
+# Needed to create release and upload assets
+permissions:
+  contents: write
+
+jobs:
+  release:
+    # Retrieve tag and create release
+    name: Create Release
+    runs-on: ubuntu-latest
+    outputs:
+      upload_url: ${{ steps.create_release.outputs.upload_url }}
+    steps:
+      - name: Checkout
+        uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+
+      - name: Extract branch info
+        shell: bash
+        run: |
+          echo "release_tag=${GITHUB_REF#refs/*/}" >> "$GITHUB_ENV"
+
+      - name: Create Release
+        id: create_release
+        uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
+        env:
+          RELEASE_TAG: ${{ env.release_tag }}
+        with:
+          github-token: "${{ secrets.GITHUB_TOKEN }}"
+          script: |
+            const script = require('.github/workflows/scripts/create_release.js')
+            await script(github, context, core)
+
+  # NOTE(simon): No longer build wheel using GitHub Actions. See buildkite's release workflow. 
+  # wheel:
+  #   name: Build Wheel
+  #   runs-on: ${{ matrix.os }}
+  #   needs: release
+
+  #   strategy:
+  #     fail-fast: false
+  #     matrix:
+  #         os: ['ubuntu-20.04']
+  #         python-version: ['3.9', '3.10', '3.11', '3.12']
+  #         pytorch-version: ['2.4.0']  # Must be the most recent version that meets requirements/cuda.txt.
+  #         cuda-version: ['11.8', '12.1']
+
+  #   steps:
+  #     - name: Checkout
+  #       uses: actions/checkout@11bd71901bbe5b1630ceea73d27597364c9af683 # v4.2.2
+
+  #     - name: Setup ccache
+  #       uses: hendrikmuhs/ccache-action@ed74d11c0b343532753ecead8a951bb09bb34bc9 # v1.2.14
+  #       with:
+  #         create-symlink: true
+  #         key: ${{ github.job }}-${{ matrix.python-version }}-${{ matrix.cuda-version }}
+
+  #     - name: Set up Linux Env
+  #       if: ${{ runner.os == 'Linux' }}
+  #       run: |
+  #         bash -x .github/workflows/scripts/env.sh
+
+  #     - name: Set up Python
+  #       uses: actions/setup-python@0b93645e9fea7318ecaed2b359559ac225c90a2b # v5.3.0
+  #       with:
+  #           python-version: ${{ matrix.python-version }}
+
+  #     - name: Install CUDA ${{ matrix.cuda-version }}
+  #       run: |
+  #         bash -x .github/workflows/scripts/cuda-install.sh ${{ matrix.cuda-version }} ${{ matrix.os }}
+
+  #     - name: Install PyTorch ${{ matrix.pytorch-version }} with CUDA ${{ matrix.cuda-version }}
+  #       run: |
+  #         bash -x .github/workflows/scripts/pytorch-install.sh ${{ matrix.python-version }} ${{ matrix.pytorch-version }} ${{ matrix.cuda-version }}
+
+  #     - name: Build wheel
+  #       shell: bash
+  #       env:
+  #         CMAKE_BUILD_TYPE: Release # do not compile with debug symbol to reduce wheel size
+  #       run: |
+  #         bash -x .github/workflows/scripts/build.sh ${{ matrix.python-version }} ${{ matrix.cuda-version }}
+  #         wheel_name=$(find dist -name "*whl" -print0 | xargs -0 -n 1 basename)
+  #         asset_name=${wheel_name//"linux"/"manylinux1"}
+  #         echo "wheel_name=${wheel_name}" >> "$GITHUB_ENV"
+  #         echo "asset_name=${asset_name}" >> "$GITHUB_ENV"
+
+  #     - name: Upload Release Asset
+  #       uses: actions/upload-release-asset@e8f9f06c4b078e705bd2ea027f0926603fc9b4d5 # v1.0.2
+  #       env:
+  #         GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+  #       with:
+  #         upload_url: ${{ needs.release.outputs.upload_url }}
+  #         asset_path: ./dist/${{ env.wheel_name }}
+  #         asset_name: ${{ env.asset_name }}
+  #         asset_content_type: application/*
+
+      # (Danielkinz): This last step will publish the .whl to pypi. Warning: untested
+      # - name: Publish package
+      #   uses: pypa/gh-action-pypi-publish@release/v1.8
+      #   with:
+      #     repository-url: https://test.pypi.org/legacy/
+      #     password: ${{ secrets.PYPI_API_TOKEN }}
+      #     skip-existing: true
--- a/.github/workflows/reminder_comment.yml
+++ b/.github/workflows/reminder_comment.yml
@ -12,43 +12,16 @@ jobs:
        uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v7.0.1
        with:
          script: |
-            try {
-              // Get the PR author
-              const prAuthor = context.payload.pull_request.user.login;
-              
-              // Check if this is the author's first PR in this repository
-              // Use GitHub's search API to find all PRs by this author
-              const { data: searchResults } = await github.rest.search.issuesAndPullRequests({
-                q: `repo:${context.repo.owner}/${context.repo.repo} type:pr author:${prAuthor}`,
-                per_page: 100  
-              });
-              
-              const authorPRCount = searchResults.total_count;
-              
-              console.log(`Found ${authorPRCount} PRs by ${prAuthor}`);
-              
-              // Only post comment if this is the first PR (only one PR by this author)
-              if (authorPRCount === 1) {
-                console.log(`Posting welcome comment for first-time contributor: ${prAuthor}`);
-                await github.rest.issues.createComment({
-                owner: context.repo.owner,
-                repo: context.repo.repo,
-                issue_number: context.issue.number,
-                body: '👋 Hi! Thank you for contributing to the vLLM project.\n\n' +
-                  '💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.\n\n' +
-                  'Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run `fastcheck` CI which starts running only a small and essential subset of CI tests to quickly catch errors. \n\n' +
-                  'You ask your reviewers to trigger select CI tests on top of `fastcheck` CI. \n\n' +
-                  'Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.\n\n' +
-                  'To run CI, PR reviewers can either: Add `ready` label to the PR or enable auto-merge.\n\n' +
-                  'If you have any questions, please reach out to us on Slack at https://slack.vllm.ai.\n\n' +
-                  '🚀'
-                });
-              } else {
-                console.log(`Skipping comment for ${prAuthor} - not their first PR (${authorPRCount} PRs found)`);
-              }
-            } catch (error) {
-              console.error('Error checking PR history or posting comment:', error);
-              // Don't fail the workflow, just log the error
-            }
+            github.rest.issues.createComment({
+              owner: context.repo.owner,
+              repo: context.repo.repo,
+              issue_number: context.issue.number,
+              body: '👋 Hi! Thank you for contributing to the vLLM project.\n\n' +
+                '💬 Join our developer Slack at https://slack.vllm.ai to discuss your PR in #pr-reviews, coordinate on features in #feat- channels, or join special interest groups in #sig- channels.\n\n' +
+                'Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run `fastcheck` CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your `fastcheck` build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping `simon-mo` or `khluu` to add you in our Buildkite org.\n\n' +
+                'Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.\n\n' +
+                'To run CI, PR reviewers can either: Add `ready` label to the PR or enable auto-merge.\n\n' +
+                '🚀'
+            })
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
--- a/.github/workflows/scripts/build.sh
+++ b/.github/workflows/scripts/build.sh
@ -15,6 +15,7 @@ $python_executable -m pip install -r requirements/build.txt -r requirements/cuda
 export MAX_JOBS=1
 # Make sure release wheels are built for the following architectures
 export TORCH_CUDA_ARCH_LIST="7.0 7.5 8.0 8.6 8.9 9.0+PTX"
+export VLLM_FA_CMAKE_GPU_ARCHES="80-real;90-real"

 bash tools/check_repo.sh

--- a/.gitignore
+++ b/.gitignore
@ -4,9 +4,6 @@
 # vllm-flash-attn built from source
 vllm/vllm_flash_attn/*

-# triton jit 
-.triton
-
 # Byte-compiled / optimized / DLL files
 __pycache__/
 *.py[cod]
@ -149,9 +146,7 @@ venv.bak/

 # mkdocs documentation
 /site
-docs/argparse
-docs/examples/*
-!docs/examples/README.md
+docs/examples

 # mypy
 .mypy_cache/
@ -205,8 +200,5 @@ benchmarks/**/*.json
 actionlint
 shellcheck*/

-# Ignore moe/marlin_moe gen code
+# Ingore moe/marlin_moe gen code
 csrc/moe/marlin_moe_wna16/kernel_*
-
-# Ignore ep_kernels_workspace folder
-ep_kernels_workspace/
--- a/.markdownlint.yaml
+++ b/.markdownlint.yaml
@ -1,13 +0,0 @@
-MD007:
-  indent: 4
-MD013: false
-MD024:
-  siblings_only: true
-MD033: false
-MD042: false
-MD045: false
-MD046: false
-MD051: false
-MD052: false
-MD053: false
-MD059: false
--- a/.pre-commit-config.yaml
+++ b/.pre-commit-config.yaml
@ -11,8 +11,6 @@ repos:
  hooks:
  - id: yapf
    args: [--in-place, --verbose]
-    # Keep the same list from yapfignore here to avoid yapf failing without any inputs
-    exclude: '(.buildkite|benchmarks|build|examples)/.*'
 - repo: https://github.com/astral-sh/ruff-pre-commit
  rev: v0.11.7
  hooks:
@ -20,10 +18,12 @@ repos:
    args: [--output-format, github, --fix]
  - id: ruff-format
    files: ^(.buildkite|benchmarks|examples)/.*
- repo: https://github.com/crate-ci/typos
-  rev: v1.35.5
+- repo: https://github.com/codespell-project/codespell
+  rev: v2.4.1
  hooks:
-  - id: typos
+  - id: codespell
+    additional_dependencies: ['tomli']
+    args: ['--toml', 'pyproject.toml']
 - repo: https://github.com/PyCQA/isort
  rev: 6.0.1
  hooks:
@ -35,12 +35,12 @@ repos:
    exclude: 'csrc/(moe/topk_softmax_kernels.cu|quantization/gguf/(ggml-common.h|dequantize.cuh|vecdotq.cuh|mmq.cuh|mmvq.cuh))|vllm/third_party/.*'
    types_or: [c++, cuda]
    args: [--style=file, --verbose]
- repo: https://github.com/igorshubovych/markdownlint-cli
-  rev: v0.45.0
+- repo: https://github.com/jackdewinter/pymarkdown
+  rev: v0.9.29
  hooks:
-  - id: markdownlint
+  - id: pymarkdown
    exclude: '.*\.inc\.md'
-    stages: [manual] # Only run in CI
+    args: [fix]
 - repo: https://github.com/rhysd/actionlint
  rev: v1.7.7
  hooks:
@ -53,11 +53,6 @@ repos:
      files: ^requirements/test\.(in|txt)$
 - repo: local
  hooks:
-  - id: format-torch-nightly-test
-    name: reformat nightly_torch_test.txt to be in sync with test.in
-    language: python
-    entry: python tools/generate_nightly_torch_test.py
-    files: ^requirements/test\.(in|txt)$
  - id: mypy-local
    name: Run mypy for local Python installation
    entry: tools/mypy.sh 0 "local"
@ -120,11 +115,6 @@ repos:
    entry: python tools/check_spdx_header.py
    language: python
    types: [python]
-  - id: check-root-lazy-imports
-    name: Check root lazy imports
-    entry: python tools/check_init_lazy_imports.py
-    language: python
-    types: [python]
  - id: check-filenames
    name: Check for spaces in all filenames
    entry: bash
@ -153,24 +143,10 @@ repos:
    types: [python]
    pass_filenames: false
    additional_dependencies: [regex]
-  - id: check-pickle-imports
-    name: Prevent new pickle/cloudpickle imports
-    entry: python tools/check_pickle_imports.py
-    language: python
-    types: [python]
-    pass_filenames: false
-    additional_dependencies: [pathspec, regex]
-  - id: validate-config
-    name: Validate configuration has default values and that each field has a docstring
-    entry: python tools/validate_config.py
-    language: python
-    types: [python]
-    pass_filenames: true
-    files: vllm/config.py|tests/test_config.py|vllm/entrypoints/openai/cli_args.py
  # Keep `suggestion` last
  - id: suggestion
    name: Suggestion
-    entry: bash -c 'echo "To bypass all the pre-commit hooks, add --no-verify to git commit. To skip a specific hook, prefix the commit command with SKIP=<hook-id>."'
+    entry: bash -c 'echo "To bypass pre-commit hooks, add --no-verify to git commit."'
    language: system
    verbose: true
    pass_filenames: false
--- a/.readthedocs.yaml
+++ b/.readthedocs.yaml
@ -7,9 +7,6 @@ build:
  os: ubuntu-22.04
  tools:
    python: "3.12"
-  jobs:
-    post_checkout:
-      - git fetch --unshallow || true

 mkdocs:
  configuration: mkdocs.yaml
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@ -30,7 +30,7 @@ install(CODE "set(CMAKE_INSTALL_LOCAL_ONLY TRUE)" ALL_COMPONENTS)
 # Supported python versions.  These versions will be searched in order, the
 # first match will be selected.  These should be kept in sync with setup.py.
 #
-set(PYTHON_SUPPORTED_VERSIONS "3.9" "3.10" "3.11" "3.12" "3.13")
+set(PYTHON_SUPPORTED_VERSIONS "3.9" "3.10" "3.11" "3.12")

 # Supported AMD GPU architectures.
 set(HIP_SUPPORTED_ARCHS "gfx906;gfx908;gfx90a;gfx942;gfx950;gfx1030;gfx1100;gfx1101;gfx1200;gfx1201")
@ -45,7 +45,7 @@ set(HIP_SUPPORTED_ARCHS "gfx906;gfx908;gfx90a;gfx942;gfx950;gfx1030;gfx1100;gfx1
 # requirements.txt files and should be kept consistent.  The ROCm torch
 # versions are derived from docker/Dockerfile.rocm
 #
-set(TORCH_SUPPORTED_VERSION_CUDA "2.7.1")
+set(TORCH_SUPPORTED_VERSION_CUDA "2.7.0")
 set(TORCH_SUPPORTED_VERSION_ROCM "2.7.0")

 #
@ -171,6 +171,7 @@ if(NVCC_THREADS AND VLLM_GPU_LANG STREQUAL "CUDA")
  list(APPEND VLLM_GPU_FLAGS "--threads=${NVCC_THREADS}")
 endif()

+
 #
 # Use FetchContent for C++ dependencies that are compiled as part of vLLM's build process.
 # setup.py will override FETCHCONTENT_BASE_DIR to play nicely with sccache.
@ -231,6 +232,7 @@ endif()

 set(VLLM_EXT_SRC
  "csrc/mamba/mamba_ssm/selective_scan_fwd.cu"
+  "csrc/mamba/causal_conv1d/causal_conv1d.cu"
  "csrc/cache_kernels.cu"
  "csrc/attention/paged_attention_v1.cu"
  "csrc/attention/paged_attention_v2.cu"
@ -249,6 +251,7 @@ set(VLLM_EXT_SRC
  "csrc/quantization/gguf/gguf_kernel.cu"
  "csrc/quantization/activation_kernels.cu"
  "csrc/cuda_utils_kernels.cu"
+  "csrc/prepare_inputs/advance_step.cu"
  "csrc/custom_all_reduce.cu"
  "csrc/torch_bindings.cpp")

@ -256,7 +259,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
  SET(CUTLASS_ENABLE_HEADERS_ONLY ON CACHE BOOL "Enable only the header library")

  # Set CUTLASS_REVISION. Used for FetchContent. Also fixes some bogus messages when building.
-  set(CUTLASS_REVISION "v4.0.0" CACHE STRING "CUTLASS revision to use")
+  set(CUTLASS_REVISION "v3.9.2" CACHE STRING "CUTLASS revision to use")

  # Use the specified CUTLASS source directory for compilation if VLLM_CUTLASS_SRC_DIR is provided
  if (DEFINED ENV{VLLM_CUTLASS_SRC_DIR})
@ -286,6 +289,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
  FetchContent_MakeAvailable(cutlass)

  list(APPEND VLLM_EXT_SRC
+    "csrc/quantization/aqlm/gemm_kernels.cu"
    "csrc/quantization/awq/gemm_kernels.cu"
    "csrc/permute_cols.cu"
    "csrc/quantization/cutlass_w8a8/scaled_mm_entry.cu"
@ -294,8 +298,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
    "csrc/quantization/fp4/nvfp4_blockwise_moe_kernel.cu"
    "csrc/sparse/cutlass/sparse_scaled_mm_entry.cu"
    "csrc/cutlass_extensions/common.cpp"
-    "csrc/attention/mla/cutlass_mla_entry.cu"
-    "csrc/quantization/fp8/per_token_group_quant.cu")
+    "csrc/attention/mla/cutlass_mla_entry.cu")

  set_gencode_flags_for_srcs(
    SRCS "${VLLM_EXT_SRC}"
@ -305,7 +308,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
  # Keep building Marlin for 9.0 as there are some group sizes and shapes that
  # are not supported by Machete yet.
  # 9.0 for latest bf16 atomicAdd PTX
-  cuda_archs_loose_intersection(MARLIN_ARCHS "8.0;8.7;9.0+PTX" "${CUDA_ARCHS}")
+  cuda_archs_loose_intersection(MARLIN_ARCHS "8.0;9.0+PTX" "${CUDA_ARCHS}")
  if (MARLIN_ARCHS)

    #
@ -349,27 +352,20 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
    set_gencode_flags_for_srcs(
      SRCS "${MARLIN_TEMPLATE_KERNEL_SRC}"
      CUDA_ARCHS "${MARLIN_ARCHS}")
-    if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8)
-      set_source_files_properties(${MARLIN_TEMPLATE_KERNEL_SRC}
-        PROPERTIES COMPILE_FLAGS "-static-global-template-stub=false")
-    endif()

    list(APPEND VLLM_EXT_SRC ${MARLIN_TEMPLATE_KERNEL_SRC})

    set(MARLIN_SRCS
+       "csrc/quantization/marlin/dense/marlin_cuda_kernel.cu"
       "csrc/quantization/marlin/sparse/marlin_24_cuda_kernel.cu"
+       "csrc/quantization/marlin/qqq/marlin_qqq_gemm_kernel.cu"
       "csrc/quantization/gptq_marlin/gptq_marlin.cu"
       "csrc/quantization/gptq_marlin/gptq_marlin_repack.cu"
       "csrc/quantization/gptq_marlin/awq_marlin_repack.cu")
    set_gencode_flags_for_srcs(
      SRCS "${MARLIN_SRCS}"
      CUDA_ARCHS "${MARLIN_ARCHS}")
-    if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8)
-      set_source_files_properties("csrc/quantization/gptq_marlin/gptq_marlin.cu"
-        PROPERTIES COMPILE_FLAGS "-static-global-template-stub=false")
-    endif()
    list(APPEND VLLM_EXT_SRC "${MARLIN_SRCS}")
-
    message(STATUS "Building Marlin kernels for archs: ${MARLIN_ARCHS}")
  else()
    message(STATUS "Not building Marlin kernels as no compatible archs found"
@ -397,7 +393,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
  # The cutlass_scaled_mm kernels for Hopper (c3x, i.e. CUTLASS 3.x) require
  # CUDA 12.0 or later
  cuda_archs_loose_intersection(SCALED_MM_ARCHS "9.0a;" "${CUDA_ARCHS}")
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.0 AND SCALED_MM_ARCHS)
+  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER 12.0 AND SCALED_MM_ARCHS)
    set(SRCS
       "csrc/quantization/cutlass_w8a8/scaled_mm_c3x_sm90.cu"
       "csrc/quantization/cutlass_w8a8/c3x/scaled_mm_sm90_fp8.cu"
@ -413,7 +409,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
    list(APPEND SCALED_MM_3X_ARCHS "${SCALED_MM_ARCHS}")
    message(STATUS "Building scaled_mm_c3x_sm90 for archs: ${SCALED_MM_ARCHS}")
  else()
-    if (NOT ${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.0 AND SCALED_MM_ARCHS)
+    if (NOT ${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER 12.0 AND SCALED_MM_ARCHS)
      message(STATUS "Not building scaled_mm_c3x_sm90 as CUDA Compiler version is "
                     "not >= 12.0, we recommend upgrading to CUDA 12.0 or "
                     "later if you intend on running FP8 quantized models on "
@ -424,41 +420,10 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
    endif()
  endif()

-
-  # The cutlass_scaled_mm kernels for Geforce Blackwell SM120 (c3x, i.e. CUTLASS 3.x) require
+  # The cutlass_scaled_mm kernels for Blackwell (c3x, i.e. CUTLASS 3.x) require
  # CUDA 12.8 or later
-  cuda_archs_loose_intersection(SCALED_MM_ARCHS "12.0;12.0a" "${CUDA_ARCHS}")
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8 AND SCALED_MM_ARCHS)
-    set(SRCS
-      "csrc/quantization/cutlass_w8a8/scaled_mm_c3x_sm120.cu"
-      "csrc/quantization/cutlass_w8a8/c3x/scaled_mm_sm120_fp8.cu"
-      "csrc/quantization/cutlass_w8a8/c3x/scaled_mm_blockwise_sm120_fp8.cu"
-    )
-    set_gencode_flags_for_srcs(
-      SRCS "${SRCS}"
-      CUDA_ARCHS "${SCALED_MM_ARCHS}")
-    list(APPEND VLLM_EXT_SRC "${SRCS}")
-    list(APPEND VLLM_GPU_FLAGS "-DENABLE_SCALED_MM_SM120=1")
-    # Let scaled_mm_c2x know it doesn't need to build these arches
-    list(APPEND SCALED_MM_3X_ARCHS "${SCALED_MM_ARCHS}")
-    message(STATUS "Building scaled_mm_c3x_sm120 for archs: ${SCALED_MM_ARCHS}")
-  else()
-    if (NOT ${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8 AND SCALED_MM_ARCHS)
-      message(STATUS "Not building scaled_mm_c3x_sm120 as CUDA Compiler version is "
-                     "not >= 12.8, we recommend upgrading to CUDA 12.8 or "
-                     "later if you intend on running FP8 quantized models on "
-                     "Blackwell.")
-    else()
-      message(STATUS "Not building scaled_mm_c3x_120 as no compatible archs found "
-                     "in CUDA target architectures")
-    endif()
-  endif()
-
-
-  # The cutlass_scaled_mm kernels for Blackwell SM100 (c3x, i.e. CUTLASS 3.x)
-  # require CUDA 12.8 or later
-  cuda_archs_loose_intersection(SCALED_MM_ARCHS "10.0a;10.1a" "${CUDA_ARCHS}")
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8 AND SCALED_MM_ARCHS)
+  cuda_archs_loose_intersection(SCALED_MM_ARCHS "10.0a;10.1a;12.0a" "${CUDA_ARCHS}")
+  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER 12.8 AND SCALED_MM_ARCHS)
    set(SRCS
      "csrc/quantization/cutlass_w8a8/scaled_mm_c3x_sm100.cu"
      "csrc/quantization/cutlass_w8a8/c3x/scaled_mm_sm100_fp8.cu"
@ -473,7 +438,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
    list(APPEND SCALED_MM_3X_ARCHS "${SCALED_MM_ARCHS}")
    message(STATUS "Building scaled_mm_c3x_sm100 for archs: ${SCALED_MM_ARCHS}")
  else()
-    if (NOT ${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8 AND SCALED_MM_ARCHS)
+    if (NOT ${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER 12.8 AND SCALED_MM_ARCHS)
      message(STATUS "Not building scaled_mm_c3x_sm100 as CUDA Compiler version is "
                     "not >= 12.8, we recommend upgrading to CUDA 12.8 or "
                     "later if you intend on running FP8 quantized models on "
@ -489,7 +454,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
  # kernels for the remaining archs that are not already built for 3x.
  # (Build 8.9 for FP8)
  cuda_archs_loose_intersection(SCALED_MM_2X_ARCHS
-    "7.5;8.0;8.7;8.9+PTX" "${CUDA_ARCHS}")
+    "7.5;8.0;8.9+PTX" "${CUDA_ARCHS}")
  # subtract out the archs that are already built for 3x
  list(REMOVE_ITEM SCALED_MM_2X_ARCHS ${SCALED_MM_3X_ARCHS})
  if (SCALED_MM_2X_ARCHS)
@ -516,7 +481,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
  # The 2:4 sparse kernels cutlass_scaled_sparse_mm and cutlass_compressor
  # require CUDA 12.2 or later (and only work on Hopper).
  cuda_archs_loose_intersection(SCALED_MM_ARCHS "9.0a;" "${CUDA_ARCHS}")
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.2 AND SCALED_MM_ARCHS)
+  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER 12.2 AND SCALED_MM_ARCHS)
    set(SRCS "csrc/sparse/cutlass/sparse_scaled_mm_c3x.cu")
    set_gencode_flags_for_srcs(
      SRCS "${SRCS}"
@ -525,7 +490,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
    list(APPEND VLLM_GPU_FLAGS "-DENABLE_SPARSE_SCALED_MM_C3X=1")
    message(STATUS "Building sparse_scaled_mm_c3x for archs: ${SCALED_MM_ARCHS}")
  else()
-    if (NOT ${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.2 AND SCALED_MM_ARCHS)
+    if (NOT ${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER 12.2 AND SCALED_MM_ARCHS)
      message(STATUS "Not building sparse_scaled_mm_c3x kernels as CUDA Compiler version is "
                     "not >= 12.2, we recommend upgrading to CUDA 12.2 or later "
                     "if you intend on running FP8 sparse quantized models on Hopper.")
@ -535,28 +500,9 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
    endif()
  endif()

-  # The nvfp4_scaled_mm_sm120 kernels for Geforce Blackwell SM120 require
-  # CUDA 12.8 or later
-  cuda_archs_loose_intersection(FP4_ARCHS "12.0;12.0a" "${CUDA_ARCHS}")
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8 AND FP4_ARCHS)
-    set(SRCS
-      "csrc/quantization/fp4/nvfp4_quant_kernels.cu"
-      "csrc/quantization/fp4/nvfp4_scaled_mm_sm120_kernels.cu")
-    set_gencode_flags_for_srcs(
-      SRCS "${SRCS}"
-      CUDA_ARCHS "${FP4_ARCHS}")
-    list(APPEND VLLM_EXT_SRC "${SRCS}")
-    list(APPEND VLLM_GPU_FLAGS "-DENABLE_NVFP4_SM120=1")
-    message(STATUS "Building NVFP4 for archs: ${FP4_ARCHS}")
-  else()
-    message(STATUS "Not building NVFP4 as no compatible archs were found.")
-    # clear FP4_ARCHS
-    set(FP4_ARCHS)
-  endif()
-
  # FP4 Archs and flags
  cuda_archs_loose_intersection(FP4_ARCHS "10.0a" "${CUDA_ARCHS}")
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8 AND FP4_ARCHS)
+  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER 12.8 AND FP4_ARCHS)
    set(SRCS
      "csrc/quantization/fp4/nvfp4_quant_kernels.cu"
      "csrc/quantization/fp4/nvfp4_experts_quant.cu"
@ -566,8 +512,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
      SRCS "${SRCS}"
      CUDA_ARCHS "${FP4_ARCHS}")
    list(APPEND VLLM_EXT_SRC "${SRCS}")
-    list(APPEND VLLM_GPU_FLAGS "-DENABLE_NVFP4_SM100=1")
-    list(APPEND VLLM_GPU_FLAGS "-DENABLE_CUTLASS_MOE_SM100=1")
+    list(APPEND VLLM_GPU_FLAGS "-DENABLE_NVFP4=1")
    message(STATUS "Building NVFP4 for archs: ${FP4_ARCHS}")
  else()
    message(STATUS "Not building NVFP4 as no compatible archs were found.")
@ -577,10 +522,9 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")

  # CUTLASS MLA Archs and flags
  cuda_archs_loose_intersection(MLA_ARCHS "10.0a" "${CUDA_ARCHS}")
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8 AND MLA_ARCHS)
+  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER 12.8 AND MLA_ARCHS)
    set(SRCS
-      "csrc/attention/mla/cutlass_mla_kernels.cu"
-      "csrc/attention/mla/sm100_cutlass_mla_kernel.cu")
+      "csrc/attention/mla/cutlass_mla_kernels.cu")
    set_gencode_flags_for_srcs(
      SRCS "${SRCS}"
      CUDA_ARCHS "${MLA_ARCHS}")
@ -598,12 +542,13 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")

  # CUTLASS MoE kernels

-  # The MoE kernel cutlass_moe_mm requires CUDA 12.3 or later (and ONLY works
-  # on Hopper). get_cutlass_(pplx_)moe_mm_data should only be compiled
-  # if it's possible to compile MoE kernels that use its output.
-  cuda_archs_loose_intersection(SCALED_MM_ARCHS "9.0a" "${CUDA_ARCHS}")
+  # The MoE kernel cutlass_moe_mm requires CUDA 12.3 or later (and only works
+  # on Hopper). get_cutlass_moe_mm_data should only be compiled if it's possible
+  # to compile MoE kernels that use its output.
+  cuda_archs_loose_intersection(SCALED_MM_ARCHS "9.0a;10.0a" "${CUDA_ARCHS}")
  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.3 AND SCALED_MM_ARCHS)
-    set(SRCS "csrc/quantization/cutlass_w8a8/moe/grouped_mm_c3x_sm90.cu")
+    set(SRCS "csrc/quantization/cutlass_w8a8/moe/grouped_mm_c3x.cu"
+             "csrc/quantization/cutlass_w8a8/moe/moe_data.cu")
    set_gencode_flags_for_srcs(
      SRCS "${SRCS}"
      CUDA_ARCHS "${SCALED_MM_ARCHS}")
@ -617,66 +562,6 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
                     "if you intend on running FP8 quantized MoE models on Hopper.")
    else()
      message(STATUS "Not building grouped_mm_c3x as no compatible archs found "
-                     "in CUDA target architectures.")
-    endif()
-  endif()
-
-  cuda_archs_loose_intersection(SCALED_MM_ARCHS "10.0a" "${CUDA_ARCHS}")
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8 AND SCALED_MM_ARCHS)
-    set(SRCS "csrc/quantization/cutlass_w8a8/moe/grouped_mm_c3x_sm100.cu")
-    set_gencode_flags_for_srcs(
-      SRCS "${SRCS}"
-      CUDA_ARCHS "${SCALED_MM_ARCHS}")
-    list(APPEND VLLM_EXT_SRC "${SRCS}")
-    list(APPEND VLLM_GPU_FLAGS "-DENABLE_CUTLASS_MOE_SM100=1")
-    message(STATUS "Building grouped_mm_c3x for archs: ${SCALED_MM_ARCHS}")
-  else()
-    if (NOT ${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8 AND SCALED_MM_ARCHS)
-      message(STATUS "Not building grouped_mm_c3x kernels as CUDA Compiler version is "
-                     "not >= 12.8, we recommend upgrading to CUDA 12.8 or later "
-                     "if you intend on running FP8 quantized MoE models on Blackwell.")
-    else()
-      message(STATUS "Not building grouped_mm_c3x as no compatible archs found "
-                     "in CUDA target architectures.")
-    endif()
-  endif()
-
-  # moe_data.cu is used by all CUTLASS MoE kernels.
-  cuda_archs_loose_intersection(CUTLASS_MOE_DATA_ARCHS "9.0a;10.0a" "${CUDA_ARCHS}")
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.3 AND CUTLASS_MOE_DATA_ARCHS)
-    set(SRCS "csrc/quantization/cutlass_w8a8/moe/moe_data.cu")
-    set_gencode_flags_for_srcs(
-      SRCS "${SRCS}"
-      CUDA_ARCHS "${CUTLASS_MOE_DATA_ARCHS}")
-    list(APPEND VLLM_EXT_SRC "${SRCS}")
-    message(STATUS "Building moe_data for archs: ${CUTLASS_MOE_DATA_ARCHS}")
-  else()
-    if (NOT ${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.3 AND CUTLASS_MOE_DATA_ARCHS)
-      message(STATUS "Not building moe_data as CUDA Compiler version is "
-                     "not >= 12.3, we recommend upgrading to CUDA 12.3 or later "
-                     "if you intend on running FP8 quantized MoE models on Hopper or Blackwell.")
-    else()
-      message(STATUS "Not building moe_data as no compatible archs found "
-                     "in CUDA target architectures.")
-    endif()
-  endif()
-
-  cuda_archs_loose_intersection(SCALED_MM_ARCHS "10.0a" "${CUDA_ARCHS}")
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8 AND SCALED_MM_ARCHS)
-    set(SRCS "csrc/quantization/cutlass_w8a8/moe/blockwise_scaled_group_mm_sm100.cu")
-    set_gencode_flags_for_srcs(
-      SRCS "${SRCS}"
-      CUDA_ARCHS "${SCALED_MM_ARCHS}")
-    list(APPEND VLLM_EXT_SRC "${SRCS}")
-    list(APPEND VLLM_GPU_FLAGS "-DENABLE_CUTLASS_MOE_SM100=1")
-    message(STATUS "Building blockwise_scaled_group_mm_sm100 for archs: ${SCALED_MM_ARCHS}")
-  else()
-    if (NOT ${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8 AND SCALED_MM_ARCHS)
-      message(STATUS "Not building blockwise_scaled_group_mm_sm100 kernels as CUDA Compiler version is "
-                     "not >= 12.8, we recommend upgrading to CUDA 12.8 or later "
-                     "if you intend on running FP8 quantized MoE models on Blackwell.")
-    else()
-      message(STATUS "Not building blockwise_scaled_group_mm_sm100 as no compatible archs found "
                     "in CUDA target architectures")
    endif()
  endif()
@ -687,7 +572,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
  # The machete kernels only work on hopper and require CUDA 12.0 or later.
  # Only build Machete kernels if we are building for something compatible with sm90a
  cuda_archs_loose_intersection(MACHETE_ARCHS "9.0a" "${CUDA_ARCHS}")
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.0 AND MACHETE_ARCHS)
+  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER 12.0 AND MACHETE_ARCHS)
    #
    # For the Machete kernels we automatically generate sources for various
    # preselected input type pairs and schedules.
@ -739,7 +624,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")

    message(STATUS "Building Machete kernels for archs: ${MACHETE_ARCHS}")
  else()
-    if (NOT ${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.0
+    if (NOT ${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER 12.0
        AND MACHETE_ARCHS)
      message(STATUS "Not building Machete kernels as CUDA Compiler version is "
                     "not >= 12.0, we recommend upgrading to CUDA 12.0 or "
@ -750,44 +635,9 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
                     "found in CUDA target architectures")
    endif()
  endif()
-
-  # Only build W4A8 kernels if we are building for something compatible with sm90a
-  cuda_archs_loose_intersection(W4A8_ARCHS "9.0a" "${CUDA_ARCHS}")
-  if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.0 AND W4A8_ARCHS)
-    set(SRCS
-       "csrc/quantization/cutlass_w4a8/w4a8_mm_entry.cu")
-
-    set_gencode_flags_for_srcs(
-      SRCS "${SRCS}"
-      CUDA_ARCHS "${W4A8_ARCHS}")
-
-    list(APPEND VLLM_EXT_SRC "${SRCS}")
-
-    message(STATUS "Building W4A8 kernels for archs: ${W4A8_ARCHS}")
-  else()
-    if (NOT ${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.0
-        AND W4A8_ARCHS)
-      message(STATUS "Not building W4A8 kernels as CUDA Compiler version is "
-                     "not >= 12.0, we recommend upgrading to CUDA 12.0 or "
-                     "later if you intend on running w4a16 quantized models on "
-                     "Hopper.")
-    else()
-      message(STATUS "Not building W4A8 kernels as no compatible archs "
-                     "found in CUDA target architectures")
-    endif()
-  endif()
-
 # if CUDA endif
 endif()

-if (VLLM_GPU_LANG STREQUAL "HIP")
-  # Add QuickReduce kernels
-  list(APPEND VLLM_EXT_SRC
-    "csrc/custom_quickreduce.cu"
-  )
-# if ROCM endif
-endif()
-
 message(STATUS "Enabling C extension.")
 define_gpu_extension_target(
  _C
@ -817,17 +667,7 @@ set(VLLM_MOE_EXT_SRC
  "csrc/moe/topk_softmax_kernels.cu")

 if(VLLM_GPU_LANG STREQUAL "CUDA")
-  list(APPEND VLLM_MOE_EXT_SRC
-    "csrc/moe/moe_wna16.cu"
-    "csrc/moe/grouped_topk_kernels.cu")
-endif()
-
-if(VLLM_GPU_LANG STREQUAL "CUDA")
-  set(MOE_PERMUTE_SRC
-      "csrc/moe/permute_unpermute_kernels/moe_permute_unpermute_kernel.cu"
-      "csrc/moe/moe_permute_unpermute_op.cu")
-
-  list(APPEND VLLM_MOE_EXT_SRC "${MOE_PERMUTE_SRC}")
+  list(APPEND VLLM_MOE_EXT_SRC "csrc/moe/moe_wna16.cu")
 endif()

 set_gencode_flags_for_srcs(
@ -844,7 +684,7 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")

  list(APPEND VLLM_MOE_EXT_SRC "${VLLM_MOE_WNA16_SRC}")
  # 9.0 for latest bf16 atomicAdd PTX
-  cuda_archs_loose_intersection(MARLIN_MOE_ARCHS "8.0;8.7;9.0+PTX" "${CUDA_ARCHS}")
+  cuda_archs_loose_intersection(MARLIN_MOE_ARCHS "8.0;9.0+PTX" "${CUDA_ARCHS}")
  if (MARLIN_MOE_ARCHS)

    #
@ -888,10 +728,6 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
    set_gencode_flags_for_srcs(
      SRCS "${MOE_WNAA16_MARLIN_SRC}"
      CUDA_ARCHS "${MARLIN_MOE_ARCHS}")
-    if(${CMAKE_CUDA_COMPILER_VERSION} VERSION_GREATER_EQUAL 12.8)
-      set_source_files_properties(${MOE_WNAA16_MARLIN_SRC}
-        PROPERTIES COMPILE_FLAGS "-static-global-template-stub=false")
-    endif()

    list(APPEND VLLM_MOE_EXT_SRC ${MOE_WNAA16_MARLIN_SRC})

@ -902,6 +738,17 @@ if(VLLM_GPU_LANG STREQUAL "CUDA")
  endif()
 endif()

+if(VLLM_GPU_LANG STREQUAL "CUDA")
+  set(MOE_PERMUTE_SRC
+      "csrc/moe/permute_unpermute_kernels/moe_permute_unpermute_kernel.cu"
+      "csrc/moe/moe_permute_unpermute_op.cu")
+
+  set_gencode_flags_for_srcs(
+    SRCS "${MARLIN_PERMUTE_SRC}"
+    CUDA_ARCHS "${MOE_PERMUTE_ARCHS}")
+
+  list(APPEND VLLM_MOE_EXT_SRC "${MOE_PERMUTE_SRC}")
+endif()
 message(STATUS "Enabling moe extension.")
 define_gpu_extension_target(
  _moe_C
--- a/README.md
+++ b/README.md
@ -1,4 +1,3 @@
-<!-- markdownlint-disable MD001 MD041 -->
 <p align="center">
  <picture>
    <source media="(prefers-color-scheme: dark)" srcset="https://raw.githubusercontent.com/vllm-project/vllm/main/docs/assets/logos/vllm-logo-text-dark.png">
@ -17,18 +16,14 @@ Easy, fast, and cheap LLM serving for everyone
 ---

 *Latest News* 🔥
-
- [2025/08] We hosted [vLLM Shanghai Meetup](https://mp.weixin.qq.com/s/pDmAXHcN7Iqc8sUKgJgGtg) focusing on building, developing, and integrating with vLLM! Please find the meetup slides [here](https://drive.google.com/drive/folders/1OvLx39wnCGy_WKq8SiVKf7YcxxYI3WCH).
- [2025/08] We hosted [vLLM Korea Meetup](https://luma.com/cgcgprmh) with Red Hat and Rebellions! We shared the latest advancements in vLLM along with project spotlights from the vLLM Korea community. Please find the meetup slides [here](https://drive.google.com/file/d/1bcrrAE1rxUgx0mjIeOWT6hNe2RefC5Hm/view).
- [2025/08] We hosted [vLLM Beijing Meetup](https://mp.weixin.qq.com/s/dgkWg1WFpWGO2jCdTqQHxA) focusing on large-scale LLM deployment! Please find the meetup slides [here](https://drive.google.com/drive/folders/1Pid6NSFLU43DZRi0EaTcPgXsAzDvbBqF) and the recording [here](https://www.chaspark.com/#/live/1166916873711665152).
+- [2025/05] We hosted [NYC vLLM Meetup](https://lu.ma/c1rqyf1f)! Please find the meetup slides [here](https://docs.google.com/presentation/d/1_q_aW_ioMJWUImf1s1YM-ZhjXz8cUeL0IJvaquOYBeA/edit?usp=sharing).
 - [2025/05] vLLM is now a hosted project under PyTorch Foundation! Please find the announcement [here](https://pytorch.org/blog/pytorch-foundation-welcomes-vllm/).
+- [2025/04] We hosted [Asia Developer Day](https://www.sginnovate.com/event/limited-availability-morning-evening-slots-remaining-inaugural-vllm-asia-developer-day)! Please find the meetup slides from the vLLM team [here](https://docs.google.com/presentation/d/19cp6Qu8u48ihB91A064XfaXruNYiBOUKrBxAmDOllOo/edit?usp=sharing).
 - [2025/01] We are excited to announce the alpha release of vLLM V1: A major architectural upgrade with 1.7x speedup! Clean code, optimized execution loop, zero-overhead prefix caching, enhanced multimodal support, and more. Please check out our blog post [here](https://blog.vllm.ai/2025/01/27/v1-alpha-release.html).

 <details>
 <summary>Previous News</summary>

- [2025/05] We hosted [NYC vLLM Meetup](https://lu.ma/c1rqyf1f)! Please find the meetup slides [here](https://docs.google.com/presentation/d/1_q_aW_ioMJWUImf1s1YM-ZhjXz8cUeL0IJvaquOYBeA/edit?usp=sharing).
- [2025/04] We hosted [Asia Developer Day](https://www.sginnovate.com/event/limited-availability-morning-evening-slots-remaining-inaugural-vllm-asia-developer-day)! Please find the meetup slides from the vLLM team [here](https://docs.google.com/presentation/d/19cp6Qu8u48ihB91A064XfaXruNYiBOUKrBxAmDOllOo/edit?usp=sharing).
 - [2025/03] We hosted [vLLM x Ollama Inference Night](https://lu.ma/vllm-ollama)! Please find the meetup slides from the vLLM team [here](https://docs.google.com/presentation/d/16T2PDD1YwRnZ4Tu8Q5r6n53c5Lr5c73UV9Vd2_eBo4U/edit?usp=sharing).
 - [2025/03] We hosted [the first vLLM China Meetup](https://mp.weixin.qq.com/s/n77GibL2corAtQHtVEAzfg)! Please find the meetup slides from vLLM team [here](https://docs.google.com/presentation/d/1REHvfQMKGnvz6p3Fd23HhSO4c8j5WPGZV0bKYLwnHyQ/edit?usp=sharing).
 - [2025/03] We hosted [the East Coast vLLM Meetup](https://lu.ma/7mu4k4xx)! Please find the meetup slides [here](https://docs.google.com/presentation/d/1NHiv8EUFF1NLd3fEYODm56nDmL26lEeXCaDgyDlTsRs/edit#slide=id.g31441846c39_0_0).
@ -51,7 +46,6 @@ Easy, fast, and cheap LLM serving for everyone
 </details>

 ---
-
 ## About

 vLLM is a fast and easy-to-use library for LLM inference and serving.
@ -69,11 +63,13 @@ vLLM is fast with:
 - Speculative decoding
 - Chunked prefill

+**Performance benchmark**: We include a performance benchmark at the end of [our blog post](https://blog.vllm.ai/2024/09/05/perf-update.html). It compares the performance of vLLM against other LLM serving engines ([TensorRT-LLM](https://github.com/NVIDIA/TensorRT-LLM), [SGLang](https://github.com/sgl-project/sglang) and [LMDeploy](https://github.com/InternLM/lmdeploy)). The implementation is under [nightly-benchmarks folder](.buildkite/nightly-benchmarks/) and you can [reproduce](https://github.com/vllm-project/vllm/issues/8176) this benchmark using our one-click runnable script.
+
 vLLM is flexible and easy to use with:

 - Seamless integration with popular Hugging Face models
 - High-throughput serving with various decoding algorithms, including *parallel sampling*, *beam search*, and more
- Tensor, pipeline, data and expert parallelism support for distributed inference
+- Tensor parallelism and pipeline parallelism support for distributed inference
 - Streaming outputs
 - OpenAI-compatible API server
 - Support NVIDIA GPUs, AMD CPUs and GPUs, Intel CPUs and GPUs, PowerPC CPUs, TPU, and AWS Neuron
@ -81,7 +77,6 @@ vLLM is flexible and easy to use with:
 - Multi-LoRA support

 vLLM seamlessly supports most popular open-source models on HuggingFace, including:
-
 - Transformer-like LLMs (e.g., Llama)
 - Mixture-of-Expert LLMs (e.g., Mixtral, Deepseek-V2 and V3)
 - Embedding Models (e.g., E5-Mistral)
@ -98,7 +93,6 @@ pip install vllm
 ```

 Visit our [documentation](https://docs.vllm.ai/en/latest/) to learn more.
-
 - [Installation](https://docs.vllm.ai/en/latest/getting_started/installation.html)
 - [Quickstart](https://docs.vllm.ai/en/latest/getting_started/quickstart.html)
 - [List of Supported Models](https://docs.vllm.ai/en/latest/models/supported_models.html)
@ -115,7 +109,6 @@ vLLM is a community project. Our compute resources for development and testing a
 <!-- Note: Please sort them in alphabetical order. -->
 <!-- Note: Please keep these consistent with docs/community/sponsors.md -->
 Cash Donations:
-
 - a16z
 - Dropbox
 - Sequoia Capital
@ -123,8 +116,6 @@ Cash Donations:
 - ZhenFund

 Compute Resources:
-
- Alibaba Cloud
 - AMD
 - Anyscale
 - AWS
@ -163,13 +154,11 @@ If you use vLLM for your research, please cite our [paper](https://arxiv.org/abs

 ## Contact Us

-<!-- --8<-- [start:contact-us] -->
- For technical questions and feature requests, please use GitHub [Issues](https://github.com/vllm-project/vllm/issues)
+- For technical questions and feature requests, please use GitHub [Issues](https://github.com/vllm-project/vllm/issues) or [Discussions](https://github.com/vllm-project/vllm/discussions)
 - For discussing with fellow users, please use the [vLLM Forum](https://discuss.vllm.ai)
- For coordinating contributions and development, please use [Slack](https://slack.vllm.ai)
+- coordinating contributions and development, please use [Slack](https://slack.vllm.ai)
 - For security disclosures, please use GitHub's [Security Advisories](https://github.com/vllm-project/vllm/security/advisories) feature
 - For collaborations and partnerships, please contact us at [vllm-questions@lists.berkeley.edu](mailto:vllm-questions@lists.berkeley.edu)
-<!-- --8<-- [end:contact-us] -->

 ## Media Kit

--- a/RELEASE.md
+++ b/RELEASE.md
@ -52,39 +52,3 @@ After branch cut, we approach finalizing the release branch with clear criteria
 * Release branch specific changes (e.g. change version identifiers or CI fixes)

 Please note: **No feature work allowed for cherry picks**. All PRs that are considered for cherry-picks need to be merged on trunk, the only exception are Release branch specific changes.
-
-## Manual validations
-
-### E2E Performance Validation
-
-Before each release, we perform end-to-end performance validation to ensure no regressions are introduced. This validation uses the [vllm-benchmark workflow](https://github.com/pytorch/pytorch-integration-testing/actions/workflows/vllm-benchmark.yml) on PyTorch CI.
-
-**Current Coverage:**
-
-* Models: Llama3, Llama4, and Mixtral
-* Hardware: NVIDIA H100 and AMD MI300x
-* _Note: Coverage may change based on new model releases and hardware availability_
-
-**Performance Validation Process:**
-
-**Step 1: Get Access**
-Request write access to the [pytorch/pytorch-integration-testing](https://github.com/pytorch/pytorch-integration-testing) repository to run the benchmark workflow.
-
-**Step 2: Review Benchmark Setup**
-Familiarize yourself with the benchmark configurations:
-
-* [CUDA setup](https://github.com/pytorch/pytorch-integration-testing/tree/main/vllm-benchmarks/benchmarks/cuda)
-* [ROCm setup](https://github.com/pytorch/pytorch-integration-testing/tree/main/vllm-benchmarks/benchmarks/rocm)
-
-**Step 3: Run the Benchmark**
-Navigate to the [vllm-benchmark workflow](https://github.com/pytorch/pytorch-integration-testing/actions/workflows/vllm-benchmark.yml) and configure:
-
-* **vLLM branch**: Set to the release branch (e.g., `releases/v0.9.2`)
-* **vLLM commit**: Set to the RC commit hash
-
-**Step 4: Review Results**
-Once the workflow completes, benchmark results will be available on the [vLLM benchmark dashboard](https://hud.pytorch.org/benchmark/llms?repoName=vllm-project%2Fvllm) under the corresponding branch and commit.
-
-**Step 5: Performance Comparison**
-Compare the current results against the previous release to verify no performance regressions have occurred. Here is an
-example of [v0.9.1 vs v0.9.2](https://hud.pytorch.org/benchmark/llms?startTime=Thu%2C%2017%20Apr%202025%2021%3A43%3A50%20GMT&stopTime=Wed%2C%2016%20Jul%202025%2021%3A43%3A50%20GMT&granularity=week&lBranch=releases/v0.9.1&lCommit=b6553be1bc75f046b00046a4ad7576364d03c835&rBranch=releases/v0.9.2&rCommit=a5dd03c1ebc5e4f56f3c9d3dc0436e9c582c978f&repoName=vllm-project%2Fvllm&benchmarkName=&modelName=All%20Models&backendName=All%20Backends&modeName=All%20Modes&dtypeName=All%20DType&deviceName=All%20Devices&archName=All%20Platforms).
--- a/SECURITY.md
+++ b/SECURITY.md
@ -1,50 +1,13 @@
 # Security Policy

-## Reporting security issues
+## Reporting a Vulnerability

-Please report security issues privately using [the vulnerability submission form](https://github.com/vllm-project/vllm/security/advisories/new).
+If you believe you have found a security vulnerability in vLLM, we encourage you to let us know right away. We will investigate all legitimate reports and do our best to quickly fix the problem.

-## Issue triage
+Please report security issues privately using [the vulnerability submission form](https://github.com/vllm-project/vllm/security/advisories/new). Reports will then be triaged by the [vulnerability management team](https://docs.vllm.ai/en/latest/contributing/vulnerability_management.html).

-Reports will then be triaged by the [vulnerability management team](https://docs.vllm.ai/en/latest/contributing/vulnerability_management.html).
-
-## Threat model
+---

 Please see the [Security Guide in the vLLM documentation](https://docs.vllm.ai/en/latest/usage/security.html) for more information on vLLM's security assumptions and recommendations.

 Please see [PyTorch's Security Policy](https://github.com/pytorch/pytorch/blob/main/SECURITY.md) for more information and recommendations on how to securely interact with models.
-
-## Issue severity
-
-We will determine the risk of each issue, taking into account our experience dealing with past issues, versions affected, common defaults, and use cases. We use the following severity categories:
-
-### CRITICAL Severity
-
-Vulnerabilities that allow remote attackers to execute arbitrary code, take full control of the system, or significantly compromise confidentiality, integrity, or availability without any interaction or privileges needed, examples include remote code execution via network, deserialization issues that allow exploit chains. Generally those issues which are rated as CVSS  ≥ 9.0.
-
-### HIGH Severity
-
-Serious security flaws that allow elevated impact—like RCE in specific, limited contexts or significant data loss—but require advanced conditions or some trust, examples include RCE in advanced deployment modes (e.g. multi-node), or high impact issues where some sort of privileged network access is required. These issues typically have CVSS scores between 7.0 and 8.9
-
-### MODERATE Severity
-
-Vulnerabilities that cause denial of service or partial disruption, but do not allow arbitrary code execution or data breach and have limited impact. These issues have a CVSS rating between 4.0 and 6.9
-
-### LOW Severity
-
-Minor issues such as informational disclosures, logging errors, non-exploitable flaws, or weaknesses that require local or high-privilege access and offer negligible impact. Examples include side channel attacks or hash collisions. These issues often have CVSS scores less than 4.0
-
-## Prenotification policy
-
-For certain security issues of CRITICAL, HIGH, or MODERATE severity level, we may prenotify certain organizations or vendors that ship vLLM. The purpose of this prenotification is to allow for a coordinated release of fixes for severe issues.
-
-* This prenotification will be in the form of a private email notification. It may also include adding security contacts to the GitHub security advisory, typically a few days before release.
-
-* If you wish to be added to the prenotification group, please send an email copying all the members of the [vulnerability management team](https://docs.vllm.ai/en/latest/contributing/vulnerability_management.html). Each vendor contact will be analyzed on a case-by-case basis.
-
-* Organizations and vendors who either ship or use vLLM, are eligible to join the prenotification group if they meet at least one of the following qualifications
-    * Substantial internal deployment leveraging the upstream vLLM project.
-    * Established internal security teams and comprehensive compliance measures.
-    * Active and consistent contributions to the upstream vLLM project.
-
-* We may withdraw organizations from receiving future prenotifications if they release fixes or any other information about issues before they are public. Group membership may also change based on policy refinements for who may be included.
--- a/benchmarks/README.md
+++ b/benchmarks/README.md
@ -22,25 +22,6 @@ become available.
      <td style="text-align: center;">✅</td>
      <td><code>wget https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/resolve/main/ShareGPT_V3_unfiltered_cleaned_split.json</code></td>
    </tr>
-    <tr>
-      <td><strong>ShareGPT4V (Image)</strong></td>
-      <td style="text-align: center;">✅</td>
-      <td style="text-align: center;">✅</td>
-      <td>
-        <code>wget https://huggingface.co/datasets/Lin-Chen/ShareGPT4V/blob/main/sharegpt4v_instruct_gpt4-vision_cap100k.json</code>
-        <br>
-        <div>Note that the images need to be downloaded separately. For example, to download COCO's 2017 Train images:</div>
-        <code>wget http://images.cocodataset.org/zips/train2017.zip</code>
-      </td>
-    </tr>
-        <tr>
-      <td><strong>ShareGPT4Video (Video)</strong></td>
-      <td style="text-align: center;">✅</td>
-      <td style="text-align: center;">✅</td>
-      <td>
-        <code>git clone https://huggingface.co/datasets/ShareGPT4Video/ShareGPT4Video</code>
-      </td>
-    </tr>
    <tr>
      <td><strong>BurstGPT</strong></td>
      <td style="text-align: center;">✅</td>
@ -48,7 +29,7 @@ become available.
      <td><code>wget https://github.com/HPMLL/BurstGPT/releases/download/v1.1/BurstGPT_without_fails_2.csv</code></td>
    </tr>
    <tr>
-      <td><strong>Sonnet (deprecated)</strong></td>
+      <td><strong>Sonnet</strong></td>
      <td style="text-align: center;">✅</td>
      <td style="text-align: center;">✅</td>
      <td>Local file: <code>benchmarks/sonnet.txt</code></td>
@ -59,18 +40,6 @@ become available.
      <td style="text-align: center;">✅</td>
      <td><code>synthetic</code></td>
    </tr>
-    <tr>
-      <td><strong>RandomMultiModal (Image/Video)</strong></td>
-      <td style="text-align: center;">🟡</td>
-      <td style="text-align: center;">🚧</td>
-      <td><code>synthetic</code> </td>
-    </tr>
-    <tr>
-      <td><strong>Prefix Repetition</strong></td>
-      <td style="text-align: center;">✅</td>
-      <td style="text-align: center;">✅</td>
-      <td><code>synthetic</code></td>
-    </tr>
    <tr>
      <td><strong>HuggingFace-VisionArena</strong></td>
      <td style="text-align: center;">✅</td>
@ -112,17 +81,13 @@ become available.

 **Note**: HuggingFace dataset's `dataset-name` should be set to `hf`

-## 🚀 Example - Online Benchmark
-
-<details>
-<summary>Show more</summary>
-
-<br/>
+---
+## Example - Online Benchmark

 First start serving your model

 ```bash
-vllm serve NousResearch/Hermes-3-Llama-3.1-8B
+vllm serve NousResearch/Hermes-3-Llama-3.1-8B --disable-log-requests
 ```

 Then run the benchmarking script
@ -130,7 +95,7 @@ Then run the benchmarking script
 ```bash
 # download dataset
 # wget https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/resolve/main/ShareGPT_V3_unfiltered_cleaned_split.json
-vllm bench serve \
+python3 vllm/benchmarks/benchmark_serving.py \
  --backend vllm \
  --model NousResearch/Hermes-3-Llama-3.1-8B \
  --endpoint /v1/completions \
@ -141,48 +106,47 @@ vllm bench serve \

 If successful, you will see the following output

-```text
+```
 ============ Serving Benchmark Result ============
-Successful requests:                     10
-Benchmark duration (s):                  5.78
-Total input tokens:                      1369
-Total generated tokens:                  2212
-Request throughput (req/s):              1.73
-Output token throughput (tok/s):         382.89
-Total Token throughput (tok/s):          619.85
+Successful requests:                     10        
+Benchmark duration (s):                  5.78      
+Total input tokens:                      1369      
+Total generated tokens:                  2212      
+Request throughput (req/s):              1.73      
+Output token throughput (tok/s):         382.89    
+Total Token throughput (tok/s):          619.85    
 ---------------Time to First Token----------------
-Mean TTFT (ms):                          71.54
-Median TTFT (ms):                        73.88
-P99 TTFT (ms):                           79.49
+Mean TTFT (ms):                          71.54     
+Median TTFT (ms):                        73.88     
+P99 TTFT (ms):                           79.49     
 -----Time per Output Token (excl. 1st token)------
-Mean TPOT (ms):                          7.91
-Median TPOT (ms):                        7.96
-P99 TPOT (ms):                           8.03
+Mean TPOT (ms):                          7.91      
+Median TPOT (ms):                        7.96      
+P99 TPOT (ms):                           8.03      
 ---------------Inter-token Latency----------------
-Mean ITL (ms):                           7.74
-Median ITL (ms):                         7.70
-P99 ITL (ms):                            8.39
+Mean ITL (ms):                           7.74      
+Median ITL (ms):                         7.70      
+P99 ITL (ms):                            8.39      
 ==================================================
 ```

 ### Custom Dataset
-
 If the dataset you want to benchmark is not supported yet in vLLM, even then you can benchmark on it using `CustomDataset`. Your data needs to be in `.jsonl` format and needs to have "prompt" field per entry, e.g., data.jsonl

-```json
+```
 {"prompt": "What is the capital of India?"}
 {"prompt": "What is the capital of Iran?"}
 {"prompt": "What is the capital of China?"}
-```
+``` 

 ```bash
 # start server
-VLLM_USE_V1=1 vllm serve meta-llama/Llama-3.1-8B-Instruct
+VLLM_USE_V1=1 vllm serve meta-llama/Llama-3.1-8B-Instruct --disable-log-requests
 ```

 ```bash
 # run benchmarking script
-vllm bench serve --port 9001 --save-result --save-detailed \
+python3 benchmarks/benchmark_serving.py --port 9001 --save-result --save-detailed \
  --backend vllm \
  --model meta-llama/Llama-3.1-8B-Instruct \
  --endpoint /v1/completions \
@ -202,13 +166,12 @@ You can skip applying chat template if your data already has it by using `--cust

 ```bash
 # need a model with vision capability here
-vllm serve Qwen/Qwen2-VL-7B-Instruct
+vllm serve Qwen/Qwen2-VL-7B-Instruct --disable-log-requests
 ```

 ```bash
-vllm bench serve \
+python3 vllm/benchmarks/benchmark_serving.py \
  --backend openai-chat \
-  --endpoint-type openai-chat \
  --model Qwen/Qwen2-VL-7B-Instruct \
  --endpoint /v1/chat/completions \
  --dataset-name hf \
@ -227,7 +190,7 @@ VLLM_USE_V1=1 vllm serve meta-llama/Meta-Llama-3-8B-Instruct \
 ```

 ``` bash
-vllm bench serve \
+python3 benchmarks/benchmark_serving.py \
    --model meta-llama/Meta-Llama-3-8B-Instruct \
    --dataset-name hf \
    --dataset-path likaixin/InstructCoder \
@ -237,15 +200,14 @@ vllm bench serve \
 ### Other HuggingFaceDataset Examples

 ```bash
-vllm serve Qwen/Qwen2-VL-7B-Instruct
+vllm serve Qwen/Qwen2-VL-7B-Instruct --disable-log-requests
 ```

-`lmms-lab/LLaVA-OneVision-Data`:
+**`lmms-lab/LLaVA-OneVision-Data`**

 ```bash
-vllm bench serve \
+python3 vllm/benchmarks/benchmark_serving.py \
  --backend openai-chat \
-  --endpoint-type openai-chat \
  --model Qwen/Qwen2-VL-7B-Instruct \
  --endpoint /v1/chat/completions \
  --dataset-name hf \
@ -255,12 +217,11 @@ vllm bench serve \
  --num-prompts 10
 ```

-`Aeala/ShareGPT_Vicuna_unfiltered`:
+**`Aeala/ShareGPT_Vicuna_unfiltered`**

 ```bash
-vllm bench serve \
+python3 vllm/benchmarks/benchmark_serving.py \
  --backend openai-chat \
-  --endpoint-type openai-chat \
  --model Qwen/Qwen2-VL-7B-Instruct \
  --endpoint /v1/chat/completions \
  --dataset-name hf \
@ -269,10 +230,10 @@ vllm bench serve \
  --num-prompts 10
 ```

-`AI-MO/aimo-validation-aime`:
+**`AI-MO/aimo-validation-aime`**

 ``` bash
-vllm bench serve \
+python3 vllm/benchmarks/benchmark_serving.py \
    --model Qwen/QwQ-32B \
    --dataset-name hf \
    --dataset-path AI-MO/aimo-validation-aime \
@ -280,10 +241,10 @@ vllm bench serve \
    --seed 42
 ```

-`philschmid/mt-bench`:
+**`philschmid/mt-bench`**

 ``` bash
-vllm bench serve \
+python3 vllm/benchmarks/benchmark_serving.py \
    --model Qwen/QwQ-32B \
    --dataset-name hf \
    --dataset-path philschmid/mt-bench \
@ -296,7 +257,7 @@ When using OpenAI-compatible backends such as `vllm`, optional sampling
 parameters can be specified. Example client command:

 ```bash
-vllm bench serve \
+python3 vllm/benchmarks/benchmark_serving.py \
  --backend vllm \
  --model NousResearch/Hermes-3-Llama-3.1-8B \
  --endpoint /v1/completions \
@ -308,34 +269,11 @@ vllm bench serve \
  --num-prompts 10
 ```

-### Running With Ramp-Up Request Rate
-
-The benchmark tool also supports ramping up the request rate over the
-duration of the benchmark run. This can be useful for stress testing the
-server or finding the maximum throughput that it can handle, given some latency budget.
-
-Two ramp-up strategies are supported:
-
- `linear`: Increases the request rate linearly from a start value to an end value.
- `exponential`: Increases the request rate exponentially.
-
-The following arguments can be used to control the ramp-up:
-
- `--ramp-up-strategy`: The ramp-up strategy to use (`linear` or `exponential`).
- `--ramp-up-start-rps`: The request rate at the beginning of the benchmark.
- `--ramp-up-end-rps`: The request rate at the end of the benchmark.
-
-</details>
-
-## 📈 Example - Offline Throughput Benchmark
-
-<details>
-<summary>Show more</summary>
-
-<br/>
+---
+## Example - Offline Throughput Benchmark

 ```bash
-vllm bench throughput \
+python3 vllm/benchmarks/benchmark_throughput.py \
  --model NousResearch/Hermes-3-Llama-3.1-8B \
  --dataset-name sonnet \
  --dataset-path vllm/benchmarks/sonnet.txt \
@ -344,7 +282,7 @@ vllm bench throughput \

 If successful, you will see the following output

-```text
+```
 Throughput: 7.15 requests/s, 4656.00 total tokens/s, 1072.15 output tokens/s
 Total num prompt tokens:  5014
 Total num output tokens:  1500
@ -352,8 +290,8 @@ Total num output tokens:  1500

 ### VisionArena Benchmark for Vision Language Models

-```bash
-vllm bench throughput \
+``` bash
+python3 vllm/benchmarks/benchmark_throughput.py \
  --model Qwen/Qwen2-VL-7B-Instruct \
  --backend vllm-chat \
  --dataset-name hf \
@ -364,7 +302,7 @@ vllm bench throughput \

 The `num prompt tokens` now includes image token counts

-```text
+```
 Throughput: 2.55 requests/s, 4036.92 total tokens/s, 326.90 output tokens/s
 Total num prompt tokens:  14527
 Total num output tokens:  1280
@ -375,7 +313,7 @@ Total num output tokens:  1280
 ``` bash
 VLLM_WORKER_MULTIPROC_METHOD=spawn \
 VLLM_USE_V1=1 \
-vllm bench throughput \
+python3 vllm/benchmarks/benchmark_throughput.py \
    --dataset-name=hf \
    --dataset-path=likaixin/InstructCoder \
    --model=meta-llama/Meta-Llama-3-8B-Instruct \
@ -388,7 +326,7 @@ vllm bench throughput \
    "prompt_lookup_min": 2}'
 ```

-```text
+```
 Throughput: 104.77 requests/s, 23836.22 total tokens/s, 10477.10 output tokens/s
 Total num prompt tokens:  261136
 Total num output tokens:  204800
@ -396,10 +334,10 @@ Total num output tokens:  204800

 ### Other HuggingFaceDataset Examples

-`lmms-lab/LLaVA-OneVision-Data`:
+**`lmms-lab/LLaVA-OneVision-Data`**

 ```bash
-vllm bench throughput \
+python3 vllm/benchmarks/benchmark_throughput.py \
  --model Qwen/Qwen2-VL-7B-Instruct \
  --backend vllm-chat \
  --dataset-name hf \
@ -409,10 +347,10 @@ vllm bench throughput \
  --num-prompts 10
 ```

-`Aeala/ShareGPT_Vicuna_unfiltered`:
+**`Aeala/ShareGPT_Vicuna_unfiltered`**

 ```bash
-vllm bench throughput \
+python3 vllm/benchmarks/benchmark_throughput.py \
  --model Qwen/Qwen2-VL-7B-Instruct \
  --backend vllm-chat \
  --dataset-name hf \
@ -421,10 +359,10 @@ vllm bench throughput \
  --num-prompts 10
 ```

-`AI-MO/aimo-validation-aime`:
+**`AI-MO/aimo-validation-aime`**

 ```bash
-vllm bench throughput \
+python3 benchmarks/benchmark_throughput.py \
  --model Qwen/QwQ-32B \
  --backend vllm \
  --dataset-name hf \
@ -433,12 +371,12 @@ vllm bench throughput \
  --num-prompts 10
 ```

-Benchmark with LoRA adapters:
+### Benchmark with LoRA Adapters

 ``` bash
 # download dataset
 # wget https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/resolve/main/ShareGPT_V3_unfiltered_cleaned_split.json
-vllm bench throughput \
+python3 vllm/benchmarks/benchmark_throughput.py \
  --model meta-llama/Llama-2-7b-hf \
  --backend vllm \
  --dataset_path <your data path>/ShareGPT_V3_unfiltered_cleaned_split.json \
@ -449,354 +387,3 @@ vllm bench throughput \
  --enable-lora \
  --lora-path yard1/llama-2-7b-sql-lora-test
  ```
-
-</details>
-
-## 🛠️ Example - Structured Output Benchmark
-
-<details>
-<summary>Show more</summary>
-
-<br/>
-
-Benchmark the performance of structured output generation (JSON, grammar, regex).
-
-### Server Setup
-
-```bash
-vllm serve NousResearch/Hermes-3-Llama-3.1-8B
-```
-
-### JSON Schema Benchmark
-
-```bash
-python3 benchmarks/benchmark_serving_structured_output.py \
-  --backend vllm \
-  --model NousResearch/Hermes-3-Llama-3.1-8B \
-  --dataset json \
-  --structured-output-ratio 1.0 \
-  --request-rate 10 \
-  --num-prompts 1000
-```
-
-### Grammar-based Generation Benchmark
-
-```bash
-python3 benchmarks/benchmark_serving_structured_output.py \
-  --backend vllm \
-  --model NousResearch/Hermes-3-Llama-3.1-8B \
-  --dataset grammar \
-  --structure-type grammar \
-  --request-rate 10 \
-  --num-prompts 1000
-```
-
-### Regex-based Generation Benchmark
-
-```bash
-python3 benchmarks/benchmark_serving_structured_output.py \
-  --backend vllm \
-  --model NousResearch/Hermes-3-Llama-3.1-8B \
-  --dataset regex \
-  --request-rate 10 \
-  --num-prompts 1000
-```
-
-### Choice-based Generation Benchmark
-
-```bash
-python3 benchmarks/benchmark_serving_structured_output.py \
-  --backend vllm \
-  --model NousResearch/Hermes-3-Llama-3.1-8B \
-  --dataset choice \
-  --request-rate 10 \
-  --num-prompts 1000
-```
-
-### XGrammar Benchmark Dataset
-
-```bash
-python3 benchmarks/benchmark_serving_structured_output.py \
-  --backend vllm \
-  --model NousResearch/Hermes-3-Llama-3.1-8B \
-  --dataset xgrammar_bench \
-  --request-rate 10 \
-  --num-prompts 1000
-```
-
-</details>
-
-## 📚 Example - Long Document QA Benchmark
-
-<details>
-<summary>Show more</summary>
-
-<br/>
-
-Benchmark the performance of long document question-answering with prefix caching.
-
-### Basic Long Document QA Test
-
-```bash
-python3 benchmarks/benchmark_long_document_qa_throughput.py \
-  --model meta-llama/Llama-2-7b-chat-hf \
-  --enable-prefix-caching \
-  --num-documents 16 \
-  --document-length 2000 \
-  --output-len 50 \
-  --repeat-count 5
-```
-
-### Different Repeat Modes
-
-```bash
-# Random mode (default) - shuffle prompts randomly
-python3 benchmarks/benchmark_long_document_qa_throughput.py \
-  --model meta-llama/Llama-2-7b-chat-hf \
-  --enable-prefix-caching \
-  --num-documents 8 \
-  --document-length 3000 \
-  --repeat-count 3 \
-  --repeat-mode random
-
-# Tile mode - repeat entire prompt list in sequence
-python3 benchmarks/benchmark_long_document_qa_throughput.py \
-  --model meta-llama/Llama-2-7b-chat-hf \
-  --enable-prefix-caching \
-  --num-documents 8 \
-  --document-length 3000 \
-  --repeat-count 3 \
-  --repeat-mode tile
-
-# Interleave mode - repeat each prompt consecutively
-python3 benchmarks/benchmark_long_document_qa_throughput.py \
-  --model meta-llama/Llama-2-7b-chat-hf \
-  --enable-prefix-caching \
-  --num-documents 8 \
-  --document-length 3000 \
-  --repeat-count 3 \
-  --repeat-mode interleave
-```
-
-</details>
-
-## 🗂️ Example - Prefix Caching Benchmark
-
-<details>
-<summary>Show more</summary>
-
-<br/>
-
-Benchmark the efficiency of automatic prefix caching.
-
-### Fixed Prompt with Prefix Caching
-
-```bash
-python3 benchmarks/benchmark_prefix_caching.py \
-  --model meta-llama/Llama-2-7b-chat-hf \
-  --enable-prefix-caching \
-  --num-prompts 1 \
-  --repeat-count 100 \
-  --input-length-range 128:256
-```
-
-### ShareGPT Dataset with Prefix Caching
-
-```bash
-# download dataset
-# wget https://huggingface.co/datasets/anon8231489123/ShareGPT_Vicuna_unfiltered/resolve/main/ShareGPT_V3_unfiltered_cleaned_split.json
-
-python3 benchmarks/benchmark_prefix_caching.py \
-  --model meta-llama/Llama-2-7b-chat-hf \
-  --dataset-path /path/ShareGPT_V3_unfiltered_cleaned_split.json \
-  --enable-prefix-caching \
-  --num-prompts 20 \
-  --repeat-count 5 \
-  --input-length-range 128:256
-```
-
-### Prefix Repetition Dataset
-
-```bash
-vllm bench serve \
-  --backend openai \
-  --model meta-llama/Llama-2-7b-chat-hf \
-  --dataset-name prefix_repetition \
-  --num-prompts 100 \
-  --prefix-repetition-prefix-len 512 \
-  --prefix-repetition-suffix-len 128 \
-  --prefix-repetition-num-prefixes 5 \
-  --prefix-repetition-output-len 128
-```
-
-</details>
-
-## ⚡ Example - Request Prioritization Benchmark
-
-<details>
-<summary>Show more</summary>
-
-<br/>
-
-Benchmark the performance of request prioritization in vLLM.
-
-### Basic Prioritization Test
-
-```bash
-python3 benchmarks/benchmark_prioritization.py \
-  --model meta-llama/Llama-2-7b-chat-hf \
-  --input-len 128 \
-  --output-len 64 \
-  --num-prompts 100 \
-  --scheduling-policy priority
-```
-
-### Multiple Sequences per Prompt
-
-```bash
-python3 benchmarks/benchmark_prioritization.py \
-  --model meta-llama/Llama-2-7b-chat-hf \
-  --input-len 128 \
-  --output-len 64 \
-  --num-prompts 100 \
-  --scheduling-policy priority \
-  --n 2
-```
-
-</details>
-
-## 👁️ Example - Multi-Modal Benchmark
-
-<details>
-<summary>Show more</summary>
-
-<br/>
-
-Benchmark the performance of multi-modal requests in vLLM.
-
-### Images (ShareGPT4V)
-
-Start vLLM:
-
-```bash
-python -m vllm.entrypoints.openai.api_server \
-  --model Qwen/Qwen2.5-VL-7B-Instruct \
-  --dtype bfloat16 \
-  --limit-mm-per-prompt '{"image": 1}' \
-  --allowed-local-media-path /path/to/sharegpt4v/images
-```
-
-Send requests with images:
-
-```bash
-python benchmarks/benchmark_serving.py \
-  --backend openai-chat \
-  --model Qwen/Qwen2.5-VL-7B-Instruct \
-  --dataset-name sharegpt \
-  --dataset-path /path/to/ShareGPT4V/sharegpt4v_instruct_gpt4-vision_cap100k.json \
-  --num-prompts 100 \
-  --save-result \
-  --result-dir ~/vllm_benchmark_results \
-  --save-detailed \
-  --endpoint /v1/chat/completion
-```
-
-### Videos (ShareGPT4Video)
-
-Start vLLM:
-
-```bash
-python -m vllm.entrypoints.openai.api_server \
-  --model Qwen/Qwen2.5-VL-7B-Instruct \
-  --dtype bfloat16 \
-  --limit-mm-per-prompt '{"video": 1}' \
-  --allowed-local-media-path /path/to/sharegpt4video/videos
-```
-
-Send requests with videos:
-
-```bash
-python benchmarks/benchmark_serving.py \
-  --backend openai-chat \
-  --model Qwen/Qwen2.5-VL-7B-Instruct \
-  --dataset-name sharegpt \
-  --dataset-path /path/to/ShareGPT4Video/llava_v1_5_mix665k_with_video_chatgpt72k_share4video28k.json \
-  --num-prompts 100 \
-  --save-result \
-  --result-dir ~/vllm_benchmark_results \
-  --save-detailed \
-  --endpoint /v1/chat/completion
-```
-
-### Synthetic Random Images (random-mm)
-
-Generate synthetic image inputs alongside random text prompts to stress-test vision models without external datasets.
-
-Notes:
-
- Works only with online benchmark via the OpenAI  backend (`--backend openai-chat`) and endpoint `/v1/chat/completions`.
- Video sampling is not yet implemented.
-
-Start the server (example):
-
-```bash
-vllm serve Qwen/Qwen2.5-VL-3B-Instruct \
-  --dtype bfloat16 \
-  --max-model-len 16384 \
-  --limit-mm-per-prompt '{"image": 3, "video": 0}' \
-  --mm-processor-kwargs max_pixels=1003520
-```
-
-Benchmark. It is recommended to use the flag `--ignore-eos` to simulate real responses. You can set the size of the output via the arg `random-output-len`.
-
-Ex.1: Fixed number of items and a single image resolution, enforcing generation of approx 40 tokens:
-
-```bash
-vllm bench serve \
-  --backend openai-chat \
-  --model Qwen/Qwen2.5-VL-3B-Instruct \
-  --endpoint /v1/chat/completions \
-  --dataset-name random-mm \
-  --num-prompts 100 \
-  --max-concurrency 10 \
-  --random-prefix-len 25 \
-  --random-input-len 300 \
-  --random-output-len 40 \
-  --random-range-ratio 0.2 \
-  --random-mm-base-items-per-request 2 \
-  --random-mm-limit-mm-per-prompt '{"image": 3, "video": 0}' \
-  --random-mm-bucket-config '{(224, 224, 1): 1.0}' \
-  --request-rate inf \
-  --ignore-eos \
-  --seed 42
-```
-
-The number of items per request can be controlled by passing multiple image buckets:
-
-```bash
-  --random-mm-base-items-per-request 2 \
-  --random-mm-num-mm-items-range-ratio 0.5 \
-  --random-mm-limit-mm-per-prompt '{"image": 4, "video": 0}' \
-  --random-mm-bucket-config '{(256, 256, 1): 0.7, (720, 1280, 1): 0.3}' \
-```
-
-Flags specific to `random-mm`:
-
- `--random-mm-base-items-per-request`: base number of multimodal items per request.
- `--random-mm-num-mm-items-range-ratio`: vary item count uniformly in the closed integer range [floor(n·(1−r)), ceil(n·(1+r))]. Set r=0 to keep it fixed; r=1 allows 0 items.
- `--random-mm-limit-mm-per-prompt`: per-modality hard caps, e.g. '{"image": 3, "video": 0}'.
- `--random-mm-bucket-config`: dict mapping (H, W, T) → probability. Entries with probability 0 are removed; remaining probabilities are renormalized to sum to 1. Use T=1 for images. Set any T>1 for videos (video sampling not yet supported).
-
-Behavioral notes:
-
- If the requested base item count cannot be satisfied under the provided per-prompt limits, the tool raises an error rather than silently clamping.
-
-How sampling works:
-
- Determine per-request item count k by sampling uniformly from the integer range defined by `--random-mm-base-items-per-request` and `--random-mm-num-mm-items-range-ratio`, then clamp k to at most the sum of per-modality limits.
- For each of the k items, sample a bucket (H, W, T) according to the normalized probabilities in `--random-mm-bucket-config`, while tracking how many items of each modality have been added.
- If a modality (e.g., image) reaches its limit from `--random-mm-limit-mm-per-prompt`, all buckets of that modality are excluded and the remaining bucket probabilities are renormalized before continuing.
-This should be seen as an edge case, and if this behavior can be avoided by setting `--random-mm-limit-mm-per-prompt` to a large number. Note that this might result in errors due to engine config `--limit-mm-per-prompt`.
- The resulting request contains synthetic image data in `multi_modal_data` (OpenAI Chat format). When `random-mm` is used with the OpenAI Chat backend, prompts remain text and MM content is attached via `multi_modal_data`.
-
-</details>
--- a/benchmarks/auto_tune.sh
+++ b/benchmarks/auto_tune.sh
@ -0,0 +1,212 @@
+#!/bin/bash
+
+# This script aims to tune the best server parameter combinations to maximize throughput for given requirement. 
+# The current server parameter combination is  max_num_seqs and max_num_batched_tokens
+# It also supports additional requirement: e2e latency and prefix cache. 
+
+# Pre-requisite:
+# 1. Checkout to your branch, install/ update the correct running env. For TPU, activate conda env and install the corresponding torch, xla version. 
+# 2. If the model is customized, replace the MODEL's config with the customized config.
+# 3. Set variables (ALL REQUIRED)
+#   BASE: your directory for vllm repo
+#   MODEL: the model served by vllm
+#   DOWNLOAD_DIR: directory to download and load model weights.
+#   INPUT_LEN: request input len
+#   OUTPUT_LEN: request output len
+#   MIN_CACHE_HIT_PCT: prefix cache rate
+#   MAX_LATENCY_ALLOWED_MS: (e2e) latency requirement. If there's no latency requirement, set it to a large number like 1000000000
+# 4. Run the script, it might take a long time, you can use tmux to avoid the script stop if disconnection happens.
+# 5. The final result will be saved in RESULT file. 
+
+
+# Example use cases 
+# 1. Given input_len=1800, output_len=20, what's the best max_num_seqs and max_num_batched_tokens to get highest throughput?
+# Use INPUT_LEN=1800,  OUTPUT_LEN=20, MIN_CACHE_HIT_PCT=0, MAX_LATENCY_ALLOWED_MS=100000000000
+# 2. If we have latency requirement to be lower than 500ms, what's the best server parameter?
+# Use INPUT_LEN=1800,  OUTPUT_LEN=20, MIN_CACHE_HIT_PCT=0, MAX_LATENCY_ALLOWED_MS=500
+# 3. If we want to reach 60% prefix cache, what's the best server parameter? 
+# Use INPUT_LEN=1800,  OUTPUT_LEN=20, MIN_CACHE_HIT_PCT=60, MAX_LATENCY_ALLOWED_MS=500
+
+TAG=$(date +"%Y_%m_%d_%H_%M")
+BASE=""
+MODEL="meta-llama/Llama-3.1-8B-Instruct"
+DOWNLOAD_DIR=""
+INPUT_LEN=4000
+OUTPUT_LEN=16
+MIN_CACHE_HIT_PCT_PCT=0
+MAX_LATENCY_ALLOWED_MS=100000000000
+
+LOG_FOLDER="$BASE/auto-benchmark/$TAG"
+RESULT="$LOG_FOLDER/result.txt"
+
+echo "result file$ $RESULT"
+echo "model: $MODEL"
+echo
+
+rm -rf $LOG_FOLDER
+mkdir -p $LOG_FOLDER
+
+cd "$BASE/vllm"
+# create sonnet-4x.txt so that we can sample 2048 tokens for input
+echo "" > benchmarks/sonnet_4x.txt
+for _ in {1..4}
+do
+cat benchmarks/sonnet.txt >> benchmarks/sonnet_4x.txt
+done
+
+pip install datasets
+
+current_hash=$(git rev-parse HEAD)
+echo "hash:$current_hash" >> "$RESULT"
+echo "current_hash: $current_hash"
+
+best_throughput=0
+best_max_num_seqs=0
+best_num_batched_tokens=0
+best_goodput=0
+run_benchmark() {
+    local max_num_seqs=$1
+    local max_num_batched_tokens=$2
+    echo "max_num_seq: $max_num_seqs, max_num_batched_tokens: $max_num_batched_tokens"
+    local vllm_log="$LOG_FOLDER/vllm_log_${max_num_seqs}_${max_num_batched_tokens}.txt"
+    echo "vllm_log: $vllm_log"
+    echo
+    rm -f $vllm_log
+
+    # start the server
+    VLLM_USE_V1=1 VLLM_SERVER_DEV_MODE=1 vllm serve $MODEL \
+        --disable-log-requests \
+        --port 8004 \
+        --gpu-memory-utilization 0.98 \
+        --max-num-seqs $max_num_seqs \
+        --max-num-batched-tokens $max_num_batched_tokens \
+        --tensor-parallel-size 1 \
+        --enable-prefix-caching \
+        --load-format dummy \
+        --download-dir $DOWNLOAD_DIR \
+        --max-model-len $(( INPUT_LEN+OUTPUT_LEN )) > "$vllm_log" 2>&1 &
+    echo "wait for 10 minutes.."
+    echo
+    # wait for 10 minutes...
+    server_started=0
+    for i in {1..60}; do        
+        if grep -Fq "Application startup complete" "$vllm_log"; then
+            echo "Application started"
+            server_started=1
+            break
+        else
+            # echo "wait for 10 seconds..."
+            sleep 10
+        fi
+    done
+ 
+    if (( ! server_started )); then
+        echo "server did not start within 10 minutes, terminate the benchmarking. Please check server log at $vllm_log"
+        echo "pkill -f vllm"
+        echo
+        pkill vllm
+        sleep 10
+        return 1
+    fi
+    
+    echo "run benchmark test..."
+    echo
+    meet_latency_requirement=0
+    # get a basic qps by using request-rate inf
+    bm_log="$LOG_FOLDER/bm_log_${max_num_seqs}_${max_num_batched_tokens}_requestrate_inf.txt"
+    prefix_len=$(( INPUT_LEN * MIN_CACHE_HIT_PCT / 100 ))
+    python benchmarks/benchmark_serving.py \
+        --backend vllm \
+        --model $MODEL  \
+        --dataset-name sonnet \
+        --dataset-path benchmarks/sonnet_4x.txt \
+        --sonnet-input-len $INPUT_LEN \
+        --sonnet-output-len $OUTPUT_LEN \
+        --ignore-eos \
+        --disable-tqdm \
+        --request-rate inf \
+        --percentile-metrics ttft,tpot,itl,e2el \
+        --goodput e2el:$MAX_LATENCY_ALLOWED_MS \
+        --num-prompts 100 \
+        --sonnet-prefix-len $prefix_len \
+        --port 8004 > "$bm_log"
+    through_put=$(grep "Request throughput (req/s):" "$bm_log" | sed 's/[^0-9.]//g')
+    e2el=$(grep "P99 E2EL (ms):" "$bm_log" | awk '{print $NF}')
+    goodput=$(grep "Request goodput (req/s):" "$bm_log" | sed 's/[^0-9.]//g')
+
+    if (( $(echo "$e2el <= $MAX_LATENCY_ALLOWED_MS" | bc -l) )); then
+        meet_latency_requirement=1
+    fi
+
+    if (( ! meet_latency_requirement )); then
+    # start from request-rate as int(through_put) + 1
+        request_rate=$((${through_put%.*} + 1))
+        while ((request_rate > 0)); do
+            # clear prefix cache
+            curl -X POST http://0.0.0.0:8004/reset_prefix_cache
+            sleep 5
+            bm_log="$LOG_FOLDER/bm_log_${max_num_seqs}_${max_num_batched_tokens}_requestrate_${request_rate}.txt"
+            python benchmarks/benchmark_serving.py \
+                --backend vllm \
+                --model $MODEL  \
+                --dataset-name sonnet \
+                --dataset-path benchmarks/sonnet_4x.txt \
+                --sonnet-input-len $INPUT_LEN \
+                --sonnet-output-len $OUTPUT_LEN \
+                --ignore_eos \
+                --disable-tqdm \
+                --request-rate $request_rate \
+                --percentile-metrics ttft,tpot,itl,e2el \
+                --goodput e2el:$MAX_LATENCY_ALLOWED_MS \
+                --num-prompts 100 \
+                --sonnet-prefix-len $prefix_len \
+                --port 8004 > "$bm_log"
+            through_put=$(grep "Request throughput (req/s):" "$bm_log" | sed 's/[^0-9.]//g')
+            e2el=$(grep "P99 E2EL (ms):" "$bm_log" | awk '{print $NF}')
+            goodput=$(grep "Request goodput (req/s):" "$bm_log" | sed 's/[^0-9.]//g')
+            if (( $(echo "$e2el <= $MAX_LATENCY_ALLOWED_MS" | bc -l) )); then
+                meet_latency_requirement=1
+                break
+            fi
+            request_rate=$((request_rate-1))
+        done
+    fi
+    # write the results and update the best result.
+    if ((meet_latency_requirement)); then
+        echo "max_num_seqs: $max_num_seqs, max_num_batched_tokens: $max_num_batched_tokens, request_rate: $request_rate, e2el: $e2el, through put: $through_put, goodput: $goodput"
+        echo "max_num_seqs: $max_num_seqs, max_num_batched_tokens: $max_num_batched_tokens, request_rate: $request_rate, e2el: $e2el, through put: $through_put, goodput: $goodput" >> "$RESULT"
+        if (( $(echo "$through_put > $best_throughput" | bc -l) )); then
+            best_throughput=$through_put
+            best_max_num_seqs=$max_num_seqs
+            best_num_batched_tokens=$max_num_batched_tokens
+            best_goodput=$goodput
+        fi
+    else
+        echo "max_num_seqs: $max_num_seqs, max_num_batched_tokens: $max_num_batched_tokens does not meet latency requirement ${MAX_LATENCY_ALLOWED_MS}"
+        echo "max_num_seqs: $max_num_seqs, max_num_batched_tokens: $max_num_batched_tokens does not meet latency requirement ${MAX_LATENCY_ALLOWED_MS}" >> "$RESULT"
+    fi
+
+    echo "best_max_num_seqs: $best_max_num_seqs, best_num_batched_tokens: $best_num_batched_tokens, best_throughput: $best_throughput"
+
+    echo "pkill -f vllm"
+    echo
+    pkill vllm
+    sleep 10
+    rm -f $vllm_log
+    printf '=%.0s' $(seq 1 20)
+    return 0
+}
+
+
+num_seqs_list="128 256"
+num_batched_tokens_list="512 1024 2048 4096"
+for num_seqs in $num_seqs_list; do
+    for num_batched_tokens in $num_batched_tokens_list; do
+        run_benchmark $num_seqs $num_batched_tokens
+        exit 0
+    done
+done
+echo "finish permutations"
+echo "best_max_num_seqs: $best_max_num_seqs, best_num_batched_tokens: $best_num_batched_tokens, best_throughput: $best_throughput"
+echo "best_max_num_seqs: $best_max_num_seqs, best_num_batched_tokens: $best_num_batched_tokens, best_throughput: $best_throughput" >> "$RESULT"
+
--- a/benchmarks/auto_tune/README.md
+++ b/benchmarks/auto_tune/README.md
@ -1,145 +0,0 @@
-# Automated vLLM Server Parameter Tuning
-
-This script automates the process of finding the optimal server parameter combination (`max-num-seqs` and `max-num-batched-tokens`) to maximize throughput for a vLLM server. It also supports additional constraints such as E2E latency and prefix cache hit rate.
-
-## Table of Contents
-
- [Prerequisites](#prerequisites)
- [Configuration](#configuration)
- [How to Run](#how-to-run)
- [Example Use Cases](#example-use-cases)
- [Output](#output)
- [How It Works](#how-it-works)
-
-## Prerequisites
-
-Before running the script, please ensure the following steps are completed:
-
-1. **Clone vLLM & Set Up Branch**: Clone the vLLM repository and check out to your desired branch.
-
-```bash
-git clone https://github.com/vllm-project/vllm.git
-cd vllm
-# git checkout <your-branch>
-```
-
-1. **Install Environment**: Install or update the correct running environment. For TPU usage, activate your `conda` environment and install the corresponding `torch` and `torch_xla` versions.
-
-2. **Model Configuration**: If you are using a customized model, ensure its configuration files are correctly placed and accessible.
-
-## Configuration
-
-You must set the following variables at the top of the script before execution.
-
-| Variable | Description | Example Value |
-| --- | --- | --- |
-| `BASE` | **Required.** The absolute path to the parent directory of your vLLM repository directory. | `"$HOME"` |
-| `MODEL` | **Required.** The Hugging Face model identifier to be served by vllm. | `"meta-llama/Llama-3.1-8B-Instruct"` |
-| `SYSTEM`| **Required.** The hardware you are running on. Choices: `TPU` or `GPU`. (For other systems, it might not support saving profiles) | `"TPU"` |
-| `TP` | **Required.** The tensor-parallelism size. | `1` |
-| `DOWNLOAD_DIR` | **Required.** Directory to download and load model weights from. | `""` (default download path) |
-| `INPUT_LEN` | **Required.** Request input length. | `4000` |
-| `OUTPUT_LEN` | **Required.** Request output length. | `16` |
-| `MAX_MODEL_LEN` | **Required.** Max model length. | `4096` |
-| `MIN_CACHE_HIT_PCT` | Prefix cache hit rate in percentage (0-100). Set to `0` to disable. | `60` |
-| `MAX_LATENCY_ALLOWED_MS` | The maximum allowed P99 end-to-end latency in milliseconds. Set to a very large number (e.g., `100000000000`) to effectively ignore the latency constraint. | `500` |
-| `NUM_SEQS_LIST` | A space-separated string of `max-num-seqs` values to test. | `"128 256"` |
-| `NUM_BATCHED_TOKENS_LIST` | A space-separated string of `max-num-batched-tokens` values to test. | `"1024 2048 4096"` |
-
-**Note**: The default `NUM_SEQS_LIST` and `NUM_BATCHED_TOKENS_LIST` are set for medium-sized inputs/outputs. For very short contexts (e.g., 20 input, 20 output tokens), you may need to test larger values for `max-num-seqs`.
-
-## How to Run
-
-1. **Configure**: Edit the script and set the variables in the [Configuration](#configuration) section.
-2. **Execute**: Run the script. Since the process can take a long time, it is highly recommended to use a terminal multiplexer like `tmux` or `screen` to prevent the script from stopping if your connection is lost.
-
-```bash
-cd <FOLDER_OF_THIS_SCRIPT>
-bash auto_tune.sh
-```
-
-    Please note that the `bash auto_tune.sh` command cannot contain full or partial path with keyword `vllm`, otherwise `pkill -f vllm` command will also kill this script itself.
-
-## Example Use Cases
-
-Here are a few examples of how to configure the script for different goals:
-
-### 1. Maximize Throughput (No Latency Constraint)
-
- **Goal**: Find the best `max-num-seqs` and `max-num-batched-tokens` to get the highest possible throughput for 1800 input tokens and 20 output tokens.
- **Configuration**:
-
-```bash
-INPUT_LEN=1800
-OUTPUT_LEN=20
-MAX_MODEL_LEN=2048
-MIN_CACHE_HIT_PCT=0
-MAX_LATENCY_ALLOWED_MS=100000000000 # A very large number
-```
-
-#### 2. Maximize Throughput with a Latency Requirement
-
- **Goal**: Find the best server parameters when P99 end-to-end latency must be below 500ms.
- **Configuration**:
-
-```bash
-INPUT_LEN=1800
-OUTPUT_LEN=20
-MAX_MODEL_LEN=2048
-MIN_CACHE_HIT_PCT=0
-MAX_LATENCY_ALLOWED_MS=500
-```
-
-#### 3. Maximize Throughput with Prefix Caching and Latency Requirements
-
- **Goal**: Find the best server parameters assuming a 60% prefix cache hit rate and a latency requirement of 500ms.
- **Configuration**:
-
-```bash
-INPUT_LEN=1800
-OUTPUT_LEN=20
-MAX_MODEL_LEN=2048
-MIN_CACHE_HIT_PCT=60
-MAX_LATENCY_ALLOWED_MS=500
-```
-
-## Output
-
-After the script finishes, you will find the results in a new, timestamped directory created inside `$BASE/auto-benchmark/`.
-
- **Log Files**: The directory (`$BASE/auto-benchmark/YYYY_MM_DD_HH_MM/`) contains detailed logs for each run:
-    - `vllm_log_...txt`: The log output from the vLLM server for each parameter combination.
-    - `bm_log_...txt`: The log output from the `vllm bench serve` command for each benchmark run.
-
- **Final Result Summary**: A file named `result.txt` is created in the log directory. It contains a summary of each tested combination and concludes with the overall best parameters found.
-
-```text
-# Example result.txt content
-hash:a1b2c3d4...
-max_num_seqs: 128, max_num_batched_tokens: 2048, request_rate: 10.0, e2el: 450.5, throughput: 9.8, goodput: 9.8
-max_num_seqs: 128, max_num_batched_tokens: 4096 does not meet latency requirement 500
-...
-best_max_num_seqs: 256, best_num_batched_tokens: 2048, best_throughput: 12.5, profile saved in: /home/user/vllm/auto-benchmark/2024_08_01_10_30/profile
-```
-
-  If it cannot find the best parameters, the final row will be `best_max_num_seqs: 0, best_num_batched_tokens: 0, best_throughput: 0`. This can be due to either the server not starting properly, or the latency requirement being too strict.
-
- **Profiler Trace**: A directory named `profile` is created inside the log directory. It contains the profiler trace file (e.g., `.xplane.pb` for TPU or a `.json` trace for GPU) from the single best-performing run.
-
-## How It Works
-
-The script follows a systematic process to find the optimal parameters:
-
-1. **Find Max GPU Memory Utilization**: The script first determines the highest safe `gpu-memory-utilization` (starting from 0.98 and decreasing) that does not cause an Out-Of-Memory (OOM) error when launching the server. This ensures the benchmark runs use the maximum available memory without crashing.
-
-2. **Iterate and Benchmark**: It then enters a nested loop, iterating through every combination of `max-num-seqs` and `max-num-batched-tokens` provided in the configuration lists.
-
-3. **Latency-Aware Throughput Search**: For each parameter combination:
-    - The vLLM server is started.
-    - A benchmark is first run with an infinite request rate (`--request-rate inf`).
-    - If the resulting P99 E2E latency is within the `MAX_LATENCY_ALLOWED_MS` limit, this throughput is considered the maximum for this configuration.
-    - If the latency is too high, the script performs a search by iteratively decreasing the request rate until the latency constraint is met. This finds the highest sustainable throughput for the given parameters and latency requirement.
-
-4. **Track Best Result**: Throughout the process, the script tracks the parameter combination that has yielded the highest valid throughput so far.
-
-5. **Profile Collection**: For the best-performing run, the script saves the vLLM profiler output, which can be used for deep-dive performance analysis with tools like TensorBoard.
--- a/benchmarks/auto_tune/auto_tune.sh
+++ b/benchmarks/auto_tune/auto_tune.sh
@ -1,292 +0,0 @@
-#!/bin/bash
-
-# This script aims to tune the best server parameter combinations to maximize throughput for given requirement.
-# See details in README (benchmarks/auto_tune/README.md).
-
-TAG=$(date +"%Y_%m_%d_%H_%M")
-SCRIPT_DIR=$( cd -- "$( dirname -- "${BASH_SOURCE[0]}" )" &> /dev/null && pwd )
-BASE="$SCRIPT_DIR/../../.."
-MODEL="meta-llama/Llama-3.1-8B-Instruct"
-SYSTEM="TPU"
-TP=1
-DOWNLOAD_DIR=""
-INPUT_LEN=4000
-OUTPUT_LEN=16
-MAX_MODEL_LEN=4096
-MIN_CACHE_HIT_PCT=0
-MAX_LATENCY_ALLOWED_MS=100000000000
-NUM_SEQS_LIST="128 256"
-NUM_BATCHED_TOKENS_LIST="512 1024 2048 4096"
-
-LOG_FOLDER="$BASE/auto-benchmark/$TAG"
-RESULT="$LOG_FOLDER/result.txt"
-PROFILE_PATH="$LOG_FOLDER/profile"
-
-echo "result file: $RESULT"
-echo "model: $MODEL"
-
-rm -rf $LOG_FOLDER
-rm -rf $PROFILE_PATH
-mkdir -p $LOG_FOLDER
-mkdir -p $PROFILE_PATH
-
-cd "$BASE/vllm"
-
-pip install -q datasets
-
-current_hash=$(git rev-parse HEAD)
-echo "hash:$current_hash" >> "$RESULT"
-echo "current_hash: $current_hash"
-
-TOTAL_LEN=$((INPUT_LEN + OUTPUT_LEN))
-RED='\033[0;31m'
-if (( TOTAL_LEN > MAX_MODEL_LEN )); then
-    echo -e "${RED}FAILED: INPUT_LEN($INPUT_LEN) + OUTPUT_LEN($OUTPUT_LEN) = $TOTAL_LEN, which is > MAX_MODEL_LEN = $MAX_MODEL_LEN.\033[0m" >&2
-    exit 1
-fi
-
-best_throughput=0
-best_max_num_seqs=0
-best_num_batched_tokens=0
-best_goodput=0
-best_request_rate=0
-
-start_server() {
-    local gpu_memory_utilization=$1
-    local max_num_seqs=$2
-    local max_num_batched_tokens=$3
-    local vllm_log=$4
-    local profile_dir=$5
-
-    pkill -if vllm
-
-    # Define the common arguments as a bash array.
-    # Each argument and its value are separate elements.
-    local common_args_array=(
-        "$MODEL"
-        "--disable-log-requests"
-        "--port" "8004"
-        "--gpu-memory-utilization" "$gpu_memory_utilization"
-        "--max-num-seqs" "$max_num_seqs"
-        "--max-num-batched-tokens" "$max_num_batched_tokens"
-        "--tensor-parallel-size" "$TP"
-        "--enable-prefix-caching"
-        "--load-format" "dummy"
-        "--download-dir" "$DOWNLOAD_DIR"
-        "--max-model-len" "$MAX_MODEL_LEN"
-    )
-
-    # Use the array expansion "${common_args_array[@]}"
-    # This correctly passes each element as a separate argument.
-    if [[ -n "$profile_dir" ]]; then
-        # Start server with profiling enabled
-        VLLM_USE_V1=1 VLLM_SERVER_DEV_MODE=1 VLLM_TORCH_PROFILER_DIR=$profile_dir \
-            vllm serve "${common_args_array[@]}" > "$vllm_log" 2>&1 &
-    else
-        # Start server without profiling
-        VLLM_USE_V1=1 VLLM_SERVER_DEV_MODE=1 \
-            vllm serve "${common_args_array[@]}" > "$vllm_log" 2>&1 &
-    fi
-
-    # wait for 10 minutes...
-    server_started=0
-    for i in {1..60}; do
-        RESPONSE=$(curl -s -X GET "http://0.0.0.0:8004/health" -w "%{http_code}" -o /dev/stdout)
-        STATUS_CODE=$(echo "$RESPONSE" | tail -n 1)
-        if [[ "$STATUS_CODE" -eq 200 ]]; then
-            server_started=1
-            break
-        else
-            sleep 10
-        fi
-    done
-
-    if (( ! server_started )); then
-        echo "server did not start within 10 minutes. Please check server log at $vllm_log".
-        return 1
-    else
-        return 0
-    fi
-}
-
-run_benchmark() {
-    local max_num_seqs=$1
-    local max_num_batched_tokens=$2
-    local gpu_memory_utilization=$3
-    echo "max_num_seq: $max_num_seqs, max_num_batched_tokens: $max_num_batched_tokens"
-    local vllm_log="$LOG_FOLDER/vllm_log_${max_num_seqs}_${max_num_batched_tokens}.txt"
-    echo "vllm_log: $vllm_log"
-    echo
-    rm -f $vllm_log
-    pkill -if vllm
-
-    echo "starting server..."
-    # Call start_server without a profile_dir to avoid profiling overhead
-    start_server $gpu_memory_utilization $max_num_seqs $max_num_batched_tokens $vllm_log ""
-    result=$?
-    if [[ "$result" -eq 1 ]]; then
-        echo "server failed to start. gpu_memory_utilization:$gpu_memory_utilization, max_num_seqs:$max_num_seqs, max_num_batched_tokens: $max_num_batched_tokens"
-    else
-        echo "server started."
-    fi
-    echo
-
-    echo "run benchmark test..."
-    meet_latency_requirement=0
-    # get a basic qps by using request-rate inf
-    bm_log="$LOG_FOLDER/bm_log_${max_num_seqs}_${max_num_batched_tokens}_requestrate_inf.txt"
-    prefix_len=$(( INPUT_LEN * MIN_CACHE_HIT_PCT / 100 ))
-    adjusted_input_len=$(( INPUT_LEN - prefix_len ))
-    # --profile flag is removed from this call
-    vllm bench serve \
-        --backend vllm \
-        --model $MODEL  \
-        --dataset-name random \
-        --random-input-len $adjusted_input_len \
-        --random-output-len $OUTPUT_LEN \
-        --ignore-eos \
-        --disable-tqdm \
-        --request-rate inf \
-        --percentile-metrics ttft,tpot,itl,e2el \
-        --goodput e2el:$MAX_LATENCY_ALLOWED_MS \
-        --num-prompts 1000 \
-        --random-prefix-len $prefix_len \
-        --port 8004 &> "$bm_log"
-    throughput=$(grep "Request throughput (req/s):" "$bm_log" | sed 's/[^0-9.]//g')
-    e2el=$(grep "P99 E2EL (ms):" "$bm_log" | awk '{print $NF}')
-    goodput=$(grep "Request goodput (req/s):" "$bm_log" | sed 's/[^0-9.]//g')
-
-    if (( $(echo "$e2el <= $MAX_LATENCY_ALLOWED_MS" | bc -l) )); then
-        meet_latency_requirement=1
-        request_rate=inf
-    fi
-
-    if (( ! meet_latency_requirement )); then
-    # start from request-rate as int(throughput) + 1
-        request_rate=$((${throughput%.*} + 1))
-        while ((request_rate > 0)); do
-            # clear prefix cache
-            curl -X POST http://0.0.0.0:8004/reset_prefix_cache
-            sleep 5
-            bm_log="$LOG_FOLDER/bm_log_${max_num_seqs}_${max_num_batched_tokens}_requestrate_${request_rate}.txt"
-            vllm bench serve \
-                --backend vllm \
-                --model $MODEL  \
-                --dataset-name random \
-                --random-input-len $adjusted_input_len \
-                --random-output-len $OUTPUT_LEN \
-                --ignore-eos \
-                --disable-tqdm \
-                --request-rate $request_rate \
-                --percentile-metrics ttft,tpot,itl,e2el \
-                --goodput e2el:$MAX_LATENCY_ALLOWED_MS \
-                --num-prompts 100 \
-                --random-prefix-len $prefix_len \
-                --port 8004 &> "$bm_log"
-            throughput=$(grep "Request throughput (req/s):" "$bm_log" | sed 's/[^0-9.]//g')
-            e2el=$(grep "P99 E2EL (ms):" "$bm_log" | awk '{print $NF}')
-            goodput=$(grep "Request goodput (req/s):" "$bm_log" | sed 's/[^0-9.]//g')
-            if (( $(echo "$e2el <= $MAX_LATENCY_ALLOWED_MS" | bc -l) )); then
-                meet_latency_requirement=1
-                break
-            fi
-            request_rate=$((request_rate-1))
-        done
-    fi
-    # write the results and update the best result.
-    if ((meet_latency_requirement)); then
-        echo "max_num_seqs: $max_num_seqs, max_num_batched_tokens: $max_num_batched_tokens, request_rate: $request_rate, e2el: $e2el, throughput: $throughput, goodput: $goodput"
-        echo "max_num_seqs: $max_num_seqs, max_num_batched_tokens: $max_num_batched_tokens, request_rate: $request_rate, e2el: $e2el, throughput: $throughput, goodput: $goodput" >> "$RESULT"
-        if (( $(echo "$throughput > $best_throughput" | bc -l) )); then
-            best_throughput=$throughput
-            best_max_num_seqs=$max_num_seqs
-            best_num_batched_tokens=$max_num_batched_tokens
-            best_goodput=$goodput
-            best_request_rate=$request_rate
-        fi
-    else
-        echo "max_num_seqs: $max_num_seqs, max_num_batched_tokens: $max_num_batched_tokens does not meet latency requirement ${MAX_LATENCY_ALLOWED_MS}"
-        echo "max_num_seqs: $max_num_seqs, max_num_batched_tokens: $max_num_batched_tokens does not meet latency requirement ${MAX_LATENCY_ALLOWED_MS}" >> "$RESULT"
-    fi
-
-    echo "best_max_num_seqs: $best_max_num_seqs, best_num_batched_tokens: $best_num_batched_tokens, best_throughput: $best_throughput"
-
-    pkill -if vllm
-    sleep 10
-    printf '=%.0s' $(seq 1 20)
-    return 0
-}
-
-read -r -a num_seqs_list <<< "$NUM_SEQS_LIST"
-read -r -a num_batched_tokens_list <<< "$NUM_BATCHED_TOKENS_LIST"
-
-# first find out the max gpu-memory-utilization without HBM OOM.
-gpu_memory_utilization=0.98
-find_gpu_memory_utilization=0
-while (( $(echo "$gpu_memory_utilization >= 0.9" | bc -l) )); do
-    # Pass empty string for profile_dir argument
-    start_server $gpu_memory_utilization "${num_seqs_list[-1]}" "${num_batched_tokens_list[-1]}" "$LOG_FOLDER/vllm_log_gpu_memory_utilization_$gpu_memory_utilization.log" ""
-    result=$?
-    if [[ "$result" -eq 0 ]]; then
-        find_gpu_memory_utilization=1
-        break
-    else
-        gpu_memory_utilization=$(echo "$gpu_memory_utilization - 0.01" | bc)
-    fi
-done
-
-if [[ "$find_gpu_memory_utilization" -eq 1 ]]; then
-    echo "Using gpu_memory_utilization=$gpu_memory_utilization to serve model."
-else
-    echo "Cannot find a proper gpu_memory_utilization over 0.9 to serve the model, please check logs in $LOG_FOLDER."
-    exit 1
-fi
-
-for num_seqs in "${num_seqs_list[@]}"; do
-    for num_batched_tokens in "${num_batched_tokens_list[@]}"; do
-        run_benchmark $num_seqs $num_batched_tokens $gpu_memory_utilization
-    done
-done
-echo "finish permutations"
-
-# =================================================================================
-# FINAL PROFILING RUN FOR THE BEST CONFIGURATION
-# =================================================================================
-if (( $(echo "$best_throughput > 0" | bc -l) )); then
-    echo
-    echo "Benchmark tuning finished. Now running profiling on the best configuration found..."
-    echo "Best config: max_num_seqs: $best_max_num_seqs, max_num_batched_tokens: $best_num_batched_tokens, throughput: $best_throughput"
-    echo
-
-    vllm_log="$LOG_FOLDER/vllm_log_BEST_PROFILE.txt"
-    bm_log="$LOG_FOLDER/bm_log_BEST_PROFILE.txt"
-
-    # Start server with the best params and profiling ENABLED
-    echo "Starting server for profiling..."
-    start_server $gpu_memory_utilization $best_max_num_seqs $best_num_batched_tokens "$vllm_log" "$PROFILE_PATH"
-
-    # Run benchmark with the best params and the --profile flag
-    echo "Running benchmark with profiling..."
-    prefix_len=$(( INPUT_LEN * MIN_CACHE_HIT_PCT / 100 ))
-    adjusted_input_len=$(( INPUT_LEN - prefix_len ))
-    vllm bench serve \
-        --backend vllm \
-        --model $MODEL \
-        --dataset-name random \
-        --random-input-len $adjusted_input_len \
-        --random-output-len $OUTPUT_LEN \
-        --ignore-eos \
-        --disable-tqdm \
-        --request-rate $best_request_rate \
-        --percentile-metrics ttft,tpot,itl,e2el \
-        --goodput e2el:$MAX_LATENCY_ALLOWED_MS \
-        --num-prompts 100 \
-        --random-prefix-len $prefix_len \
-        --port 8004 \
-        --profile &> "$bm_log"
-else
-    echo "No configuration met the latency requirements. Skipping final profiling run."
-fi
-pkill -if vllm
-echo "best_max_num_seqs: $best_max_num_seqs, best_num_batched_tokens: $best_num_batched_tokens, best_throughput: $best_throughput, profile saved in: $PROFILE_PATH"
-echo "best_max_num_seqs: $best_max_num_seqs, best_num_batched_tokens: $best_num_batched_tokens, best_throughput: $best_throughput, profile saved in: $PROFILE_PATH" >> "$RESULT"
--- a/benchmarks/backend_request_func.py
+++ b/benchmarks/backend_request_func.py
@ -31,10 +31,9 @@ class RequestFuncInput:
    model_name: Optional[str] = None
    logprobs: Optional[int] = None
    extra_body: Optional[dict] = None
-    multi_modal_content: Optional[dict | list[dict]] = None
+    multi_modal_content: Optional[dict] = None
    ignore_eos: bool = False
    language: Optional[str] = None
-    request_id: Optional[str] = None


@dataclass
@ -72,9 +71,6 @@ async def async_request_tgi(
            "inputs": request_func_input.prompt,
            "parameters": params,
        }
-        headers = None
-        if request_func_input.request_id:
-            headers = {"x-request-id": request_func_input.request_id}
        output = RequestFuncOutput()
        output.prompt_len = request_func_input.prompt_len
        if request_func_input.ignore_eos:
@ -86,9 +82,7 @@ async def async_request_tgi(
        st = time.perf_counter()
        most_recent_timestamp = st
        try:
-            async with session.post(
-                url=api_url, json=payload, headers=headers
-            ) as response:
+            async with session.post(url=api_url, json=payload) as response:
                if response.status == 200:
                    async for chunk_bytes in response.content:
                        chunk_bytes = chunk_bytes.strip()
@ -151,9 +145,6 @@ async def async_request_trt_llm(
        }
        if request_func_input.ignore_eos:
            payload["min_length"] = request_func_input.output_len
-        headers = None
-        if request_func_input.request_id:
-            headers = {"x-request-id": request_func_input.request_id}
        output = RequestFuncOutput()
        output.prompt_len = request_func_input.prompt_len

@ -161,9 +152,7 @@ async def async_request_trt_llm(
        st = time.perf_counter()
        most_recent_timestamp = st
        try:
-            async with session.post(
-                url=api_url, json=payload, headers=headers
-            ) as response:
+            async with session.post(url=api_url, json=payload) as response:
                if response.status == 200:
                    async for chunk_bytes in response.content:
                        chunk_bytes = chunk_bytes.strip()
@ -222,8 +211,6 @@ async def async_request_deepspeed_mii(
            "top_p": 1.0,
        }
        headers = {"Authorization": f"Bearer {os.environ.get('OPENAI_API_KEY')}"}
-        if request_func_input.request_id:
-            headers["x-request-id"] = request_func_input.request_id

        output = RequestFuncOutput()
        output.prompt_len = request_func_input.prompt_len
@ -296,8 +283,6 @@ async def async_request_openai_completions(
        if request_func_input.extra_body:
            payload.update(request_func_input.extra_body)
        headers = {"Authorization": f"Bearer {os.environ.get('OPENAI_API_KEY')}"}
-        if request_func_input.request_id:
-            headers["x-request-id"] = request_func_input.request_id

        output = RequestFuncOutput()
        output.prompt_len = request_func_input.prompt_len
@ -379,15 +364,7 @@ async def async_request_openai_chat_completions(
    ) as session:
        content = [{"type": "text", "text": request_func_input.prompt}]
        if request_func_input.multi_modal_content:
-            mm_content = request_func_input.multi_modal_content
-            if isinstance(mm_content, list):
-                content.extend(mm_content)
-            elif isinstance(mm_content, dict):
-                content.append(mm_content)
-            else:
-                raise TypeError(
-                    "multi_modal_content must be a dict or list[dict] for openai-chat"
-                )
+            content.append(request_func_input.multi_modal_content)
        payload = {
            "model": request_func_input.model_name
            if request_func_input.model_name
@ -410,8 +387,6 @@ async def async_request_openai_chat_completions(
            "Content-Type": "application/json",
            "Authorization": f"Bearer {os.environ.get('OPENAI_API_KEY')}",
        }
-        if request_func_input.request_id:
-            headers["x-request-id"] = request_func_input.request_id

        output = RequestFuncOutput()
        output.prompt_len = request_func_input.prompt_len
@ -429,14 +404,8 @@ async def async_request_openai_chat_completions(
                        chunk_bytes = chunk_bytes.strip()
                        if not chunk_bytes:
                            continue
-                        chunk_bytes = chunk_bytes.decode("utf-8")
-                        # NOTE: SSE comments (often used as pings) start with a colon.
-                        # These are not JSON data payload and should be skipped.
-                        if chunk_bytes.startswith(":"):
-                            continue
-
-                        chunk = chunk_bytes.removeprefix("data: ")

+                        chunk = chunk_bytes.decode("utf-8").removeprefix("data: ")
                        if chunk != "[DONE]":
                            timestamp = time.perf_counter()
                            data = json.loads(chunk)
@ -508,8 +477,6 @@ async def async_request_openai_audio(
        headers = {
            "Authorization": f"Bearer {os.environ.get('OPENAI_API_KEY')}",
        }
-        if request_func_input.request_id:
-            headers["x-request-id"] = request_func_input.request_id

        # Send audio file
        def to_bytes(y, sr):
@ -518,10 +485,7 @@ async def async_request_openai_audio(
            buffer.seek(0)
            return buffer

-        mm_audio = request_func_input.multi_modal_content
-        if not isinstance(mm_audio, dict) or "audio" not in mm_audio:
-            raise TypeError("multi_modal_content must be a dict containing 'audio'")
-        with to_bytes(*mm_audio["audio"]) as f:
+        with to_bytes(*request_func_input.multi_modal_content["audio"]) as f:
            form = aiohttp.FormData()
            form.add_field("file", f, content_type="audio/wav")
            for key, value in payload.items():
--- a/benchmarks/benchmark_block_pool.py
+++ b/benchmarks/benchmark_block_pool.py
@ -1,74 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-import gc
-
-from tabulate import tabulate
-
-from benchmark_utils import TimeCollector
-from vllm.utils import FlexibleArgumentParser
-from vllm.v1.core.block_pool import BlockPool
-
-
-def main(args):
-    rows = []
-    for allocate_block in args.allocate_blocks:
-        # Enforce a GC collect ahead to minimize the impact among runs
-        gc.collect()
-        block_pool = BlockPool(num_gpu_blocks=args.num_gpu_blocks, enable_caching=True)
-
-        get_blocks_times = TimeCollector(TimeCollector.US)
-        free_blocks_times = TimeCollector(TimeCollector.US)
-        for _ in range(args.num_iteration):
-            with get_blocks_times:
-                blocks = block_pool.get_new_blocks(allocate_block)
-            with free_blocks_times:
-                block_pool.free_blocks(blocks)
-
-        rows.append(
-            [get_blocks_times.cnt, args.num_gpu_blocks, allocate_block]
-            + get_blocks_times.dump_avg_max()
-            + free_blocks_times.dump_avg_max()
-        )
-
-    print(
-        tabulate(
-            rows,
-            headers=[
-                "Iterations",
-                "Total\nBlocks",
-                "Allocated\nBlocks",
-                "Get Blocks\nAvg (us)",
-                "Get Blocks\nMax (us)",
-                "Free Blocks\nAvg (us)",
-                "Free Blocks\nMax (us)",
-            ],
-            tablefmt="grid",
-            floatfmt=".3f",
-        )
-    )
-
-
-def invoke_main() -> None:
-    parser = FlexibleArgumentParser(
-        description="Benchmark the performance of BlockPool for KV Cache."
-    )
-    parser.add_argument("--num-gpu-blocks", type=int, default=100000)
-    parser.add_argument(
-        "--num-iteration",
-        type=int,
-        default=1000,
-        help="Number of iterations to run to stablize final data readings",
-    )
-    parser.add_argument(
-        "--allocate-blocks",
-        type=int,
-        nargs="*",
-        default=[10, 50, 100, 500, 1000],
-        help="Number of blocks to allocate",
-    )
-    args = parser.parse_args()
-    main(args)
-
-
-if __name__ == "__main__":
-    invoke_main()  # pragma: no cover
--- a/benchmarks/benchmark_dataset.py
+++ b/benchmarks/benchmark_dataset.py
@ -19,7 +19,6 @@ import logging
 import random
 from abc import ABC, abstractmethod
 from collections.abc import Mapping
-from copy import deepcopy
 from dataclasses import dataclass
 from functools import cache
 from io import BytesIO
@ -53,9 +52,8 @@ class SampleRequest:
    prompt: Union[str, Any]
    prompt_len: int
    expected_output_len: int
-    multi_modal_data: Optional[Union[MultiModalDataDict, dict, list[dict]]] = None
+    multi_modal_data: Optional[Union[MultiModalDataDict, dict]] = None
    lora_request: Optional[LoRARequest] = None
-    request_id: Optional[str] = None


 # -----------------------------------------------------------------------------
@ -157,10 +155,7 @@ class BenchmarkDataset(ABC):

    @abstractmethod
    def sample(
-        self,
-        tokenizer: PreTrainedTokenizerBase,
-        num_requests: int,
-        request_id_prefix: str = "",
+        self, tokenizer: PreTrainedTokenizerBase, num_requests: int
    ) -> list[SampleRequest]:
        """
        Abstract method to generate sample requests from the dataset.
@ -172,7 +167,6 @@ class BenchmarkDataset(ABC):
            tokenizer (PreTrainedTokenizerBase): The tokenizer to be used
             for processing the dataset's text.
            num_requests (int): The number of sample requests to generate.
-            request_id_prefix (str) The prefix of request_id.

        Returns:
            list[SampleRequest]: A list of sample requests generated from the
@ -181,10 +175,7 @@ class BenchmarkDataset(ABC):
        raise NotImplementedError("sample must be implemented in subclasses.")

    def maybe_oversample_requests(
-        self,
-        requests: list[SampleRequest],
-        num_requests: int,
-        request_id_prefix: str = "",
+        self, requests: list[SampleRequest], num_requests: int
    ) -> None:
        """
        Oversamples the list of requests if its size is less than the desired
@ -192,18 +183,11 @@ class BenchmarkDataset(ABC):

        Args:
            requests (List[SampleRequest]): The current list of sampled
-            requests.
-            num_requests (int): The target number of requests.
-            request_id_prefix (str) The prefix of the request ids.
+            requests.  num_requests (int): The target number of requests.
        """
        if len(requests) < num_requests:
            random.seed(self.random_seed)
-            additional = deepcopy(
-                random.choices(requests, k=num_requests - len(requests))
-            )
-            for i in range(len(additional)):
-                req = additional[i]
-                req.request_id = request_id_prefix + str(len(requests) + i)
+            additional = random.choices(requests, k=num_requests - len(requests))
            requests.extend(additional)
            logger.info("Oversampled requests to reach %d total samples.", num_requests)

@ -293,41 +277,6 @@ def process_image(image: Any) -> Mapping[str, Any]:
    )


-def process_video(video: Any) -> Mapping[str, Any]:
-    """
-    Process a single video input and return a multimedia content dictionary.
-
-    Supports the following input types:
-
-    1. Dictionary with raw video bytes: - Expects a dict with a 'bytes' key
-       containing raw video data.
-
-    2. String input: - Treats the string as a URL or local file path.  -
-       Prepends "file://" if the string doesn't start with "http://" or
-       "file://".  - Returns a dictionary with the image URL.
-
-    Raises:
-        ValueError: If the input is not a supported type.
-    """
-    if isinstance(video, dict) and "bytes" in video:
-        video_bytes = video["bytes"]
-        video_base64 = base64.b64encode(video_bytes).decode("utf-8")
-        return {
-            "type": "video_url",
-            "video_url": {"url": f"data:video/mp4;base64,{video_base64}"},
-        }
-
-    if isinstance(video, str):
-        video_url = (
-            video if video.startswith(("http://", "file://")) else f"file://{video}"
-        )
-        return {"type": "video_url", "video_url": {"url": video_url}}
-
-    raise ValueError(
-        f"Invalid video input {video}. Must be a string of local path/remote url, or a dictionary with raw video bytes in the form of `{{'bytes': raw_video_bytes}}`."  # noqa: E501
-    )
-
-
 # -----------------------------------------------------------------------------
 # Random Dataset Implementation (Synthetic Data)
 # -----------------------------------------------------------------------------
@ -354,7 +303,6 @@ class RandomDataset(BenchmarkDataset):
        range_ratio: float = DEFAULT_RANGE_RATIO,
        input_len: int = DEFAULT_INPUT_LEN,
        output_len: int = DEFAULT_OUTPUT_LEN,
-        request_id_prefix: str = "",
        **kwargs,
    ) -> list[SampleRequest]:
        # Enforce range_ratio < 1
@ -376,9 +324,6 @@ class RandomDataset(BenchmarkDataset):
        input_low = int(real_input_len * (1 - range_ratio))
        input_high = int(real_input_len * (1 + range_ratio))
        output_low = int(output_len * (1 - range_ratio))
-        # Ensure the lower bound for output length is at least 1 to prevent
-        # sampling 0 tokens, which can cause request failures.
-        output_low = max(output_low, 1)
        output_high = int(output_len * (1 + range_ratio))

        # Add logging for debugging
@ -404,21 +349,18 @@ class RandomDataset(BenchmarkDataset):
            # [1650, 939, 486] -> ['Ġcall', 'sh', 'ere']
            # To avoid uncontrolled change of the prompt length,
            # the encoded sequence is truncated before being decode again.
-            total_input_len = prefix_len + int(input_lens[i])
            re_encoded_sequence = tokenizer.encode(prompt, add_special_tokens=False)[
-                :total_input_len
+                : input_lens[i]
            ]
            prompt = tokenizer.decode(re_encoded_sequence)
-            total_input_len = len(re_encoded_sequence)
+            total_input_len = prefix_len + int(input_lens[i])
            requests.append(
                SampleRequest(
                    prompt=prompt,
                    prompt_len=total_input_len,
                    expected_output_len=int(output_lens[i]),
-                    request_id=request_id_prefix + str(i),
                )
            )
-
        return requests


@ -460,11 +402,9 @@ class ShareGPTDataset(BenchmarkDataset):
        max_loras: Optional[int] = None,
        output_len: Optional[int] = None,
        enable_multimodal_chat: bool = False,
-        request_id_prefix: str = "",
        **kwargs,
    ) -> list:
        samples: list = []
-        ind = 0
        for entry in self.data:
            if len(samples) >= num_requests:
                break
@ -486,26 +426,17 @@ class ShareGPTDataset(BenchmarkDataset):
                skip_min_output_len_check=output_len is not None,
            ):
                continue
-            if image_path := entry.get("image"):
-                mm_content = process_image(image_path)
-            elif video_path := entry.get("video"):
-                mm_content = process_video(video_path)
-            else:
-                mm_content = None
            if enable_multimodal_chat:
-                prompt = self.apply_multimodal_chat_transformation(prompt, mm_content)
+                prompt = self.apply_multimodal_chat_transformation(prompt, None)
            samples.append(
                SampleRequest(
                    prompt=prompt,
                    prompt_len=prompt_len,
                    expected_output_len=new_output_len,
                    lora_request=lora_request,
-                    multi_modal_data=mm_content,
-                    request_id=request_id_prefix + str(ind),
                )
            )
-            ind += 1
-        self.maybe_oversample_requests(samples, num_requests, request_id_prefix)
+        self.maybe_oversample_requests(samples, num_requests)
        return samples


@ -571,11 +502,10 @@ class CustomDataset(BenchmarkDataset):
        output_len: Optional[int] = None,
        enable_multimodal_chat: bool = False,
        skip_chat_template: bool = False,
-        request_id_prefix: str = "",
        **kwargs,
    ) -> list:
        sampled_requests = []
-        for i, item in enumerate(self.data):
+        for item in self.data:
            if len(sampled_requests) >= num_requests:
                break
            prompt = item["prompt"]
@ -594,12 +524,9 @@ class CustomDataset(BenchmarkDataset):
                    prompt=prompt,
                    prompt_len=prompt_len,
                    expected_output_len=output_len,
-                    request_id=request_id_prefix + str(i),
                )
            )
-        self.maybe_oversample_requests(
-            sampled_requests, num_requests, request_id_prefix
-        )
+        self.maybe_oversample_requests(sampled_requests, num_requests)

        return sampled_requests

@ -641,7 +568,6 @@ class SonnetDataset(BenchmarkDataset):
        input_len: int = DEFAULT_INPUT_LEN,
        output_len: int = DEFAULT_OUTPUT_LEN,
        return_prompt_formatted: bool = False,
-        request_id_prefix: str = "",
        **kwargs,
    ) -> list:
        # Calculate average token length for a poem line.
@ -667,7 +593,6 @@ class SonnetDataset(BenchmarkDataset):
        prefix_lines = self.data[:num_prefix_lines]

        samples = []
-        ind = 0
        while len(samples) < num_requests:
            extra_lines = random.choices(
                self.data, k=num_input_lines - num_prefix_lines
@ -678,17 +603,14 @@ class SonnetDataset(BenchmarkDataset):
                msg, add_generation_prompt=True, tokenize=False
            )
            prompt_len = len(tokenizer(prompt_formatted).input_ids)
-
            if prompt_len <= input_len:
                samples.append(
                    SampleRequest(
                        prompt=prompt_formatted if return_prompt_formatted else prompt,
                        prompt_len=prompt_len,
                        expected_output_len=output_len,
-                        request_id=request_id_prefix + str(ind),
                    )
                )
-                ind += 1
        return samples


@ -740,7 +662,6 @@ class BurstGPTDataset(BenchmarkDataset):
        num_requests: int,
        max_loras: Optional[int] = None,
        lora_path: Optional[str] = None,
-        request_id_prefix: str = "",
        **kwargs,
    ) -> list[SampleRequest]:
        samples = []
@ -762,7 +683,6 @@ class BurstGPTDataset(BenchmarkDataset):
                    prompt_len=input_len,
                    expected_output_len=output_len,
                    lora_request=lora_req,
-                    request_id=request_id_prefix + str(i),
                )
            )
        return samples
@ -780,7 +700,6 @@ class HuggingFaceDataset(BenchmarkDataset):
        self,
        dataset_path: str,
        dataset_split: str,
-        no_stream: bool = False,
        dataset_subset: Optional[str] = None,
        **kwargs,
    ) -> None:
@ -788,7 +707,6 @@ class HuggingFaceDataset(BenchmarkDataset):

        self.dataset_split = dataset_split
        self.dataset_subset = dataset_subset
-        self.load_stream = not no_stream
        self.load_data()

    def load_data(self) -> None:
@ -797,7 +715,7 @@ class HuggingFaceDataset(BenchmarkDataset):
            self.dataset_path,
            name=self.dataset_subset,
            split=self.dataset_split,
-            streaming=self.load_stream,
+            streaming=True,
        )
        self.data = self.data.shuffle(seed=self.random_seed)

@ -822,14 +740,12 @@ class ConversationDataset(HuggingFaceDataset):
        num_requests: int,
        output_len: Optional[int] = None,
        enable_multimodal_chat: bool = False,
-        request_id_prefix: str = "",
        **kwargs,
    ) -> list:
        # Filter examples with at least 2 conversations
        filtered_data = self.data.filter(lambda x: len(x["conversations"]) >= 2)
        sampled_requests = []
        dynamic_output = output_len is None
-        ind = 0

        for item in filtered_data:
            if len(sampled_requests) >= num_requests:
@ -857,13 +773,9 @@ class ConversationDataset(HuggingFaceDataset):
                    prompt_len=prompt_len,
                    expected_output_len=output_len,
                    multi_modal_data=mm_content,
-                    request_id=request_id_prefix + str(ind),
                )
            )
-            ind += 1
-        self.maybe_oversample_requests(
-            sampled_requests, num_requests, request_id_prefix
-        )
+        self.maybe_oversample_requests(sampled_requests, num_requests)
        return sampled_requests


@ -890,12 +802,11 @@ class VisionArenaDataset(HuggingFaceDataset):
        num_requests: int,
        output_len: Optional[int] = None,
        enable_multimodal_chat: bool = False,
-        request_id_prefix: str = "",
        **kwargs,
    ) -> list:
        output_len = output_len if output_len is not None else self.DEFAULT_OUTPUT_LEN
        sampled_requests = []
-        for i, item in enumerate(self.data):
+        for item in self.data:
            if len(sampled_requests) >= num_requests:
                break
            parser_fn = self.SUPPORTED_DATASET_PATHS.get(self.dataset_path)
@ -915,12 +826,9 @@ class VisionArenaDataset(HuggingFaceDataset):
                    prompt_len=prompt_len,
                    expected_output_len=output_len,
                    multi_modal_data=mm_content,
-                    request_id=request_id_prefix + str(i),
                )
            )
-        self.maybe_oversample_requests(
-            sampled_requests, num_requests, request_id_prefix
-        )
+        self.maybe_oversample_requests(sampled_requests, num_requests)
        return sampled_requests


@ -950,18 +858,15 @@ class InstructCoderDataset(HuggingFaceDataset):
        num_requests: int,
        output_len: Optional[int] = None,
        enable_multimodal_chat: bool = False,
-        request_id_prefix: str = "",
        **kwargs,
    ) -> list:
        output_len = output_len if output_len is not None else self.DEFAULT_OUTPUT_LEN
        sampled_requests = []
-        for i, item in enumerate(self.data):
+        for item in self.data:
            if len(sampled_requests) >= num_requests:
                break
-            prompt = (
-                f"{item['input']}\n\n{item['instruction']} Just output "
-                "the code, do not include any explanation."
-            )
+            prompt = f"{item['input']}\n\n{item['instruction']} Just output \
+            the code, do not include any explanation."

            # apply template
            prompt = tokenizer.apply_chat_template(
@ -975,12 +880,9 @@ class InstructCoderDataset(HuggingFaceDataset):
                    prompt=prompt,
                    prompt_len=prompt_len,
                    expected_output_len=output_len,
-                    request_id=request_id_prefix + str(i),
                )
            )
-        self.maybe_oversample_requests(
-            sampled_requests, num_requests, request_id_prefix
-        )
+        self.maybe_oversample_requests(sampled_requests, num_requests)
        return sampled_requests


@ -1010,13 +912,12 @@ class MTBenchDataset(HuggingFaceDataset):
        num_requests: int,
        output_len: Optional[int] = None,
        enable_multimodal_chat: bool = False,
-        request_id_prefix: str = "",
        **kwargs,
    ) -> list:
        output_len = output_len if output_len is not None else self.DEFAULT_OUTPUT_LEN
        sampled_requests = []

-        for i, item in enumerate(self.data):
+        for item in self.data:
            if len(sampled_requests) >= num_requests:
                break
            prompt = item["turns"][0]
@ -1034,12 +935,9 @@ class MTBenchDataset(HuggingFaceDataset):
                    prompt=prompt,
                    prompt_len=prompt_len,
                    expected_output_len=output_len,
-                    request_id=request_id_prefix + str(i),
                )
            )
-        self.maybe_oversample_requests(
-            sampled_requests, num_requests, request_id_prefix
-        )
+        self.maybe_oversample_requests(sampled_requests, num_requests)
        return sampled_requests


@ -1064,12 +962,10 @@ class AIMODataset(HuggingFaceDataset):
        tokenizer: PreTrainedTokenizerBase,
        num_requests: int,
        output_len: Optional[int] = None,
-        request_id_prefix: str = "",
        **kwargs,
    ) -> list:
        sampled_requests = []
        dynamic_output = output_len is None
-        ind = 0

        for item in self.data:
            if len(sampled_requests) >= num_requests:
@ -1092,13 +988,9 @@ class AIMODataset(HuggingFaceDataset):
                    prompt_len=prompt_len,
                    expected_output_len=output_len,
                    multi_modal_data=None,
-                    request_id=request_id_prefix + str(ind),
                )
            )
-            ind += 1
-        self.maybe_oversample_requests(
-            sampled_requests, num_requests, request_id_prefix
-        )
+        self.maybe_oversample_requests(sampled_requests, num_requests)
        return sampled_requests


@ -1168,18 +1060,12 @@ class NextEditPredictionDataset(HuggingFaceDataset):
        "zed-industries/zeta": _format_zeta_prompt,
    }

-    def sample(
-        self,
-        tokenizer: PreTrainedTokenizerBase,
-        num_requests: int,
-        request_id_prefix: str = "",
-        **kwargs,
-    ):
+    def sample(self, tokenizer: PreTrainedTokenizerBase, num_requests: int, **kwargs):
        formatting_prompt_func = self.MAPPING_PROMPT_FUNCS.get(self.dataset_path)
        if formatting_prompt_func is None:
            raise ValueError(f"Unsupported dataset path: {self.dataset_path}")
        samples = []
-        for i, sample in enumerate(self.data):
+        for sample in self.data:
            sample = formatting_prompt_func(sample)
            samples.append(
                SampleRequest(
@ -1188,12 +1074,11 @@ class NextEditPredictionDataset(HuggingFaceDataset):
                    expected_output_len=len(
                        tokenizer(sample["expected_output"]).input_ids
                    ),
-                    request_id=request_id_prefix + str(i),
                )
            )
            if len(samples) >= num_requests:
                break
-        self.maybe_oversample_requests(samples, num_requests, request_id_prefix)
+        self.maybe_oversample_requests(samples, num_requests)
        return samples


@ -1242,7 +1127,6 @@ class ASRDataset(HuggingFaceDataset):
        tokenizer: PreTrainedTokenizerBase,
        num_requests: int,
        output_len: Optional[int] = None,
-        request_id_prefix: str = "",
        **kwargs,
    ) -> list:
        import librosa
@ -1252,7 +1136,6 @@ class ASRDataset(HuggingFaceDataset):
        prompt_len = len(tokenizer(prompt).input_ids)
        sampled_requests = []
        skipped = 0
-        ind = 0
        for item in self.data:
            if len(sampled_requests) >= num_requests:
                break
@ -1271,10 +1154,8 @@ class ASRDataset(HuggingFaceDataset):
                    prompt_len=prompt_len,
                    expected_output_len=output_len,
                    multi_modal_data=mm_content,
-                    request_id=request_id_prefix + str(ind),
                )
            )
-            ind += 1
        if skipped:
            logger.warning(
                "%d samples discarded from dataset due to"
@ -1282,7 +1163,5 @@ class ASRDataset(HuggingFaceDataset):
                " what Whisper supports.",
                skipped,
            )
-        self.maybe_oversample_requests(
-            sampled_requests, num_requests, request_id_prefix
-        )
+        self.maybe_oversample_requests(sampled_requests, num_requests)
        return sampled_requests
--- a/benchmarks/benchmark_latency.py
+++ b/benchmarks/benchmark_latency.py
@ -11,7 +11,6 @@ from typing import Any, Optional

 import numpy as np
 from tqdm import tqdm
-from typing_extensions import deprecated

 import vllm.envs as envs
 from benchmark_utils import convert_to_pytorch_benchmark_format, write_to_json
@ -35,10 +34,6 @@ def save_to_pytorch_benchmark_format(
        write_to_json(pt_file, pt_records)


-@deprecated(
-    "benchmark_latency.py is deprecated and will be removed in a "
-    "future version. Please use 'vllm bench latency' instead.",
-)
 def main(args: argparse.Namespace):
    print(args)

@ -128,7 +123,7 @@ def main(args: argparse.Namespace):
        save_to_pytorch_benchmark_format(args, results)


-def create_argument_parser():
+if __name__ == "__main__":
    parser = FlexibleArgumentParser(
        description="Benchmark the latency of processing a single batch of "
        "requests till completion."
@ -176,12 +171,6 @@ def create_argument_parser():
    # V1 enables prefix caching by default which skews the latency
    # numbers. We need to disable prefix caching by default.
    parser.set_defaults(enable_prefix_caching=False)
-
-    return parser
-
-
-if __name__ == "__main__":
-    parser = create_argument_parser()
    args = parser.parse_args()
    if args.profile and not envs.VLLM_TORCH_PROFILER_DIR:
        raise OSError(
--- a/benchmarks/benchmark_long_document_qa_throughput.py
+++ b/benchmarks/benchmark_long_document_qa_throughput.py
@ -142,7 +142,7 @@ def main(args):
    )


-def create_argument_parser():
+if __name__ == "__main__":
    parser = FlexibleArgumentParser(
        description="Benchmark the performance with or "
        "without automatic prefix caching."
@ -192,11 +192,5 @@ def create_argument_parser():
    )

    parser = EngineArgs.add_cli_args(parser)
-
-    return parser
-
-
-if __name__ == "__main__":
-    parser = create_argument_parser()
    args = parser.parse_args()
    main(args)
--- a/benchmarks/benchmark_ngram_proposer.py
+++ b/benchmarks/benchmark_ngram_proposer.py
@ -1,112 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-import gc
-
-import numpy as np
-from tabulate import tabulate
-
-from benchmark_utils import TimeCollector
-from vllm.config import ModelConfig, SpeculativeConfig, VllmConfig
-from vllm.utils import FlexibleArgumentParser
-from vllm.v1.spec_decode.ngram_proposer import NgramProposer
-
-
-def main(args):
-    rows = []
-    for max_ngram in args.max_ngram:
-        collector = TimeCollector(TimeCollector.US)
-
-        model_config = ModelConfig(
-            model="facebook/opt-125m",
-            task="generate",
-            max_model_len=args.num_token + args.num_spec_token,
-            tokenizer="facebook/opt-125m",
-            tokenizer_mode="auto",
-            dtype="auto",
-            seed=None,
-            trust_remote_code=False,
-        )
-        proposer = NgramProposer(
-            vllm_config=VllmConfig(
-                model_config=model_config,
-                speculative_config=SpeculativeConfig(
-                    prompt_lookup_min=args.min_ngram,
-                    prompt_lookup_max=max_ngram,
-                    num_speculative_tokens=args.num_spec_token,
-                    method="ngram",
-                ),
-            )
-        )
-
-        # Warm up
-        proposer.propose(np.random.randint(0, 20, (args.num_token,)))
-
-        gc.collect()
-        for _ in range(args.num_iteration):
-            tokens = np.random.randint(0, 20, (args.num_req, args.num_token))
-            with collector:
-                for i in range(args.num_req):
-                    proposer.propose(tokens[i, :])
-        rows.append(
-            [args.num_req, args.num_token, args.min_ngram, max_ngram]
-            + collector.dump_avg_max()
-        )
-
-    print(
-        tabulate(
-            rows,
-            headers=[
-                "# Request",
-                "# Token",
-                "Min Ngram",
-                "Max Ngram",
-                "Avg (us)",
-                "Max (us)",
-            ],
-            tablefmt="grid",
-            floatfmt=".3f",
-        )
-    )
-
-
-def invoke_main() -> None:
-    parser = FlexibleArgumentParser(
-        description="Benchmark the performance of N-gram speculative decode drafting"
-    )
-    parser.add_argument(
-        "--num-iteration",
-        type=int,
-        default=100,
-        help="Number of iterations to run to stablize final data readings",
-    )
-    parser.add_argument(
-        "--num-req", type=int, default=128, help="Number of requests in the batch"
-    )
-    parser.add_argument(
-        "--num-token", type=int, default=1500, help="Number of tokens for each request"
-    )
-    parser.add_argument(
-        "--min-ngram",
-        type=int,
-        default=3,
-        help="Minimum n-gram to match",
-    )
-    parser.add_argument(
-        "--max-ngram",
-        type=int,
-        nargs="*",
-        default=[5, 7, 10, 15, 20],
-        help="Maximum n-gram to match",
-    )
-    parser.add_argument(
-        "--num-spec-token",
-        type=int,
-        default=3,
-        help="Number of speculative tokens to generate",
-    )
-    args = parser.parse_args()
-    main(args)
-
-
-if __name__ == "__main__":
-    invoke_main()  # pragma: no cover
--- a/benchmarks/benchmark_prefix_caching.py
+++ b/benchmarks/benchmark_prefix_caching.py
@ -218,7 +218,7 @@ def main(args):
    )


-def create_argument_parser():
+if __name__ == "__main__":
    parser = FlexibleArgumentParser(
        description="Benchmark the performance with or without "
        "automatic prefix caching."
@ -268,11 +268,5 @@ def create_argument_parser():
    )

    parser = EngineArgs.add_cli_args(parser)
-
-    return parser
-
-
-if __name__ == "__main__":
-    parser = create_argument_parser()
    args = parser.parse_args()
    main(args)
--- a/benchmarks/benchmark_prioritization.py
+++ b/benchmarks/benchmark_prioritization.py
@ -161,7 +161,7 @@ def main(args: argparse.Namespace):
            json.dump(results, f, indent=4)


-def create_argument_parser():
+if __name__ == "__main__":
    parser = FlexibleArgumentParser(description="Benchmark the throughput.")
    parser.add_argument(
        "--backend", type=str, choices=["vllm", "hf", "mii"], default="vllm"
@ -204,12 +204,6 @@ def create_argument_parser():
    )

    parser = EngineArgs.add_cli_args(parser)
-
-    return parser
-
-
-if __name__ == "__main__":
-    parser = create_argument_parser()
    args = parser.parse_args()
    if args.tokenizer is None:
        args.tokenizer = args.model
--- a/benchmarks/benchmark_serving.py
+++ b/benchmarks/benchmark_serving.py
@ -5,7 +5,8 @@ r"""Benchmark online serving throughput.
 On the server side, run one of the following commands:
    vLLM OpenAI API server
    vllm serve <your_model> \
-        --swap-space 16
+        --swap-space 16 \
+        --disable-log-requests

 On the client side, run:
    python benchmarks/benchmark_serving.py \
@ -29,15 +30,14 @@ import os
 import random
 import time
 import warnings
-from collections.abc import Iterable
+from collections.abc import AsyncGenerator, Iterable
 from dataclasses import dataclass
 from datetime import datetime
-from typing import Any, Literal, Optional
+from typing import Any, Optional

 import numpy as np
 from tqdm.asyncio import tqdm
 from transformers import PreTrainedTokenizerBase
-from typing_extensions import deprecated

 from backend_request_func import (
    ASYNC_REQUEST_FUNCS,
@ -73,7 +73,6 @@ from benchmark_dataset import (
    VisionArenaDataset,
 )
 from benchmark_utils import convert_to_pytorch_benchmark_format, write_to_json
-from vllm.benchmarks.serve import get_request

 MILLISECONDS_TO_SECONDS_CONVERSION = 1000

@ -108,6 +107,51 @@ class BenchmarkMetrics:
    percentiles_e2el_ms: list[tuple[float, float]]


+async def get_request(
+    input_requests: list[SampleRequest],
+    request_rate: float,
+    burstiness: float = 1.0,
+) -> AsyncGenerator[SampleRequest, None]:
+    """
+    Asynchronously generates requests at a specified rate
+    with OPTIONAL burstiness.
+
+    Args:
+        input_requests:
+            A list of input requests, each represented as a SampleRequest.
+        request_rate:
+            The rate at which requests are generated (requests/s).
+        burstiness (optional):
+            The burstiness factor of the request generation.
+            Only takes effect when request_rate is not inf.
+            Default value is 1, which follows a Poisson process.
+            Otherwise, the request intervals follow a gamma distribution.
+            A lower burstiness value (0 < burstiness < 1) results
+            in more bursty requests, while a higher burstiness value
+            (burstiness > 1) results in a more uniform arrival of requests.
+    """
+    input_requests: Iterable[SampleRequest] = iter(input_requests)
+
+    # Calculate scale parameter theta to maintain the desired request_rate.
+    assert burstiness > 0, (
+        f"A positive burstiness factor is expected, but given {burstiness}."
+    )
+    theta = 1.0 / (request_rate * burstiness)
+
+    for request in input_requests:
+        yield request
+
+        if request_rate == float("inf"):
+            # If the request rate is infinity, then we don't need to wait.
+            continue
+
+        # Sample the request interval from the gamma distribution.
+        # If burstiness is 1, it follows exponential distribution.
+        interval = np.random.gamma(shape=burstiness, scale=theta)
+        # The next request will be sent after the interval.
+        await asyncio.sleep(interval)
+
+
 def calculate_metrics(
    input_requests: list[SampleRequest],
    outputs: list[RequestFuncOutput],
@ -246,9 +290,6 @@ async def benchmark(
    max_concurrency: Optional[int],
    lora_modules: Optional[Iterable[str]],
    extra_body: Optional[dict],
-    ramp_up_strategy: Optional[Literal["linear", "exponential"]] = None,
-    ramp_up_start_rps: Optional[int] = None,
-    ramp_up_end_rps: Optional[int] = None,
 ):
    if backend in ASYNC_REQUEST_FUNCS:
        request_func = ASYNC_REQUEST_FUNCS[backend]
@ -263,14 +304,7 @@ async def benchmark(
        input_requests[0].multi_modal_data,
    )

-    assert (
-        test_mm_content is None
-        or isinstance(test_mm_content, dict)
-        or (
-            isinstance(test_mm_content, list)
-            and all(isinstance(item, dict) for item in test_mm_content)
-        )
-    ), "multi_modal_data must be a dict or list[dict]"
+    assert test_mm_content is None or isinstance(test_mm_content, dict)
    test_input = RequestFuncInput(
        model=model_id,
        model_name=model_name,
@ -319,15 +353,7 @@ async def benchmark(

    distribution = "Poisson process" if burstiness == 1.0 else "Gamma distribution"

-    if ramp_up_strategy is not None:
-        print(
-            f"Traffic ramp-up strategy: {ramp_up_strategy}. Will increase "
-            f"RPS from {ramp_up_start_rps} to {ramp_up_end_rps} RPS over "
-            "the duration of the benchmark."
-        )
-    else:
-        print(f"Traffic request rate: {request_rate} RPS.")
-
+    print(f"Traffic request rate: {request_rate}")
    print(f"Burstiness factor: {burstiness} ({distribution})")
    print(f"Maximum request concurrency: {max_concurrency}")

@ -347,40 +373,12 @@ async def benchmark(

    benchmark_start_time = time.perf_counter()
    tasks: list[asyncio.Task] = []
-
-    rps_change_events = []
-    last_int_rps = -1
-    if ramp_up_strategy is not None and ramp_up_start_rps is not None:
-        last_int_rps = ramp_up_start_rps
-        rps_change_events.append(
-            {
-                "rps": last_int_rps,
-                "timestamp": datetime.now().isoformat(),
-            }
-        )
-
-    async for request, current_request_rate in get_request(
-        input_requests,
-        request_rate,
-        burstiness,
-        ramp_up_strategy,
-        ramp_up_start_rps,
-        ramp_up_end_rps,
-    ):
-        if ramp_up_strategy is not None:
-            current_int_rps = int(current_request_rate)
-            if current_int_rps > last_int_rps:
-                timestamp = datetime.now().isoformat()
-                for rps_val in range(last_int_rps + 1, current_int_rps + 1):
-                    rps_change_events.append({"rps": rps_val, "timestamp": timestamp})
-                last_int_rps = current_int_rps
-
-        prompt, prompt_len, output_len, mm_content, request_id = (
+    async for request in get_request(input_requests, request_rate, burstiness):
+        prompt, prompt_len, output_len, mm_content = (
            request.prompt,
            request.prompt_len,
            request.expected_output_len,
            request.multi_modal_data,
-            request.request_id,
        )
        req_model_id, req_model_name = model_id, model_name
        if lora_modules:
@ -398,12 +396,28 @@ async def benchmark(
            multi_modal_content=mm_content,
            ignore_eos=ignore_eos,
            extra_body=extra_body,
-            request_id=request_id,
        )
-        task = limited_request_func(request_func_input=request_func_input, pbar=pbar)
-        tasks.append(asyncio.create_task(task))
+        tasks.append(
+            asyncio.create_task(
+                limited_request_func(request_func_input=request_func_input, pbar=pbar)
+            )
+        )
    outputs: list[RequestFuncOutput] = await asyncio.gather(*tasks)

+    if profile:
+        print("Stopping profiler...")
+        profile_input = RequestFuncInput(
+            model=model_id,
+            prompt=test_prompt,
+            api_url=base_url + "/stop_profile",
+            prompt_len=test_prompt_len,
+            output_len=test_output_len,
+            logprobs=logprobs,
+        )
+        profile_output = await request_func(request_func_input=profile_input)
+        if profile_output.success:
+            print("Profiler stopped")
+
    if pbar is not None:
        pbar.close()

@ -421,10 +435,6 @@ async def benchmark(

    print("{s:{c}^{n}}".format(s=" Serving Benchmark Result ", n=50, c="="))
    print("{:<40} {:<10}".format("Successful requests:", metrics.completed))
-    if max_concurrency is not None:
-        print("{:<40} {:<10}".format("Maximum request concurrency:", max_concurrency))
-    if request_rate != float("inf"):
-        print("{:<40} {:<10.2f}".format("Request rate configured (RPS):", request_rate))
    print("{:<40} {:<10.2f}".format("Benchmark duration (s):", benchmark_duration))
    print("{:<40} {:<10}".format("Total input tokens:", metrics.total_input))
    print("{:<40} {:<10}".format("Total generated tokens:", metrics.total_output))
@ -456,7 +466,7 @@ async def benchmark(
        "total_input_tokens": metrics.total_input,
        "total_output_tokens": metrics.total_output,
        "request_throughput": metrics.request_throughput,
-        "request_goodput": metrics.request_goodput if goodput_config_dict else None,
+        "request_goodput:": metrics.request_goodput if goodput_config_dict else None,
        "output_throughput": metrics.output_throughput,
        "total_token_throughput": metrics.total_token_throughput,
        "input_lens": [output.prompt_len for output in outputs],
@ -467,9 +477,6 @@ async def benchmark(
        "errors": [output.error for output in outputs],
    }

-    if rps_change_events:
-        result["rps_change_events"] = rps_change_events
-
    def process_one_metric(
        # E.g., "ttft"
        metric_attribute_name: str,
@ -516,20 +523,6 @@ async def benchmark(

    print("=" * 50)

-    if profile:
-        print("Stopping profiler...")
-        profile_input = RequestFuncInput(
-            model=model_id,
-            prompt=test_prompt,
-            api_url=base_url + "/stop_profile",
-            prompt_len=test_prompt_len,
-            output_len=test_output_len,
-            logprobs=logprobs,
-        )
-        profile_output = await request_func(request_func_input=profile_input)
-        if profile_output.success:
-            print("Profiler stopped")
-
    return result


@ -606,10 +599,6 @@ def save_to_pytorch_benchmark_format(
        write_to_json(pt_file, pt_records)


-@deprecated(
-    "benchmark_serving.py is deprecated and will be removed in a future "
-    "version. Please use 'vllm bench serve' instead.",
-)
 def main(args: argparse.Namespace):
    print(args)
    random.seed(args.seed)
@ -621,26 +610,6 @@ def main(args: argparse.Namespace):
    tokenizer_id = args.tokenizer if args.tokenizer is not None else args.model
    tokenizer_mode = args.tokenizer_mode

-    # Validate ramp-up arguments
-    if args.ramp_up_strategy is not None:
-        if args.request_rate != float("inf"):
-            raise ValueError(
-                "When using ramp-up, do not specify --request-rate. "
-                "The request rate will be controlled by ramp-up parameters. "
-                "Please remove the --request-rate argument."
-            )
-        if args.ramp_up_start_rps is None or args.ramp_up_end_rps is None:
-            raise ValueError(
-                "When using --ramp-up-strategy, both --ramp-up-start-rps and "
-                "--ramp-up-end-rps must be specified"
-            )
-        if args.ramp_up_start_rps < 0 or args.ramp_up_end_rps < 0:
-            raise ValueError("Ramp-up start and end RPS must be non-negative")
-        if args.ramp_up_start_rps > args.ramp_up_end_rps:
-            raise ValueError("Ramp-up start RPS must be less than end RPS")
-        if args.ramp_up_strategy == "exponential" and args.ramp_up_start_rps == 0:
-            raise ValueError("For exponential ramp-up, the start RPS cannot be 0.")
-
    if args.base_url is not None:
        api_url = f"{args.base_url}{args.endpoint}"
        base_url = f"{args.base_url}"
@ -667,7 +636,6 @@ def main(args: argparse.Namespace):
            tokenizer=tokenizer,
            output_len=args.custom_output_len,
            skip_chat_template=args.custom_skip_chat_template,
-            request_id_prefix=args.request_id_prefix,
        )

    elif args.dataset_name == "sonnet":
@ -681,7 +649,6 @@ def main(args: argparse.Namespace):
                prefix_len=args.sonnet_prefix_len,
                tokenizer=tokenizer,
                return_prompt_formatted=False,
-                request_id_prefix=args.request_id_prefix,
            )
        else:
            assert tokenizer.chat_template or tokenizer.default_chat_template, (
@ -694,7 +661,6 @@ def main(args: argparse.Namespace):
                prefix_len=args.sonnet_prefix_len,
                tokenizer=tokenizer,
                return_prompt_formatted=True,
-                request_id_prefix=args.request_id_prefix,
            )

    elif args.dataset_name == "hf":
@ -751,12 +717,10 @@ def main(args: argparse.Namespace):
            dataset_subset=args.hf_subset,
            dataset_split=args.hf_split,
            random_seed=args.seed,
-            no_stream=args.no_stream,
        ).sample(
            num_requests=args.num_prompts,
            tokenizer=tokenizer,
            output_len=args.hf_output_len,
-            request_id_prefix=args.request_id_prefix,
        )

    else:
@ -768,15 +732,10 @@ def main(args: argparse.Namespace):
                tokenizer=tokenizer,
                num_requests=args.num_prompts,
                output_len=args.sharegpt_output_len,
-                request_id_prefix=args.request_id_prefix,
            ),
            "burstgpt": lambda: BurstGPTDataset(
                random_seed=args.seed, dataset_path=args.dataset_path
-            ).sample(
-                tokenizer=tokenizer,
-                num_requests=args.num_prompts,
-                request_id_prefix=args.request_id_prefix,
-            ),
+            ).sample(tokenizer=tokenizer, num_requests=args.num_prompts),
            "random": lambda: RandomDataset(dataset_path=args.dataset_path).sample(
                tokenizer=tokenizer,
                num_requests=args.num_prompts,
@ -784,7 +743,6 @@ def main(args: argparse.Namespace):
                input_len=args.random_input_len,
                output_len=args.random_output_len,
                range_ratio=args.random_range_ratio,
-                request_id_prefix=args.request_id_prefix,
            ),
        }

@ -844,9 +802,6 @@ def main(args: argparse.Namespace):
            max_concurrency=args.max_concurrency,
            lora_modules=args.lora_modules,
            extra_body=sampling_params,
-            ramp_up_strategy=args.ramp_up_strategy,
-            ramp_up_start_rps=args.ramp_up_start_rps,
-            ramp_up_end_rps=args.ramp_up_end_rps,
        )
    )

@ -879,11 +834,6 @@ def main(args: argparse.Namespace):
        result_json["burstiness"] = args.burstiness
        result_json["max_concurrency"] = args.max_concurrency

-        if args.ramp_up_strategy is not None:
-            result_json["ramp_up_strategy"] = args.ramp_up_strategy
-            result_json["ramp_up_start_rps"] = args.ramp_up_start_rps
-            result_json["ramp_up_end_rps"] = args.ramp_up_end_rps
-
        # Merge with benchmark result
        result_json = {**result_json, **benchmark_result}

@ -909,10 +859,7 @@ def main(args: argparse.Namespace):
            if args.max_concurrency is not None
            else ""
        )
-        if args.ramp_up_strategy is not None:
-            file_name = f"{backend}-ramp-up-{args.ramp_up_strategy}-{args.ramp_up_start_rps}qps-{args.ramp_up_end_rps}qps{max_concurrency_str}-{base_model_id}-{current_dt}.json"  # noqa
-        else:
-            file_name = f"{backend}-{args.request_rate}qps{max_concurrency_str}-{base_model_id}-{current_dt}.json"  # noqa
+        file_name = f"{backend}-{args.request_rate}qps{max_concurrency_str}-{base_model_id}-{current_dt}.json"  # noqa
        if args.result_filename:
            file_name = args.result_filename
        if args.result_dir:
@ -928,7 +875,7 @@ def main(args: argparse.Namespace):
        save_to_pytorch_benchmark_format(args, result_json, file_name)


-def create_argument_parser():
+if __name__ == "__main__":
    parser = FlexibleArgumentParser(
        description="Benchmark the online serving throughput."
    )
@ -967,11 +914,6 @@ def create_argument_parser():
        help="Path to the sharegpt/sonnet dataset. "
        "Or the huggingface dataset ID if using HF dataset.",
    )
-    parser.add_argument(
-        "--no-stream",
-        action="store_true",
-        help="Do not load the dataset in streaming mode.",
-    )
    parser.add_argument(
        "--max-concurrency",
        type=int,
@ -1130,13 +1072,6 @@ def create_argument_parser():
        "goodput, refer to DistServe paper: https://arxiv.org/pdf/2401.09670 "
        "and the blog: https://hao-ai-lab.github.io/blogs/distserve",
    )
-    parser.add_argument(
-        "--request-id-prefix",
-        type=str,
-        required=False,
-        default="benchmark-serving",
-        help="Specify the prefix of request id.",
-    )

    # group for dataset specific arguments
    custom_group = parser.add_argument_group("custom dataset options")
@ -1290,35 +1225,6 @@ def create_argument_parser():
        "script chooses a LoRA module at random.",
    )

-    parser.add_argument(
-        "--ramp-up-strategy",
-        type=str,
-        default=None,
-        choices=["linear", "exponential"],
-        help="The ramp-up strategy. This would be used to "
-        "ramp up the request rate from initial RPS to final "
-        "RPS rate (specified by --ramp-up-start-rps and --ramp-up-end-rps). "
-        "over the duration of the benchmark.",
-    )
-    parser.add_argument(
-        "--ramp-up-start-rps",
-        type=int,
-        default=None,
-        help="The starting request rate for ramp-up (RPS). "
-        "Needs to be specified when --ramp-up-strategy is used.",
-    )
-    parser.add_argument(
-        "--ramp-up-end-rps",
-        type=int,
-        default=None,
-        help="The ending request rate for ramp-up (RPS). "
-        "Needs to be specified when --ramp-up-strategy is used.",
-    )
-
-    return parser
-
-
-if __name__ == "__main__":
-    parser = create_argument_parser()
    args = parser.parse_args()
+
    main(args)
--- a/benchmarks/benchmark_serving_structured_output.py
+++ b/benchmarks/benchmark_serving_structured_output.py
@ -4,7 +4,7 @@ r"""Benchmark online serving throughput with structured outputs.

 On the server side, run one of the following commands:
    (vLLM OpenAI API server)
-    vllm serve <your_model>
+    vllm serve <your_model> --disable-log-requests

 On the client side, run:
    python benchmarks/benchmark_serving_structured_output.py \
@ -12,6 +12,7 @@ On the client side, run:
        --model <your_model> \
        --dataset json \
        --structured-output-ratio 1.0 \
+        --structured-output-backend auto \
        --request-rate 10 \
        --num-prompts 1000

@ -538,6 +539,20 @@ async def benchmark(
        )
    outputs: list[RequestFuncOutput] = await asyncio.gather(*tasks)

+    if profile:
+        print("Stopping profiler...")
+        profile_input = RequestFuncInput(
+            model=model_id,
+            prompt=test_request.prompt,
+            api_url=base_url + "/stop_profile",
+            prompt_len=test_request.prompt_len,
+            output_len=test_request.expected_output_len,
+            extra_body={test_request.structure_type: test_request.schema},
+        )
+        profile_output = await request_func(request_func_input=profile_input)
+        if profile_output.success:
+            print("Profiler stopped")
+
    if pbar is not None:
        pbar.close()

@ -555,10 +570,6 @@ async def benchmark(

    print("{s:{c}^{n}}".format(s=" Serving Benchmark Result ", n=50, c="="))
    print("{:<40} {:<10}".format("Successful requests:", metrics.completed))
-    if max_concurrency is not None:
-        print("{:<40} {:<10}".format("Maximum request concurrency:", max_concurrency))
-    if request_rate != float("inf"):
-        print("{:<40} {:<10.2f}".format("Request rate configured (RPS):", request_rate))
    print("{:<40} {:<10.2f}".format("Benchmark duration (s):", benchmark_duration))
    print("{:<40} {:<10}".format("Total input tokens:", metrics.total_input))
    print("{:<40} {:<10}".format("Total generated tokens:", metrics.total_output))
@ -656,20 +667,6 @@ async def benchmark(

    print("=" * 50)

-    if profile:
-        print("Stopping profiler...")
-        profile_input = RequestFuncInput(
-            model=model_id,
-            prompt=test_request.prompt,
-            api_url=base_url + "/stop_profile",
-            prompt_len=test_request.prompt_len,
-            output_len=test_request.expected_output_len,
-            extra_body={test_request.structure_type: test_request.schema},
-        )
-        profile_output = await request_func(request_func_input=profile_input)
-        if profile_output.success:
-            print("Profiler stopped")
-
    return result, ret


@ -854,7 +851,7 @@ def main(args: argparse.Namespace):
            json.dump(results, outfile, indent=4)


-def create_argument_parser():
+if __name__ == "__main__":
    parser = FlexibleArgumentParser(
        description="Benchmark the online serving throughput."
    )
@ -1038,10 +1035,5 @@ def create_argument_parser():
        help="Ratio of Structured Outputs requests",
    )

-    return parser
-
-
-if __name__ == "__main__":
-    parser = create_argument_parser()
    args = parser.parse_args()
    main(args)
--- a/benchmarks/benchmark_throughput.py
+++ b/benchmarks/benchmark_throughput.py
@ -15,7 +15,6 @@ import torch
 import uvloop
 from tqdm import tqdm
 from transformers import AutoModelForCausalLM, AutoTokenizer, PreTrainedTokenizerBase
-from typing_extensions import deprecated

 from benchmark_dataset import (
    AIMODataset,
@ -96,8 +95,9 @@ def run_vllm(
        end = time.perf_counter()
    else:
        assert lora_requests is None, "BeamSearch API does not support LoRA"
+        prompts = [request.prompt for request in requests]
        # output_len should be the same for all requests.
-        output_len = requests[0].expected_output_len
+        output_len = requests[0][2]
        for request in requests:
            assert request.expected_output_len == output_len
        start = time.perf_counter()
@ -167,8 +167,7 @@ async def run_vllm_async(
    from vllm import SamplingParams

    async with build_async_engine_client_from_engine_args(
-        engine_args,
-        disable_frontend_multiprocessing=disable_frontend_multiprocessing,
+        engine_args, disable_frontend_multiprocessing
    ) as llm:
        model_config = await llm.get_model_config()
        assert all(
@ -357,7 +356,6 @@ def get_requests(args, tokenizer):
    elif args.dataset_name == "burstgpt":
        dataset_cls = BurstGPTDataset
    elif args.dataset_name == "hf":
-        common_kwargs["no_stream"] = args.no_stream
        if args.dataset_path in VisionArenaDataset.SUPPORTED_DATASET_PATHS:
            dataset_cls = VisionArenaDataset
            common_kwargs["dataset_subset"] = None
@ -382,10 +380,6 @@ def get_requests(args, tokenizer):
    return dataset_cls(**common_kwargs).sample(**sample_kwargs)


-@deprecated(
-    "benchmark_throughput.py is deprecated and will be removed in a "
-    "future version. Please use 'vllm bench throughput' instead.",
-)
 def main(args: argparse.Namespace):
    if args.seed is None:
        args.seed = 0
@ -596,12 +590,12 @@ def validate_args(args):
    # https://github.com/vllm-project/vllm/issues/16222
    if args.data_parallel_size > 1:
        raise ValueError(
-            "Data parallel is not supported in offline benchmark, "
-            "please use benchmark serving instead"
+            "Data parallel is not supported in offline benchmark, \
+            please use benchmark serving instead"
        )


-def create_argument_parser():
+if __name__ == "__main__":
    parser = FlexibleArgumentParser(description="Benchmark the throughput.")
    parser.add_argument(
        "--backend",
@ -616,11 +610,6 @@ def create_argument_parser():
        help="Name of the dataset to benchmark on.",
        default="sharegpt",
    )
-    parser.add_argument(
-        "--no-stream",
-        action="store_true",
-        help="Do not load the dataset in streaming mode.",
-    )
    parser.add_argument(
        "--dataset",
        type=str,
@ -728,12 +717,6 @@ def create_argument_parser():
    )

    parser = AsyncEngineArgs.add_cli_args(parser)
-
-    return parser
-
-
-if __name__ == "__main__":
-    parser = create_argument_parser()
    args = parser.parse_args()
    if args.tokenizer is None:
        args.tokenizer = args.model
--- a/benchmarks/benchmark_utils.py
+++ b/benchmarks/benchmark_utils.py
@ -1,12 +1,11 @@
 # SPDX-License-Identifier: Apache-2.0
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+
 import argparse
 import json
 import math
 import os
-import time
-from types import TracebackType
-from typing import Any, Optional, Union
+from typing import Any


 def convert_to_pytorch_benchmark_format(
@ -67,59 +66,4 @@ class InfEncoder(json.JSONEncoder):

 def write_to_json(filename: str, records: list) -> None:
    with open(filename, "w") as f:
-        json.dump(
-            records,
-            f,
-            cls=InfEncoder,
-            default=lambda o: f"<{type(o).__name__} object is not JSON serializable>",
-        )
-
-
-# Collect time and generate time metrics
-#
-# Example Usage:
-#   collector = TimeCollector(TimeCollector.US)
-#   for _ in range(total_iteration):
-#      with collector:
-#          ...
-#   collector.dump_avg_max()
-class TimeCollector:
-    NS: int = 1
-    US: int = NS * 1000
-    MS: int = US * 1000
-    S: int = MS * 1000
-
-    def __init__(self, scale: int) -> None:
-        self.cnt: int = 0
-        self._sum: int = 0
-        self._max: Optional[int] = None
-        self.scale = scale
-        self.start_time: int = time.monotonic_ns()
-
-    def collect(self, v: int) -> None:
-        self.cnt += 1
-        self._sum += v
-        if self._max is None:
-            self._max = v
-        else:
-            self._max = max(self._max, v)
-
-    def avg(self) -> Union[float, str]:
-        return self._sum * 1.0 / self.cnt / self.scale if self.cnt > 0 else "N/A"
-
-    def max(self) -> Union[float, str]:
-        return self._max / self.scale if self._max else "N/A"
-
-    def dump_avg_max(self) -> list[Union[float, str]]:
-        return [self.avg(), self.max()]
-
-    def __enter__(self) -> None:
-        self.start_time = time.monotonic_ns()
-
-    def __exit__(
-        self,
-        exc_type: Optional[type[BaseException]],
-        exc_value: Optional[BaseException],
-        exc_traceback: Optional[TracebackType],
-    ) -> None:
-        self.collect(time.monotonic_ns() - self.start_time)
+        json.dump(records, f, cls=InfEncoder)
--- a/benchmarks/cutlass_benchmarks/w8a8_benchmarks.py
+++ b/benchmarks/cutlass_benchmarks/w8a8_benchmarks.py
@ -19,7 +19,7 @@ from vllm import _custom_ops as ops
 from vllm.model_executor.layers.quantization.utils.fp8_utils import (
    w8a8_block_fp8_matmul,
 )
-from vllm.utils import FlexibleArgumentParser, cdiv
+from vllm.utils import FlexibleArgumentParser

 DEFAULT_MODELS = list(WEIGHT_SHAPES.keys())
 DEFAULT_BATCH_SIZES = [1, 16, 32, 64, 128, 256, 512]
@ -117,9 +117,14 @@ def bench_fp8(
    scale_a = torch.tensor(1.0, device="cuda", dtype=torch.float32)
    scale_b = torch.tensor(1.0, device="cuda", dtype=torch.float32)

-    block_scale_a = torch.rand((m, cdiv(k, 128)), device="cuda", dtype=torch.float32)
+    def ceil_div(x: int, y: int) -> int:
+        return (x + y - 1) // y
+
+    block_scale_a = torch.rand(
+        (m, ceil_div(k, 128)), device="cuda", dtype=torch.float32
+    )
    block_scale_b = torch.rand(
-        cdiv(k, 128), cdiv(n, 128), device="cuda", dtype=torch.float32
+        ceil_div(k, 128), ceil_div(n, 128), device="cuda", dtype=torch.float32
    )
    block_scale_a_M_major = block_scale_a.t().contiguous().t()
    block_scale_b_K_major = block_scale_b.t().contiguous().t()
--- a/benchmarks/disagg_benchmarks/disagg_overhead_benchmark.sh
+++ b/benchmarks/disagg_benchmarks/disagg_overhead_benchmark.sh
@ -3,7 +3,7 @@
 # benchmark the overhead of disaggregated prefill.
 # methodology:
 # - send all request to prefill vLLM instance. It will buffer KV cache.
-# - then send all request to decode instance.
+# - then send all request to decode instance. 
 # - The TTFT of decode instance is the overhead.

 set -ex
@ -12,8 +12,6 @@ kill_gpu_processes() {
  # kill all processes on GPU.
  pgrep pt_main_thread | xargs -r kill -9
  pgrep python3 | xargs -r kill -9
-  # vLLM now names the process with VLLM prefix after https://github.com/vllm-project/vllm/pull/21445
-  pgrep VLLM | xargs -r kill -9
  sleep 10

  # remove vllm config file
@ -63,7 +61,7 @@ benchmark() {
    --gpu-memory-utilization 0.6 \
    --kv-transfer-config \
    '{"kv_connector":"PyNcclConnector","kv_role":"kv_producer","kv_rank":0,"kv_parallel_size":2,"kv_buffer_size":5e9}' &
-
+    

  CUDA_VISIBLE_DEVICES=1 python3 \
    -m vllm.entrypoints.openai.api_server \
@ -78,38 +76,38 @@ benchmark() {
  wait_for_server 8200

  # let the prefill instance finish prefill
-  vllm bench serve \
-    --backend vllm \
-    --model $model \
-    --dataset-name $dataset_name \
-    --dataset-path $dataset_path \
-    --sonnet-input-len $input_len \
-    --sonnet-output-len "$output_len" \
-    --sonnet-prefix-len $prefix_len \
-    --num-prompts $num_prompts \
-    --port 8100 \
-    --save-result \
-    --result-dir $results_folder \
-    --result-filename disagg_prefill_tp1.json \
-    --request-rate "inf"
+  python3 ../benchmark_serving.py \
+          --backend vllm \
+          --model $model \
+          --dataset-name $dataset_name \
+          --dataset-path $dataset_path \
+          --sonnet-input-len $input_len \
+          --sonnet-output-len "$output_len" \
+          --sonnet-prefix-len $prefix_len \
+          --num-prompts $num_prompts \
+          --port 8100 \
+          --save-result \
+          --result-dir $results_folder \
+          --result-filename disagg_prefill_tp1.json \
+          --request-rate "inf"


  # send the request to decode.
  # The TTFT of this command will be the overhead of disagg prefill impl.
-  vllm bench serve \
-    --backend vllm \
-    --model $model \
-    --dataset-name $dataset_name \
-    --dataset-path $dataset_path \
-    --sonnet-input-len $input_len \
-    --sonnet-output-len "$output_len" \
-    --sonnet-prefix-len $prefix_len \
-    --num-prompts $num_prompts \
-    --port 8200 \
-    --save-result \
-    --result-dir $results_folder \
-    --result-filename disagg_prefill_tp1_overhead.json \
-    --request-rate "$qps"
+  python3 ../benchmark_serving.py \
+          --backend vllm \
+          --model $model \
+          --dataset-name $dataset_name \
+          --dataset-path $dataset_path \
+          --sonnet-input-len $input_len \
+          --sonnet-output-len "$output_len" \
+          --sonnet-prefix-len $prefix_len \
+          --num-prompts $num_prompts \
+          --port 8200 \
+          --save-result \
+          --result-dir $results_folder \
+          --result-filename disagg_prefill_tp1_overhead.json \
+          --request-rate "$qps"
  kill_gpu_processes

 }
--- a/benchmarks/disagg_benchmarks/disagg_performance_benchmark.sh
+++ b/benchmarks/disagg_benchmarks/disagg_performance_benchmark.sh
@ -18,8 +18,6 @@ kill_gpu_processes() {
  # kill all processes on GPU.
  pgrep pt_main_thread | xargs -r kill -9
  pgrep python3 | xargs -r kill -9
-  # vLLM now names the process with VLLM prefix after https://github.com/vllm-project/vllm/pull/21445
-  pgrep VLLM | xargs -r kill -9
  for port in 8000 8100 8200; do lsof -t -i:$port | xargs -r kill -9; done
  sleep 1
 }
@ -60,7 +58,7 @@ launch_chunked_prefill() {


 launch_disagg_prefill() {
-  model="meta-llama/Meta-Llama-3.1-8B-Instruct"
+  model="meta-llama/Meta-Llama-3.1-8B-Instruct" 
  # disagg prefill
  CUDA_VISIBLE_DEVICES=0 python3 \
    -m vllm.entrypoints.openai.api_server \
@ -99,20 +97,20 @@ benchmark() {
  output_len=$2
  tag=$3

-  vllm bench serve \
-    --backend vllm \
-    --model $model \
-    --dataset-name $dataset_name \
-    --dataset-path $dataset_path \
-    --sonnet-input-len $input_len \
-    --sonnet-output-len "$output_len" \
-    --sonnet-prefix-len $prefix_len \
-    --num-prompts $num_prompts \
-    --port 8000 \
-    --save-result \
-    --result-dir $results_folder \
-    --result-filename "$tag"-qps-"$qps".json \
-    --request-rate "$qps"
+  python3 ../benchmark_serving.py \
+          --backend vllm \
+          --model $model \
+          --dataset-name $dataset_name \
+          --dataset-path $dataset_path \
+          --sonnet-input-len $input_len \
+          --sonnet-output-len "$output_len" \
+          --sonnet-prefix-len $prefix_len \
+          --num-prompts $num_prompts \
+          --port 8000 \
+          --save-result \
+          --result-dir $results_folder \
+          --result-filename "$tag"-qps-"$qps".json \
+          --request-rate "$qps"

  sleep 2
 }
--- a/benchmarks/disagg_benchmarks/disagg_prefill_proxy_server.py
+++ b/benchmarks/disagg_benchmarks/disagg_prefill_proxy_server.py
@ -1,199 +1,63 @@
 # SPDX-License-Identifier: Apache-2.0
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project

-import argparse
-import asyncio
-import logging
 import os

 import aiohttp
-from quart import Quart, Response, make_response, request
-from rate_limiter import RateLimiter
-from request_queue import RequestQueue
+from quart import Quart, make_response, request

-# Configure logging
-logging.basicConfig(level=logging.INFO)
-logger = logging.getLogger(__name__)
+AIOHTTP_TIMEOUT = aiohttp.ClientTimeout(total=6 * 60 * 60)
+
+app = Quart(__name__)


-def parse_args():
-    """parse command line arguments"""
-    parser = argparse.ArgumentParser(description="vLLM P/D disaggregation proxy server")
-
-    # Add args
-    parser.add_argument(
-        "--timeout",
-        type=float,
-        default=300,
-        help="Timeout for backend service requests in seconds (default: 300)",
-    )
-    parser.add_argument(
-        "--max-concurrent",
-        type=int,
-        default=100,
-        help="Maximum concurrent requests to backend services (default: 100)",
-    )
-    parser.add_argument(
-        "--queue-size",
-        type=int,
-        default=500,
-        help="Maximum number of requests in the queue (default: 500)",
-    )
-    parser.add_argument(
-        "--rate-limit",
-        type=int,
-        default=40,
-        help="Maximum requests per second (default: 40)",
-    )
-    parser.add_argument(
-        "--port",
-        type=int,
-        default=8000,
-        help="Port to run the server on (default: 8000)",
-    )
-    parser.add_argument(
-        "--prefill-url",
-        type=str,
-        default="http://localhost:8100/v1/completions",
-        help="Prefill service endpoint URL",
-    )
-    parser.add_argument(
-        "--decode-url",
-        type=str,
-        default="http://localhost:8200/v1/completions",
-        help="Decode service endpoint URL",
-    )
-
-    return parser.parse_args()
-
-
-def main():
-    """parse command line arguments"""
-    args = parse_args()
-
-    # Initialize configuration using command line parameters
-    AIOHTTP_TIMEOUT = aiohttp.ClientTimeout(total=args.timeout)
-    MAX_CONCURRENT_REQUESTS = args.max_concurrent
-    REQUEST_QUEUE_SIZE = args.queue_size
-    RATE_LIMIT = args.rate_limit
-    PREFILL_SERVICE_URL = args.prefill_url
-    DECODE_SERVICE_URL = args.decode_url
-    PORT = args.port
-
-    app = Quart(__name__)
-
-    # Initialize the rate limiter and request queue
-    rate_limiter = RateLimiter(RATE_LIMIT)
-    request_queue = RequestQueue(MAX_CONCURRENT_REQUESTS, REQUEST_QUEUE_SIZE)
-
-    # Attach the configuration object to the application instance
-    app.config.update(
-        {
-            "AIOHTTP_TIMEOUT": AIOHTTP_TIMEOUT,
-            "rate_limiter": rate_limiter,
-            "request_queue": request_queue,
-            "PREFILL_SERVICE_URL": PREFILL_SERVICE_URL,
-            "DECODE_SERVICE_URL": DECODE_SERVICE_URL,
-        }
-    )
-
-    # Start queue processing on app startup
-    @app.before_serving
-    async def startup():
-        """Start request processing task when app starts serving"""
-        asyncio.create_task(request_queue.process())
-
-    async def forward_request(url, data):
-        """Forward request to backend service with rate limiting and error handling"""
+async def forward_request(url, data):
+    async with aiohttp.ClientSession(timeout=AIOHTTP_TIMEOUT) as session:
        headers = {"Authorization": f"Bearer {os.environ.get('OPENAI_API_KEY')}"}
+        async with session.post(url=url, json=data, headers=headers) as response:
+            if response.status == 200:
+                # if response.headers.get('Transfer-Encoding') == 'chunked':
+                if True:
+                    async for chunk_bytes in response.content.iter_chunked(1024):
+                        yield chunk_bytes
+                else:
+                    content = await response.read()
+                    yield content

-        # Use rate limiter as context manager
-        async with (
-            rate_limiter,
-            aiohttp.ClientSession(timeout=AIOHTTP_TIMEOUT) as session,
+
+@app.route("/v1/completions", methods=["POST"])
+async def handle_request():
+    try:
+        original_request_data = await request.get_json()
+
+        prefill_request = original_request_data.copy()
+        # change max_tokens = 1 to let it only do prefill
+        prefill_request["max_tokens"] = 1
+
+        # finish prefill
+        async for _ in forward_request(
+            "http://localhost:8100/v1/completions", prefill_request
        ):
-            try:
-                async with session.post(
-                    url=url, json=data, headers=headers
-                ) as response:
-                    if response.status == 200:
-                        # Stream response chunks
-                        async for chunk_bytes in response.content.iter_chunked(1024):
-                            yield chunk_bytes
-                    else:
-                        # Handle backend service errors
-                        error_text = await response.text()
-                        logger.error(
-                            "Backend service error: %s - %s",
-                            response.status,
-                            error_text,
-                        )
-                        yield b'{"error": "Backend service error"}'
-            except aiohttp.ClientError as e:
-                # Handle connection errors
-                logger.error("Connection error to %s: %s", url, str(e))
-                yield b'{"error": "Service unavailable"}'
-            except asyncio.TimeoutError:
-                # Handle timeout errors
-                logger.error("Timeout connecting to %s", url)
-                yield b'{"error": "Service timeout"}'
+            continue

-    async def process_request():
-        """Process a single request through prefill and decode stages"""
-        try:
-            original_request_data = await request.get_json()
+        # return decode
+        generator = forward_request(
+            "http://localhost:8200/v1/completions", original_request_data
+        )
+        response = await make_response(generator)
+        response.timeout = None

-            # Create prefill request (max_tokens=1)
-            prefill_request = original_request_data.copy()
-            prefill_request["max_tokens"] = 1
+        return response

-            # Execute prefill stage
-            async for _ in forward_request(PREFILL_SERVICE_URL, prefill_request):
-                continue
+    except Exception as e:
+        import sys
+        import traceback

-            # Execute decode stage and stream response
-            generator = forward_request(DECODE_SERVICE_URL, original_request_data)
-            response = await make_response(generator)
-            response.timeout = None  # Disable timeout for streaming response
-            return response
-
-        except Exception:
-            logger.exception("Error processing request")
-            return Response(
-                response=b'{"error": "Internal server error"}',
-                status=500,
-                content_type="application/json",
-            )
-
-    @app.route("/v1/completions", methods=["POST"])
-    async def handle_request():
-        """Handle incoming API requests with concurrency and rate limiting"""
-        # Create task for request processing
-        task = asyncio.create_task(process_request())
-
-        # Enqueue request or reject if queue is full
-        if not await request_queue.enqueue(task):
-            return Response(
-                response=b'{"error": "Server busy, try again later"}',
-                status=503,
-                content_type="application/json",
-            )
-
-        try:
-            # Return the response from the processing task
-            return await task
-        except asyncio.CancelledError:
-            # Handle task cancellation (timeout or queue full)
-            logger.warning("Request cancelled due to timeout or queue full")
-            return Response(
-                response=b'{"error": "Request cancelled"}',
-                status=503,
-                content_type="application/json",
-            )
-
-    # Start the Quart server with host can be set to 0.0.0.0
-    app.run(port=PORT)
+        exc_info = sys.exc_info()
+        print("Error occurred in disagg prefill proxy server")
+        print(e)
+        print("".join(traceback.format_exception(*exc_info)))


 if __name__ == "__main__":
-    main()
+    app.run(port=8000)
--- a/benchmarks/disagg_benchmarks/rate_limiter.py
+++ b/benchmarks/disagg_benchmarks/rate_limiter.py
@ -1,45 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-
-import asyncio
-import time
-
-
-class RateLimiter:
-    """Token bucket rate limiter implementation"""
-
-    def __init__(self, rate_limit):
-        self.rate_limit = rate_limit  # Requests per second
-        self.num_available_tokens = rate_limit  # Available tokens
-        self.last_refill = time.monotonic()  # Last token refill time
-        self.lock = asyncio.Lock()  # Synchronization lock
-
-    async def acquire(self):
-        """Acquire a token from the rate limiter"""
-        while True:
-            async with self.lock:
-                current_time = time.monotonic()
-                elapsed = current_time - self.last_refill
-
-                # Refill num_available_tokens if more than 1 second has passed
-                if elapsed > 1.0:
-                    self.num_available_tokens = self.rate_limit
-                    self.last_refill = current_time
-
-                # Check if num_available_tokens are available
-                if self.num_available_tokens > 0:
-                    self.num_available_tokens -= 1
-                    return True
-
-                # Calculate wait time if no num_available_tokens available
-                wait_time = 1.0 - elapsed
-            await asyncio.sleep(wait_time)
-
-    async def __aenter__(self):
-        """Enter async context manager - acquire token"""
-        await self.acquire()
-        return self
-
-    async def __aexit__(self, exc_type, exc_value, traceback):
-        """Exit async context manager - no cleanup needed"""
-        pass
--- a/benchmarks/disagg_benchmarks/request_queue.py
+++ b/benchmarks/disagg_benchmarks/request_queue.py
@ -1,39 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-
-import asyncio
-from collections import deque
-
-
-class RequestQueue:
-    """Request queue manager with concurrency control"""
-
-    def __init__(self, max_concurrent, max_queue_size):
-        # Maximum concurrent requests
-        self.max_concurrent = max_concurrent
-        self.max_queue_size = max_queue_size  # Maximum queue size
-        # Concurrency control
-        self.semaphore = asyncio.Semaphore(max_concurrent)
-        self.queue = deque()  # Request queue
-        self.queue_size = 0  # Current queue size
-        self.lock = asyncio.Lock()  # Sync queue Lock
-
-    async def enqueue(self, task):
-        """Add a request task to the queue"""
-        async with self.lock:
-            if self.queue_size >= self.max_queue_size:
-                return False
-
-            self.queue.append(task)
-            self.queue_size += 1
-            return True
-
-    async def process(self):
-        """Process queued requests using semaphore for concurrency control"""
-        while True:
-            if self.queue:
-                async with self.semaphore, self.lock:
-                    task = self.queue.popleft()
-                    self.queue_size -= 1
-                    await task
-            await asyncio.sleep(0.01)  # Yield control to event loop
--- a/benchmarks/kernels/bench_fp8_gemm.py
+++ b/benchmarks/kernels/bench_fp8_gemm.py
@ -5,85 +5,11 @@ import copy
 import itertools

 import torch
+import triton
 from weight_shapes import WEIGHT_SHAPES

 from vllm._custom_ops import cutlass_scaled_mm as vllm_scaled_mm
 from vllm._custom_ops import scaled_fp8_quant as vllm_scaled_fp8_quant
-from vllm.triton_utils import triton
-
-PROVIDER_CFGS = {
-    "torch-bf16": dict(enabled=True),
-    "fp8-tensor-w-token-a": dict(
-        w="tensor", a="token", no_a_quant=False, enabled=False
-    ),
-    "fp8-tensor-w-tensor-a": dict(
-        w="tensor", a="tensor", no_a_quant=False, enabled=True
-    ),
-    "fp8-channel-w-token-a": dict(
-        w="channel", a="token", no_a_quant=False, enabled=True
-    ),
-    "fp8-channel-w-tensor-a": dict(
-        w="channel", a="tensor", no_a_quant=False, enabled=False
-    ),
-    "fp8-tensor-w-token-a-noquant": dict(
-        w="tensor", a="token", no_a_quant=True, enabled=False
-    ),
-    "fp8-tensor-w-tensor-a-noquant": dict(
-        w="tensor", a="tensor", no_a_quant=True, enabled=True
-    ),
-    "fp8-channel-w-token-a-noquant": dict(
-        w="channel", a="token", no_a_quant=True, enabled=True
-    ),
-    "fp8-channel-w-tensor-a-noquant": dict(
-        w="channel", a="tensor", no_a_quant=True, enabled=False
-    ),
-}
-
-_enabled = [k for k, v in PROVIDER_CFGS.items() if v["enabled"]]
-
-
-def _quant_weight_fp8(b: torch.Tensor, w_type: str, device: str):
-    if w_type == "tensor":
-        scale_b = torch.ones(1, device=device, dtype=torch.float32)
-        b_fp8, scale_b_fp8 = vllm_scaled_fp8_quant(b, scale_b)
-    else:
-        b_fp8, scale_b_fp8 = vllm_scaled_fp8_quant(b, use_per_token_if_dynamic=True)
-    return b_fp8.t(), scale_b_fp8
-
-
-def build_fp8_runner(cfg, a, b, dtype, device):
-    b_fp8, scale_b_fp8 = _quant_weight_fp8(b, cfg["w"], device)
-
-    scale_a_const = (
-        torch.ones(1, device=device, dtype=torch.float32)
-        if cfg["a"] == "tensor"
-        else None
-    )
-
-    if cfg["no_a_quant"]:
-        if cfg["a"] == "tensor":
-            a_fp8, scale_a_fp8 = vllm_scaled_fp8_quant(a, scale_a_const)
-        else:
-            a_fp8, scale_a_fp8 = vllm_scaled_fp8_quant(a, use_per_token_if_dynamic=True)
-
-        def run():
-            return vllm_scaled_mm(a_fp8, b_fp8, scale_a_fp8, scale_b_fp8, dtype)
-
-        return run
-
-    if cfg["a"] == "tensor":
-
-        def run():
-            a_fp8, scale_a_fp8 = vllm_scaled_fp8_quant(a, scale_a_const)
-            return vllm_scaled_mm(a_fp8, b_fp8, scale_a_fp8, scale_b_fp8, dtype)
-
-    else:
-
-        def run():
-            a_fp8, scale_a_fp8 = vllm_scaled_fp8_quant(a, use_per_token_if_dynamic=True)
-            return vllm_scaled_mm(a_fp8, b_fp8, scale_a_fp8, scale_b_fp8, dtype)
-
-    return run


@triton.testing.perf_report(
@ -92,8 +18,28 @@ def build_fp8_runner(cfg, a, b, dtype, device):
        x_vals=[1, 16, 64, 128, 256, 512, 1024, 2048, 4096, 8192, 16384],
        x_log=False,
        line_arg="provider",
-        line_vals=_enabled,
-        line_names=_enabled,
+        line_vals=[
+            "torch-bf16",
+            # "fp8-tensor-w-token-a",
+            "fp8-tensor-w-tensor-a",
+            "fp8-channel-w-token-a",
+            # "fp8-channel-w-tensor-a",
+            # "fp8-tensor-w-token-a-noquant",
+            "fp8-tensor-w-tensor-a-noquant",
+            "fp8-channel-w-token-a-noquant",
+            # "fp8-channel-w-tensor-a-noquant",
+        ],
+        line_names=[
+            "torch-bf16",
+            # "fp8-tensor-w-token-a",
+            "fp8-tensor-w-tensor-a",
+            "fp8-channel-w-token-a",
+            # "fp8-channel-w-tensor-a",
+            # "fp8-tensor-w-token-a-noquant",
+            "fp8-tensor-w-tensor-a-noquant",
+            "fp8-channel-w-token-a-noquant",
+            # "fp8-channel-w-tensor-a-noquant",
+        ],
        ylabel="TFLOP/s (larger is better)",
        plot_name="BF16 vs FP8 GEMMs",
        args={},
@ -104,34 +50,144 @@ def benchmark(batch_size, provider, N, K):
    device = "cuda"
    dtype = torch.bfloat16

+    # Create input tensors
    a = torch.randn((M, K), device=device, dtype=dtype)
    b = torch.randn((N, K), device=device, dtype=dtype)

    quantiles = [0.5, 0.2, 0.8]

-    if provider == "torch-bf16":
+    if "torch-bf16" in provider:
        ms, min_ms, max_ms = triton.testing.do_bench_cudagraph(
            lambda: torch.nn.functional.linear(a, b), quantiles=quantiles
        )
-    else:
-        cfg = PROVIDER_CFGS[provider]
-        run_quant = build_fp8_runner(cfg, a, b, dtype, device)
+
+    elif "fp8" in provider:
+        # Weights are always quantized ahead of time
+        if "noquant" in provider:
+            # For no quantization, we just measure the GEMM
+            if "tensor-w-token-a" in provider:
+                # Dynamic per-token quant for A, per-tensor quant for B
+                b_fp8, scale_b_fp8 = vllm_scaled_fp8_quant(b)
+                assert scale_b_fp8.numel() == 1
+                a_fp8, scale_a_fp8 = vllm_scaled_fp8_quant(
+                    a, use_per_token_if_dynamic=True
+                )
+
+                def run_quant():
+                    return vllm_scaled_mm(a_fp8, b_fp8, scale_a_fp8, scale_b_fp8, dtype)
+
+            elif "tensor-w-tensor-a" in provider:
+                # Static per-tensor quantization with fixed scales
+                # for both A and B
+                scale_a = torch.tensor([1.0], device=device, dtype=torch.float32)
+                scale_b = torch.tensor([1.0], device=device, dtype=torch.float32)
+                b_fp8, scale_b_fp8 = vllm_scaled_fp8_quant(b, scale_b)
+                assert scale_b_fp8.numel() == 1
+                a_fp8, scale_a_fp8 = vllm_scaled_fp8_quant(a, scale_a)
+
+                def run_quant():
+                    return vllm_scaled_mm(a_fp8, b_fp8, scale_a_fp8, scale_b_fp8, dtype)
+
+            elif "channel-w-token-a" in provider:
+                # Static per-channel quantization for weights, per-token
+                # quant for A
+                scale_b = torch.tensor((N,), device=device, dtype=torch.float32)
+                b_fp8, scale_b_fp8 = vllm_scaled_fp8_quant(b, scale_b)
+                scale_b_fp8 = scale_b_fp8.expand(N).contiguous()
+                assert scale_b_fp8.numel() == N
+                a_fp8, scale_a_fp8 = vllm_scaled_fp8_quant(
+                    a, use_per_token_if_dynamic=True
+                )
+
+                def run_quant():
+                    return vllm_scaled_mm(a_fp8, b_fp8, scale_a_fp8, scale_b_fp8, dtype)
+
+            elif "channel-w-tensor-a" in provider:
+                # Static per-channel quantization for weights, per-tensor
+                # quant for A
+                scale_a = torch.tensor([1.0], device=device, dtype=torch.float32)
+                scale_b = torch.tensor((N,), device=device, dtype=torch.float32)
+                b_fp8, scale_b_fp8 = vllm_scaled_fp8_quant(b, scale_b)
+                scale_b_fp8 = scale_b_fp8.expand(N).contiguous()
+                assert scale_b_fp8.numel() == N
+                a_fp8, scale_a_fp8 = vllm_scaled_fp8_quant(a, scale_a)
+
+                def run_quant():
+                    return vllm_scaled_mm(a_fp8, b_fp8, scale_a_fp8, scale_b_fp8, dtype)
+
+        else:
+            # In these cases, we quantize the activations during the GEMM call
+            if "tensor-w-token-a" in provider:
+                # Dynamic per-token quant for A, per-tensor quant for B
+                b_fp8, scale_b_fp8 = vllm_scaled_fp8_quant(b)
+                assert scale_b_fp8.numel() == 1
+
+                def run_quant():
+                    a_fp8, scale_a_fp8 = vllm_scaled_fp8_quant(
+                        a, use_per_token_if_dynamic=True
+                    )
+                    return vllm_scaled_mm(a_fp8, b_fp8, scale_a_fp8, scale_b_fp8, dtype)
+
+            elif "tensor-w-tensor-a" in provider:
+                # Static per-tensor quantization with fixed scales
+                # for both A and B
+                scale_a = torch.tensor([1.0], device=device, dtype=torch.float32)
+                scale_b = torch.tensor([1.0], device=device, dtype=torch.float32)
+                b_fp8, scale_b_fp8 = vllm_scaled_fp8_quant(b, scale_b)
+                assert scale_b_fp8.numel() == 1
+
+                def run_quant():
+                    a_fp8, scale_a_fp8 = vllm_scaled_fp8_quant(a, scale_a)
+                    return vllm_scaled_mm(a_fp8, b_fp8, scale_a_fp8, scale_b_fp8, dtype)
+
+            elif "channel-w-token-a" in provider:
+                # Static per-channel quantization for weights, per-token
+                # quant for A
+                scale_b = torch.tensor((N,), device=device, dtype=torch.float32)
+                b_fp8, scale_b_fp8 = vllm_scaled_fp8_quant(b, scale_b)
+                scale_b_fp8 = scale_b_fp8.expand(N).contiguous()
+                assert scale_b_fp8.numel() == N
+
+                def run_quant():
+                    a_fp8, scale_a_fp8 = vllm_scaled_fp8_quant(
+                        a, use_per_token_if_dynamic=True
+                    )
+                    return vllm_scaled_mm(a_fp8, b_fp8, scale_a_fp8, scale_b_fp8, dtype)
+
+            elif "channel-w-tensor-a" in provider:
+                # Static per-channel quantization for weights, per-tensor
+                # quant for A
+                scale_a = torch.tensor([1.0], device=device, dtype=torch.float32)
+                scale_b = torch.tensor((N,), device=device, dtype=torch.float32)
+                b_fp8, scale_b_fp8 = vllm_scaled_fp8_quant(b, scale_b)
+                scale_b_fp8 = scale_b_fp8.expand(N).contiguous()
+                assert scale_b_fp8.numel() == N
+
+                def run_quant():
+                    a_fp8, scale_a_fp8 = vllm_scaled_fp8_quant(a, scale_a)
+                    return vllm_scaled_mm(a_fp8, b_fp8, scale_a_fp8, scale_b_fp8, dtype)
+
+        b_fp8 = b_fp8.t()
+
        ms, min_ms, max_ms = triton.testing.do_bench_cudagraph(
            lambda: run_quant(), quantiles=quantiles
        )

-    to_tflops = lambda t_ms: (2 * M * N * K) * 1e-12 / (t_ms * 1e-3)
-    return to_tflops(ms), to_tflops(max_ms), to_tflops(min_ms)
+    # Calculate TFLOP/s, two flops per multiply-add
+    tflops = lambda ms: (2 * M * N * K) * 1e-12 / (ms * 1e-3)
+    return tflops(ms), tflops(max_ms), tflops(min_ms)


 def prepare_shapes(args):
-    out = []
-    for model, tp_size in itertools.product(args.models, args.tp_sizes):
-        for KN, tp_dim in copy.deepcopy(WEIGHT_SHAPES[model]):
-            KN[tp_dim] //= tp_size
+    KN_model_names = []
+    models_tps = list(itertools.product(args.models, args.tp_sizes))
+    for model, tp_size in models_tps:
+        assert model in WEIGHT_SHAPES
+        for KN, tp_split_dim in copy.deepcopy(WEIGHT_SHAPES[model]):
+            KN[tp_split_dim] = KN[tp_split_dim] // tp_size
            KN.append(model)
-            out.append(KN)
-    return out
+            KN_model_names.append(KN)
+    return KN_model_names


 if __name__ == "__main__":
@ -141,13 +197,21 @@ if __name__ == "__main__":
        nargs="+",
        type=str,
        default=["meta-llama/Llama-3.1-8B-Instruct"],
-        choices=list(WEIGHT_SHAPES.keys()),
+        choices=[*WEIGHT_SHAPES.keys()],
+        help="List of models to benchmark",
+    )
+    parser.add_argument(
+        "--tp-sizes",
+        nargs="+",
+        type=int,
+        default=[1],
+        help="List of tensor parallel sizes",
    )
-    parser.add_argument("--tp-sizes", nargs="+", type=int, default=[1])
    args = parser.parse_args()

-    for K, N, model in prepare_shapes(args):
-        print(f"{model}, N={N} K={K}, BF16 vs FP8 GEMMs TFLOP/s:")
+    KN_model_names = prepare_shapes(args)
+    for K, N, model_name in KN_model_names:
+        print(f"{model_name}, N={N} K={K}, BF16 vs FP8 GEMMs TFLOP/s:")
        benchmark.run(
            print_data=True,
            show_plots=True,
--- a/benchmarks/kernels/bench_int8_gemm.py
+++ b/benchmarks/kernels/bench_int8_gemm.py
@ -1,169 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-import argparse
-import copy
-import itertools
-
-import torch
-from weight_shapes import WEIGHT_SHAPES
-
-from vllm._custom_ops import cutlass_scaled_mm as vllm_scaled_mm
-from vllm._custom_ops import scaled_int8_quant as vllm_scaled_int8_quant
-from vllm.triton_utils import triton
-
-PROVIDER_CFGS = {
-    "torch-bf16": dict(enabled=True),
-    "int8-tensor-w-token-a": dict(
-        w="tensor", a="token", no_a_quant=False, enabled=False
-    ),
-    "int8-tensor-w-tensor-a": dict(
-        w="tensor", a="tensor", no_a_quant=False, enabled=True
-    ),
-    "int8-channel-w-token-a": dict(
-        w="channel", a="token", no_a_quant=False, enabled=True
-    ),
-    "int8-channel-w-tensor-a": dict(
-        w="channel", a="tensor", no_a_quant=False, enabled=False
-    ),
-    "int8-tensor-w-token-a-noquant": dict(
-        w="tensor", a="token", no_a_quant=True, enabled=False
-    ),
-    "int8-tensor-w-tensor-a-noquant": dict(
-        w="tensor", a="tensor", no_a_quant=True, enabled=True
-    ),
-    "int8-channel-w-token-a-noquant": dict(
-        w="channel", a="token", no_a_quant=True, enabled=True
-    ),
-    "int8-channel-w-tensor-a-noquant": dict(
-        w="channel", a="tensor", no_a_quant=True, enabled=False
-    ),
-}
-
-
-def _quant_weight(b, w_type, device):
-    if w_type == "tensor":
-        scale_b = torch.ones(1, device=device, dtype=torch.float32)
-        b_int8, scale_b_int8, _ = vllm_scaled_int8_quant(b, scale_b)
-        assert scale_b_int8.numel() == 1
-    else:  # channel
-        b_int8, scale_b_int8, _ = vllm_scaled_int8_quant(b)
-        assert scale_b_int8.numel() == b.shape[0]
-    return b_int8.t(), scale_b_int8
-
-
-def build_int8_runner(cfg, a, b, dtype, device):
-    # quant before running the kernel
-    b_int8, scale_b_int8 = _quant_weight(b, cfg["w"], device)
-
-    scale_a_const = None
-    if cfg["a"] == "tensor":
-        scale_a_const = torch.ones(1, device=device, dtype=torch.float32)
-
-    # no quant, create activation ahead
-    if cfg["no_a_quant"]:
-        if cfg["a"] == "tensor":
-            a_int8, scale_a_int8, _ = vllm_scaled_int8_quant(a, scale_a_const)
-        else:  # token
-            a_int8, scale_a_int8, _ = vllm_scaled_int8_quant(a)
-
-        def run_quant():
-            return vllm_scaled_mm(a_int8, b_int8, scale_a_int8, scale_b_int8, dtype)
-
-        return run_quant
-
-    # dynamic quant, create activation inside
-    if cfg["a"] == "tensor":
-
-        def run_quant():
-            a_int8, scale_a_int8, _ = vllm_scaled_int8_quant(a, scale_a_const)
-            return vllm_scaled_mm(a_int8, b_int8, scale_a_int8, scale_b_int8, dtype)
-
-    else:  # token
-
-        def run_quant():
-            a_int8, scale_a_int8, _ = vllm_scaled_int8_quant(a)
-            return vllm_scaled_mm(a_int8, b_int8, scale_a_int8, scale_b_int8, dtype)
-
-    return run_quant
-
-
-_enabled = [k for k, v in PROVIDER_CFGS.items() if v.get("enabled")]
-
-
-@triton.testing.perf_report(
-    triton.testing.Benchmark(
-        x_names=["batch_size"],
-        x_vals=[1, 16, 64, 128, 256, 512, 1024, 2048, 4096, 8192, 16384],
-        x_log=False,
-        line_arg="provider",
-        line_vals=_enabled,
-        line_names=[k for k in _enabled],
-        ylabel="TFLOP/s (larger is better)",
-        plot_name="BF16 vs INT8 GEMMs",
-        args={},
-    )
-)
-def benchmark(batch_size, provider, N, K):
-    M = batch_size
-    device = "cuda"
-    dtype = torch.bfloat16
-    a = torch.randn((M, K), device=device, dtype=dtype)
-    b = torch.randn((N, K), device=device, dtype=dtype)
-
-    quantiles = [0.5, 0.2, 0.8]
-
-    if provider == "torch-bf16":
-        ms, min_ms, max_ms = triton.testing.do_bench_cudagraph(
-            lambda: torch.nn.functional.linear(a, b), quantiles=quantiles
-        )
-    else:
-        cfg = PROVIDER_CFGS[provider]
-        run_quant = build_int8_runner(cfg, a, b, dtype, device)
-        ms, min_ms, max_ms = triton.testing.do_bench_cudagraph(
-            lambda: run_quant(), quantiles=quantiles
-        )
-
-    to_tflops = lambda t_ms: (2 * M * N * K) * 1e-12 / (t_ms * 1e-3)
-    return to_tflops(ms), to_tflops(max_ms), to_tflops(min_ms)
-
-
-def prepare_shapes(args):
-    KN_model_names = []
-    for model, tp_size in itertools.product(args.models, args.tp_sizes):
-        for KN, tp_dim in copy.deepcopy(WEIGHT_SHAPES[model]):
-            KN[tp_dim] //= tp_size
-            KN.append(model)
-            KN_model_names.append(KN)
-    return KN_model_names
-
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser()
-    parser.add_argument(
-        "--models",
-        nargs="+",
-        type=str,
-        default=["meta-llama/Llama-3.1-8B-Instruct"],
-        choices=list(WEIGHT_SHAPES.keys()),
-        help="List of models to benchmark",
-    )
-    parser.add_argument(
-        "--tp-sizes",
-        nargs="+",
-        type=int,
-        default=[1],
-        help="List of tensor parallel sizes",
-    )
-    args = parser.parse_args()
-
-    for K, N, model in prepare_shapes(args):
-        print(f"{model}, N={N} K={K}, BF16 vs INT8 GEMMs TFLOP/s:")
-        benchmark.run(
-            print_data=True,
-            show_plots=True,
-            save_path=f"bench_int8_res_n{N}_k{K}",
-            N=N,
-            K=K,
-        )
-
-    print("Benchmark finished!")
--- a/benchmarks/kernels/bench_nvfp4_gemm.py
+++ b/benchmarks/kernels/bench_nvfp4_gemm.py
@ -1,141 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-import argparse
-import copy
-import itertools
-
-import torch
-from weight_shapes import WEIGHT_SHAPES
-
-from vllm import _custom_ops as ops
-from vllm.platforms import current_platform
-from vllm.scalar_type import scalar_types
-from vllm.triton_utils import triton
-
-if not current_platform.has_device_capability(100):
-    raise RuntimeError("NVFP4 requires compute capability of 10.0 (Blackwell)")
-
-
-FLOAT4_E2M1_MAX = scalar_types.float4_e2m1f.max()
-FLOAT8_E4M3_MAX = torch.finfo(torch.float8_e4m3fn).max
-
-PROVIDER_CFGS = {
-    "torch-bf16": dict(enabled=True),
-    "nvfp4": dict(no_a_quant=False, enabled=True),
-    "nvfp4-noquant": dict(no_a_quant=True, enabled=True),
-}
-
-_enabled = [k for k, v in PROVIDER_CFGS.items() if v["enabled"]]
-
-
-def _quant_weight_nvfp4(b: torch.Tensor, device: str):
-    # Compute global scale for weight
-    b_amax = torch.abs(b).max().to(torch.float32)
-    b_global_scale = FLOAT8_E4M3_MAX * FLOAT4_E2M1_MAX / b_amax
-    b_fp4, scale_b_fp4 = ops.scaled_fp4_quant(b, b_global_scale)
-    return b_fp4, scale_b_fp4, b_global_scale
-
-
-def build_nvfp4_runner(cfg, a, b, dtype, device):
-    b_fp4, scale_b_fp4, b_global_scale = _quant_weight_nvfp4(b, device)
-
-    # Compute global scale for activation
-    # NOTE: This is generally provided ahead-of-time by the model checkpoint.
-    a_amax = torch.abs(a).max().to(torch.float32)
-    a_global_scale = FLOAT8_E4M3_MAX * FLOAT4_E2M1_MAX / a_amax
-
-    # Alpha for the GEMM operation
-    alpha = 1.0 / (a_global_scale * b_global_scale)
-
-    if cfg["no_a_quant"]:
-        # Pre-quantize activation
-        a_fp4, scale_a_fp4 = ops.scaled_fp4_quant(a, a_global_scale)
-
-        def run():
-            return ops.cutlass_scaled_fp4_mm(
-                a_fp4, b_fp4, scale_a_fp4, scale_b_fp4, alpha, dtype
-            )
-
-        return run
-
-    # Quantize activation on-the-fly
-    def run():
-        a_fp4, scale_a_fp4 = ops.scaled_fp4_quant(a, a_global_scale)
-        return ops.cutlass_scaled_fp4_mm(
-            a_fp4, b_fp4, scale_a_fp4, scale_b_fp4, alpha, dtype
-        )
-
-    return run
-
-
-@triton.testing.perf_report(
-    triton.testing.Benchmark(
-        x_names=["batch_size"],
-        x_vals=[1, 16, 64, 128, 256, 512, 1024, 2048, 4096, 8192, 16384],
-        x_log=False,
-        line_arg="provider",
-        line_vals=_enabled,
-        line_names=_enabled,
-        ylabel="TFLOP/s (larger is better)",
-        plot_name="BF16 vs NVFP4 GEMMs",
-        args={},
-    )
-)
-def benchmark(batch_size, provider, N, K):
-    M = batch_size
-    device = "cuda"
-    dtype = torch.bfloat16
-
-    a = torch.randn((M, K), device=device, dtype=dtype)
-    b = torch.randn((N, K), device=device, dtype=dtype)
-
-    quantiles = [0.5, 0.2, 0.8]
-
-    if provider == "torch-bf16":
-        ms, min_ms, max_ms = triton.testing.do_bench_cudagraph(
-            lambda: torch.nn.functional.linear(a, b), quantiles=quantiles
-        )
-    else:
-        cfg = PROVIDER_CFGS[provider]
-        run_quant = build_nvfp4_runner(cfg, a, b, dtype, device)
-        ms, min_ms, max_ms = triton.testing.do_bench_cudagraph(
-            lambda: run_quant(), quantiles=quantiles
-        )
-
-    to_tflops = lambda t_ms: (2 * M * N * K) * 1e-12 / (t_ms * 1e-3)
-    return to_tflops(ms), to_tflops(max_ms), to_tflops(min_ms)
-
-
-def prepare_shapes(args):
-    out = []
-    for model, tp_size in itertools.product(args.models, args.tp_sizes):
-        for KN, tp_dim in copy.deepcopy(WEIGHT_SHAPES[model]):
-            KN[tp_dim] //= tp_size
-            KN.append(model)
-            out.append(KN)
-    return out
-
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser()
-    parser.add_argument(
-        "--models",
-        nargs="+",
-        type=str,
-        default=["meta-llama/Llama-3.1-8B-Instruct"],
-        choices=list(WEIGHT_SHAPES.keys()),
-    )
-    parser.add_argument("--tp-sizes", nargs="+", type=int, default=[1])
-    args = parser.parse_args()
-
-    for K, N, model in prepare_shapes(args):
-        print(f"{model}, N={N} K={K}, BF16 vs NVFP4 GEMMs TFLOP/s:")
-        benchmark.run(
-            print_data=True,
-            show_plots=True,
-            save_path=f"bench_nvfp4_res_n{N}_k{K}",
-            N=N,
-            K=K,
-        )
-
-    print("Benchmark finished!")
--- a/benchmarks/kernels/bench_per_token_quant_fp8.py
+++ b/benchmarks/kernels/bench_per_token_quant_fp8.py
@ -1,98 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-import itertools
-from typing import Callable
-
-import torch
-
-from vllm import _custom_ops as ops
-from vllm.config import CompilationConfig, VllmConfig, set_current_vllm_config
-from vllm.model_executor.layers.quantization.input_quant_fp8 import QuantFP8
-from vllm.model_executor.layers.quantization.utils.quant_utils import GroupShape
-from vllm.triton_utils import triton
-
-
-# TODO(luka): use standalone_compile utility
-def with_dyn_arg(fn: Callable, arg_index: int, dim_index: int):
-    def inner(*args):
-        torch._dynamo.mark_dynamic(args[arg_index], dim_index)
-        return fn(*args)
-
-    return inner
-
-
-torch._dynamo.config.recompile_limit = 8888
-compilation_config = CompilationConfig(custom_ops=["none"])
-with set_current_vllm_config(VllmConfig(compilation_config=compilation_config)):
-    torch_per_token_quant_fp8 = torch.compile(
-        QuantFP8(False, GroupShape.PER_TOKEN),
-        fullgraph=True,
-        dynamic=False,  # recompile for different shapes
-    )
-
-    # First dim is explicitly dynamic to simulate vLLM usage
-    torch_per_token_quant_fp8 = with_dyn_arg(torch_per_token_quant_fp8, 0, 0)
-
-
-def cuda_per_token_quant_fp8(
-    input: torch.Tensor,
-) -> tuple[torch.Tensor, torch.Tensor]:
-    return ops.scaled_fp8_quant(input)
-
-
-def calculate_diff(batch_size: int, seq_len: int):
-    """Calculate difference between Triton and CUDA implementations."""
-    device = torch.device("cuda")
-    x = torch.rand((batch_size * seq_len, 4096), dtype=torch.float16, device=device)
-
-    torch_out, torch_scale = torch_per_token_quant_fp8(x)
-    cuda_out, cuda_scale = cuda_per_token_quant_fp8(x)
-
-    if torch.allclose(
-        cuda_out.to(torch.float32), torch_out.to(torch.float32), rtol=1e-3, atol=1e-5
-    ) and torch.allclose(cuda_scale, torch_scale, rtol=1e-3, atol=1e-5):
-        print("✅ All implementations match")
-    else:
-        print("❌ Implementations differ")
-
-
-batch_size_range = [1, 16, 32, 64, 128]
-seq_len_range = [1, 16, 64, 128, 256, 512, 1024, 2048, 4096]
-
-configs = list(itertools.product(batch_size_range, seq_len_range))
-
-
-@triton.testing.perf_report(
-    triton.testing.Benchmark(
-        x_names=["batch_size", "seq_len"],
-        x_vals=configs,
-        line_arg="provider",
-        line_vals=["torch", "cuda"],
-        line_names=["Torch", "CUDA"],
-        styles=[("blue", "-"), ("green", "-")],
-        ylabel="us",
-        plot_name="per-token-dynamic-quant-fp8-performance",
-        args={},
-    )
-)
-def benchmark_quantization(batch_size, seq_len, provider):
-    dtype = torch.float16
-    device = torch.device("cuda")
-
-    x = torch.randn(batch_size * seq_len, 4096, device=device, dtype=dtype)
-
-    quantiles = [0.5, 0.2, 0.8]
-
-    if provider == "torch":
-        fn = lambda: torch_per_token_quant_fp8(x.clone())
-    elif provider == "cuda":
-        fn = lambda: cuda_per_token_quant_fp8(x.clone())
-
-    ms, min_ms, max_ms = triton.testing.do_bench_cudagraph(fn, quantiles=quantiles)
-
-    return 1000 * ms, 1000 * max_ms, 1000 * min_ms
-
-
-if __name__ == "__main__":
-    calculate_diff(batch_size=4, seq_len=4096)
-    benchmark_quantization.run(print_data=True)
--- a/benchmarks/kernels/benchmark_aqlm.py
+++ b/benchmarks/kernels/benchmark_aqlm.py
@ -0,0 +1,345 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+
+import os
+import sys
+from typing import Optional
+
+import torch
+import torch.nn.functional as F
+
+from vllm import _custom_ops as ops
+from vllm.model_executor.layers.quantization.aqlm import (
+    dequantize_weight,
+    generic_dequantize_gemm,
+    get_int_dtype,
+    optimized_dequantize_gemm,
+)
+from vllm.utils import FlexibleArgumentParser
+
+os.environ["CUDA_VISIBLE_DEVICES"] = "0"
+
+
+def torch_mult(
+    # [..., in_features]
+    input: torch.Tensor,
+    weights: torch.Tensor,
+    # [num_out_groups, 1, 1, 1]
+    scales: torch.Tensor,
+) -> torch.Tensor:
+    output = F.linear(input, weights)
+    return output
+
+
+def dequant_out_scale(
+    # [..., in_features]
+    input: torch.Tensor,
+    # [num_out_groups, num_in_groups, num_codebooks]
+    codes: torch.IntTensor,
+    # [num_codebooks, codebook_size, out_group_size, in_group_size]
+    codebooks: torch.Tensor,
+    # [num_out_groups, 1, 1, 1]
+    scales: torch.Tensor,
+    output_partition_sizes: torch.IntTensor,
+    bias: Optional[torch.Tensor],
+) -> torch.Tensor:
+    weights = ops.aqlm_dequant(codes, codebooks, output_partition_sizes)
+
+    if bias is None:
+        output = F.linear(input, weights, bias)
+        orig_shape = output.shape
+        flattened_output = output.view(-1, output.size(-1))
+        f_scales = scales.view(-1, scales.shape[0])
+        b_scales = f_scales.expand(flattened_output.shape[0], -1)
+        flattened_output *= b_scales
+        return flattened_output.view(orig_shape)
+    else:
+        b_scales = scales.view(scales.shape[:-3] + (-1,)).expand(-1, weights.shape[1])
+        weights *= b_scales
+        return F.linear(input, weights, bias)
+
+
+def dequant_weight_scale(
+    # [..., in_features]
+    input: torch.Tensor,
+    # [num_out_groups, num_in_groups, num_codebooks]
+    codes: torch.IntTensor,
+    # [num_codebooks, codebook_size, out_group_size, in_group_size]
+    codebooks: torch.Tensor,
+    # [num_out_groups, 1, 1, 1]
+    scales: torch.Tensor,
+    output_partition_sizes: torch.IntTensor,
+    bias: Optional[torch.Tensor],
+) -> torch.Tensor:
+    weights = ops.aqlm_dequant(codes, codebooks, output_partition_sizes)
+
+    b_scales = scales.view(scales.shape[:-3] + (-1,)).expand(-1, weights.shape[1])
+    weights *= b_scales
+    return F.linear(input, weights, bias)
+
+
+def dequant_no_scale(
+    # [..., in_features]
+    input: torch.Tensor,
+    # [num_out_groups, num_in_groups, num_codebooks]
+    codes: torch.IntTensor,
+    # [num_codebooks, codebook_size, out_group_size, in_group_size]
+    codebooks: torch.Tensor,
+    # [num_out_groups, 1, 1, 1]
+    scales: torch.Tensor,
+    output_partition_sizes: torch.IntTensor,
+    bias: Optional[torch.Tensor],
+) -> torch.Tensor:
+    weights = ops.aqlm_dequant(codes, codebooks, output_partition_sizes)
+
+    return F.linear(input, weights, bias)
+
+
+# Compare the optimized 1x16 and 2x8 cuda decompression/dequant kernels against
+# the generic pytorch version.
+# Just visual comparison.
+def dequant_test(k: int, parts: torch.Tensor, nbooks: int, bits: int) -> None:
+    n = int(parts.sum().item())
+
+    device = torch.device("cuda:0")
+
+    code_range = (1 << bits) // 2
+    ingroups = 8
+
+    codes = torch.randint(
+        -code_range,
+        code_range,
+        size=(n, k // ingroups, nbooks),
+        dtype=get_int_dtype(bits),
+        device=device,
+    )
+
+    codebooks = torch.randn(
+        size=(parts.shape[0] * nbooks, 1 << bits, 1, 8),
+        dtype=torch.float16,
+        device=device,
+    )
+
+    count = 0
+    for index in range(16):
+        for i in range(8):
+            for book in range(nbooks):
+                codebooks[book, index, 0, i] = count * (10**book)
+            count += 1
+
+    print("codes shape", codes.shape)
+
+    for i in range(16):
+        for book in range(nbooks):
+            codes[0, i, book] = i
+            codes[0, -i, book] = i
+
+    weights = dequantize_weight(codes, codebooks, None)
+    weights2 = ops.aqlm_dequant(codes, codebooks, parts)
+
+    print("weights shape:", weights.shape)
+    print("weights2 shape:", weights2.shape)
+
+    print("weights are:", weights)
+    print("weights2 are:", weights2)
+
+    print("first 128 weights are", weights[0, 0:128].to(torch.int32))
+    print("first 128 weights2 are:", weights2[0, 0:128].to(torch.int32))
+
+    print("last 128 weights are", weights[0, -128:])
+    print("last 128 weights2 are:", weights2[0, -128:])
+
+
+def main():
+    parser = FlexibleArgumentParser(description="Benchmark aqlm performance.")
+
+    # Add arguments
+    parser.add_argument(
+        "--nbooks", type=int, default=1, help="Number of codebooks (default: 1)"
+    )
+    parser.add_argument(
+        "--bits",
+        type=int,
+        default=16,
+        help="Number of bits per code element (default: 16)",
+    )
+    parser.add_argument(
+        "--test",
+        type=bool,
+        default=False,
+        help="Run the decompression/dequant tester rather than benchmarking "
+        "(default: False)",
+    )
+
+    # Parse the arguments
+    args = parser.parse_args()
+
+    # Extract values
+    nbooks = args.nbooks
+    bits = args.bits
+
+    if args.test:
+        dequant_test(4096, torch.tensor((4096,)), nbooks, bits)
+        return
+
+    # Otherwise, benchmark.
+    methods = [
+        ops.aqlm_gemm,
+        dequant_out_scale,
+        generic_dequantize_gemm,
+        optimized_dequantize_gemm,
+        dequant_weight_scale,
+        torch_mult,
+        dequant_no_scale,
+    ]
+
+    filename = f"./aqlm_benchmark_{nbooks}x{bits}.csv"
+    print(f"writing benchmarks to file {filename}")
+    with open(filename, "w") as f:
+        sys.stdout = f
+
+        print("m | k | n | n parts", end="")
+        for method in methods:
+            print(f" | {method.__name__.replace('_', ' ')} (µs)", end="")
+        print("")
+
+        # These are reasonable prefill sizes.
+        ksandpartions = (
+            (4096, (4096, 4096, 4096)),
+            (4096, (4096,)),
+            (4096, (11008, 11008)),
+            (11008, (4096,)),
+        )
+
+        # reasonable ranges for m.
+        for m in [
+            1,
+            2,
+            4,
+            8,
+            10,
+            12,
+            14,
+            16,
+            24,
+            32,
+            48,
+            52,
+            56,
+            64,
+            96,
+            112,
+            128,
+            256,
+            512,
+            1024,
+            1536,
+            2048,
+            3072,
+            4096,
+        ]:
+            print(f"{m}", file=sys.__stdout__)
+            for ksp in ksandpartions:
+                run_grid(m, ksp[0], torch.tensor(ksp[1]), nbooks, bits, methods)
+
+        sys.stdout = sys.__stdout__
+
+
+def run_grid(m: int, k: int, parts: torch.Tensor, nbooks: int, bits: int, methods):
+    # I didn't see visible improvements from increasing these, but feel free :)
+    num_warmup_trials = 1
+    num_trials = 1
+
+    num_calls = 100
+
+    # warmup.
+    for method in methods:
+        for _ in range(num_warmup_trials):
+            run_timing(
+                num_calls=num_calls,
+                m=m,
+                k=k,
+                parts=parts,
+                nbooks=nbooks,
+                bits=bits,
+                method=method,
+            )
+
+    n = parts.sum().item()
+    print(f"{m} | {k} | {n} | {parts.tolist()}", end="")
+
+    for method in methods:
+        best_time_us = 1e20
+        for _ in range(num_trials):
+            kernel_dur_ms = run_timing(
+                num_calls=num_calls,
+                m=m,
+                k=k,
+                parts=parts,
+                nbooks=nbooks,
+                bits=bits,
+                method=method,
+            )
+
+            kernel_dur_us = 1000 * kernel_dur_ms
+
+            if kernel_dur_us < best_time_us:
+                best_time_us = kernel_dur_us
+
+        print(f" | {kernel_dur_us:.0f}", end="")
+
+    print("")
+
+
+def run_timing(
+    num_calls: int, m: int, k: int, parts: torch.Tensor, nbooks: int, bits: int, method
+) -> float:
+    n = int(parts.sum().item())
+
+    device = torch.device("cuda:0")
+
+    input = torch.randn((1, m, k), dtype=torch.float16, device=device)
+
+    code_range = (1 << bits) // 2
+    ingroups = 8
+
+    codes = torch.randint(
+        -code_range,
+        code_range,
+        size=(n, k // ingroups, nbooks),
+        dtype=get_int_dtype(bits),
+        device=device,
+    )
+
+    codebooks = torch.randn(
+        size=(parts.shape[0] * nbooks, 1 << bits, 1, 8),
+        dtype=torch.float16,
+        device=device,
+    )
+
+    scales = torch.randn(size=(n, 1, 1, 1), dtype=torch.float16, device=device)
+
+    # for comparison to just a pytorch mult.
+    weights = torch.randn((n, k), dtype=torch.float16, device=device)
+
+    start_event = torch.cuda.Event(enable_timing=True)
+    end_event = torch.cuda.Event(enable_timing=True)
+
+    start_event.record()
+
+    if method is torch_mult:
+        for i in range(num_calls):
+            torch_mult(input, weights, scales)
+    else:
+        for i in range(num_calls):
+            method(input, codes, codebooks, scales, parts, None)
+
+    end_event.record()
+    end_event.synchronize()
+
+    dur_ms = start_event.elapsed_time(end_event) / num_calls
+    return dur_ms
+
+
+if __name__ == "__main__":
+    sys.exit(main())
--- a/benchmarks/kernels/benchmark_bitblas.py
+++ b/benchmarks/kernels/benchmark_bitblas.py
@ -3,8 +3,6 @@
 # Copyright (c) Microsoft Corporation.
 # Licensed under the MIT License.

-from packaging import version
-
 from vllm.model_executor.layers.quantization.utils.bitblas_utils import (
    MINIMUM_BITBLAS_VERSION,
 )
@ -12,7 +10,7 @@ from vllm.model_executor.layers.quantization.utils.bitblas_utils import (
 try:
    import bitblas

-    if version.parse(bitblas.__version__) < version.parse(MINIMUM_BITBLAS_VERSION):
+    if bitblas.__version__ < MINIMUM_BITBLAS_VERSION:
        raise ImportError(
            "bitblas version is wrong. Please "
            f"install bitblas>={MINIMUM_BITBLAS_VERSION}"
--- a/benchmarks/kernels/benchmark_cutlass_fp4_moe.py
+++ b/benchmarks/kernels/benchmark_cutlass_fp4_moe.py
@ -91,7 +91,7 @@ def bench_run(

    score = torch.randn((m, num_experts), device=device, dtype=dtype)

-    topk_weights, topk_ids, _ = fused_topk(a, score, topk, renormalize=False)
+    topk_weights, topk_ids = fused_topk(a, score, topk, renormalize=False)

    quant_blocksize = 16
    w1_blockscale = torch.empty(
--- a/benchmarks/kernels/benchmark_grouped_gemm_cutlass.py
+++ b/benchmarks/kernels/benchmark_grouped_gemm_cutlass.py
@ -7,8 +7,8 @@ from benchmark_shapes import WEIGHT_SHAPES_MOE

 from vllm import _custom_ops as ops
 from vllm.config import ParallelConfig, VllmConfig, set_current_vllm_config
-from vllm.model_executor.layers.fused_moe.cutlass_moe import cutlass_moe_fp8
 from vllm.model_executor.layers.fused_moe.fused_moe import (
+    cutlass_moe_fp8,
    fused_experts,
    fused_topk,
 )
@ -70,9 +70,18 @@ def bench_run(
    w1_scale = torch.empty((num_experts, 1, 1), device="cuda", dtype=torch.float32)
    w2_scale = torch.empty((num_experts, 1, 1), device="cuda", dtype=torch.float32)

+    ab_strides1 = torch.full((num_experts,), k, device="cuda", dtype=torch.int64)
+    c_strides1 = torch.full((num_experts,), 2 * n, device="cuda", dtype=torch.int64)
+    ab_strides2 = torch.full((num_experts,), n, device="cuda", dtype=torch.int64)
+    c_strides2 = torch.full((num_experts,), k, device="cuda", dtype=torch.int64)
+
    for expert in range(num_experts):
        w1_q[expert], w1_scale[expert] = ops.scaled_fp8_quant(w1[expert])
        w2_q[expert], w2_scale[expert] = ops.scaled_fp8_quant(w2[expert])
+    w1_q_notransp = w1_q.clone()
+    w2_q_notransp = w2_q.clone()
+    w1_q = w1_q.transpose(1, 2)
+    w2_q = w2_q.transpose(1, 2)

    score = torch.randn((m, num_experts), device="cuda", dtype=dtype)

@ -80,11 +89,6 @@ def bench_run(
        a, score, topk, renormalize=False
    )

-    ab_strides1 = torch.full((num_experts,), k, device="cuda", dtype=torch.int64)
-    ab_strides2 = torch.full((num_experts,), n, device="cuda", dtype=torch.int64)
-    c_strides1 = torch.full((num_experts,), 2 * n, device="cuda", dtype=torch.int64)
-    c_strides2 = torch.full((num_experts,), k, device="cuda", dtype=torch.int64)
-
    def run_triton_moe(
        a: torch.Tensor,
        w1: torch.Tensor,
@ -116,13 +120,12 @@ def bench_run(
        w2: torch.Tensor,
        w1_scale: torch.Tensor,
        w2_scale: torch.Tensor,
-        ab_strides1: torch.Tensor,
-        ab_strides2: torch.Tensor,
-        c_strides1: torch.Tensor,
-        c_strides2: torch.Tensor,
        topk_weights: torch.Tensor,
        topk_ids: torch.Tensor,
-        per_act_token: bool,
+        ab_strides1: torch.Tensor,
+        c_strides1: torch.Tensor,
+        ab_strides2: torch.Tensor,
+        c_strides2: torch.Tensor,
        num_repeats: int,
    ):
        for _ in range(num_repeats):
@ -130,16 +133,15 @@ def bench_run(
                a,
                w1,
                w2,
-                topk_weights,
-                topk_ids,
                w1_scale,
                w2_scale,
+                topk_weights,
+                topk_ids,
                ab_strides1,
-                ab_strides2,
                c_strides1,
+                ab_strides2,
                c_strides2,
-                per_act_token,
-                a1_scale=None,
+                a1_scale=a_scale,
            )

    def run_cutlass_from_graph(
@ -149,12 +151,12 @@ def bench_run(
        w2_q: torch.Tensor,
        w1_scale: torch.Tensor,
        w2_scale: torch.Tensor,
-        ab_strides1: torch.Tensor,
-        ab_strides2: torch.Tensor,
-        c_strides1: torch.Tensor,
-        c_strides2: torch.Tensor,
        topk_weights: torch.Tensor,
        topk_ids: torch.Tensor,
+        ab_strides1: torch.Tensor,
+        c_strides1: torch.Tensor,
+        ab_strides2: torch.Tensor,
+        c_strides2: torch.Tensor,
    ):
        with set_current_vllm_config(
            VllmConfig(parallel_config=ParallelConfig(pipeline_parallel_size=1))
@ -163,16 +165,15 @@ def bench_run(
                a,
                w1_q,
                w2_q,
-                topk_weights,
-                topk_ids,
                w1_scale,
                w2_scale,
+                topk_weights,
+                topk_ids,
                ab_strides1,
-                ab_strides2,
                c_strides1,
+                ab_strides2,
                c_strides2,
-                per_act_token,
-                a1_scale=None,
+                a1_scale=a_scale,
            )

    def run_triton_from_graph(
@ -215,12 +216,12 @@ def bench_run(
            w2_q,
            w1_scale,
            w2_scale,
-            ab_strides1,
-            ab_strides2,
-            c_strides1,
-            c_strides2,
            topk_weights,
            topk_ids,
+            ab_strides1,
+            c_strides1,
+            ab_strides2,
+            c_strides2,
        )
    torch.cuda.synchronize()

@ -229,8 +230,8 @@ def bench_run(
    with torch.cuda.graph(triton_graph, stream=triton_stream):
        run_triton_from_graph(
            a,
-            w1_q,
-            w2_q,
+            w1_q_notransp,
+            w2_q_notransp,
            topk_weights,
            topk_ids,
            w1_scale,
@ -249,16 +250,17 @@ def bench_run(
        "w2": w2,
        "score": score,
        "topk": topk,
+        "w1_q_notransp": w1_q_notransp,
+        "w2_q_notransp": w2_q_notransp,
        # Cutlass params
        "a_scale": a_scale,
        "w1_q": w1_q,
        "w2_q": w2_q,
        "w1_scale": w1_scale,
        "w2_scale": w2_scale,
-        "per_act_token": per_act_token,
        "ab_strides1": ab_strides1,
-        "ab_strides2": ab_strides2,
        "c_strides1": c_strides1,
+        "ab_strides2": ab_strides2,
        "c_strides2": c_strides2,
        # cuda graph params
        "cutlass_graph": cutlass_graph,
@ -277,8 +279,8 @@ def bench_run(
    # Warmup
    run_triton_moe(
        a,
-        w1_q,
-        w2_q,
+        w1_q_notransp,
+        w2_q_notransp,
        topk_weights,
        topk_ids,
        w1_scale,
@ -289,7 +291,7 @@ def bench_run(

    results.append(
        benchmark.Timer(
-            stmt="run_triton_moe(a, w1_q, w2_q, topk_weights, topk_ids, w1_scale, w2_scale, a_scale, num_runs)",  # noqa: E501
+            stmt="run_triton_moe(a, w1_q_notransp, w2_q_notransp, topk_weights, topk_ids, w1_scale, w2_scale, a_scale, num_runs)",  # noqa: E501
            globals=globals,
            label=label,
            sub_label=sub_label,
@ -318,19 +320,18 @@ def bench_run(
        w2_q,
        w1_scale,
        w2_scale,
-        ab_strides1,
-        ab_strides2,
-        c_strides1,
-        c_strides2,
        topk_weights,
        topk_ids,
-        per_act_token,
+        ab_strides1,
+        c_strides1,
+        ab_strides2,
+        c_strides2,
        num_warmup,
    )

    results.append(
        benchmark.Timer(
-            stmt="run_cutlass_moe(a, a_scale, w1_q, w2_q, w1_scale, w2_scale, ab_strides1, ab_strides2, c_strides1, c_strides2, topk_weights, topk_ids, per_act_token, num_runs)",  # noqa: E501
+            stmt="run_cutlass_moe(a, a_scale, w1_q, w2_q, w1_scale, w2_scale, topk_weights, topk_ids, ab_strides1, c_strides1, ab_strides2, c_strides2, num_runs)",  # noqa: E501
            globals=globals,
            label=label,
            sub_label=sub_label,
--- a/benchmarks/kernels/benchmark_machete.py
+++ b/benchmarks/kernels/benchmark_machete.py
@ -234,11 +234,8 @@ def marlin_create_bench_fn(bt: BenchmarkTensors) -> Callable:

        fn = lambda: ops.gptq_marlin_gemm(
            a=bt.a,
-            c=None,
            b_q_weight=w_q,
-            b_bias=None,
            b_scales=w_s,
-            global_scale=None,
            b_zeros=w_zp,
            g_idx=g_idx,
            perm=sort_indices,
@ -253,7 +250,28 @@ def marlin_create_bench_fn(bt: BenchmarkTensors) -> Callable:
    else:
        assert bt.a.dtype == torch.int8
        assert bt.wtype == scalar_types.uint4b8
-        raise NotImplementedError("QQQ is not supported anymore")
+
+        if bt.w_ch_s is not None:
+            s_ch = bt.w_ch_s.to(torch.float32)
+        else:
+            s_ch = torch.ones(bt.w_ref.shape[1], dtype=torch.float32, device=device)
+
+        if bt.w_tok_s is not None:
+            s_tok = bt.w_tok_s.to(torch.float32)
+        else:
+            s_tok = torch.ones(bt.a.shape[0], dtype=torch.float32, device=device)
+
+        fn = lambda: ops.marlin_qqq_gemm(
+            a=bt.a,
+            b_q_weight=w_q,
+            s_group=w_s,
+            s_tok=s_tok,
+            s_ch=s_ch,
+            workspace=workspace.scratch,
+            size_m=bt.a.shape[0],
+            size_n=bt.w_ref.shape[1],
+            size_k=bt.w_ref.shape[0],
+        )

    return fn

@ -284,25 +302,6 @@ def machete_create_bench_fn(
    )


-def cutlass_w4a8_create_bench_fn(
-    bt: BenchmarkTensors, out_type=torch.dtype, schedule=None
-) -> Callable:
-    w_q = bt.w_q.t().contiguous().t()  # make col major
-    w_q = ops.cutlass_encode_and_reorder_int4b(w_q)
-    # expects fp8 scales
-    w_s = ops.cutlass_pack_scale_fp8(bt.w_g_s.to(torch.float8_e4m3fn))
-
-    return lambda: ops.cutlass_w4a8_mm(
-        a=bt.a,
-        b_q=w_q,
-        b_group_scales=w_s,
-        b_group_size=bt.group_size,
-        b_channel_scales=bt.w_ch_s,
-        a_token_scales=bt.w_tok_s,
-        maybe_schedule=schedule,
-    )
-
-
 # impl

 # bench
@ -404,20 +403,6 @@ def bench(
        )
    )

-    # cutlass w4a8
-    if types.act_type == torch.float8_e4m3fn and group_size == 128:
-        timers.append(
-            bench_fns(
-                label,
-                sub_label,
-                f"cutlass w4a8 ({name_type_string})",
-                [
-                    cutlass_w4a8_create_bench_fn(bt, out_type=types.output_type)
-                    for bt in benchmark_tensors
-                ],
-            )
-        )
-
    if sweep_schedules:
        global _SWEEP_SCHEDULES_RESULTS

--- a/benchmarks/kernels/benchmark_marlin.py
+++ b/benchmarks/kernels/benchmark_marlin.py
@ -22,16 +22,8 @@ from vllm.model_executor.layers.quantization.utils.marlin_utils import (
    MARLIN_SUPPORTED_GROUP_SIZES,
    query_marlin_supported_quant_types,
 )
-from vllm.model_executor.layers.quantization.utils.marlin_utils_fp4 import (
-    FP4_MARLIN_SUPPORTED_GROUP_SIZES,
-    rand_marlin_weight_fp4_like,
-)
-from vllm.model_executor.layers.quantization.utils.marlin_utils_fp8 import (
-    marlin_quant_fp8_torch,
-)
 from vllm.model_executor.layers.quantization.utils.marlin_utils_test import (
    MarlinWorkspace,
-    awq_marlin_quantize,
    marlin_quantize,
 )
 from vllm.model_executor.layers.quantization.utils.marlin_utils_test_24 import (
@ -43,7 +35,7 @@ from vllm.model_executor.layers.quantization.utils.quant_utils import (
    quantize_weights,
    sort_weights,
 )
-from vllm.scalar_type import ScalarType, scalar_types
+from vllm.scalar_type import ScalarType
 from vllm.utils import FlexibleArgumentParser

 DEFAULT_MODELS = ["meta-llama/Llama-2-7b-hf/TP1"]
@ -65,144 +57,80 @@ def bench_run(
    size_n: int,
 ):
    label = "Quant Matmul"
+
    sub_label = "{}, act={} k_full={}, q={}, g={}, MKN=({}x{}x{})".format(
        model, act_order, is_k_full, str(quant_type), group_size, size_m, size_k, size_n
    )
+
    print(f"Testing: {sub_label}")

    a = torch.randn(size_m, size_k).to(torch.half).cuda()
    b = torch.rand(size_k, size_n).to(torch.half).cuda()
-    has_zp = quant_type in [scalar_types.uint4, scalar_types.uint8]
-    if act_order and (group_size == -1 or group_size == size_k or has_zp):
-        return
-    if size_k % group_size != 0:
-        return

-    marlin_24_supported = (
-        quant_type in GPTQ_MARLIN_24_SUPPORTED_QUANT_TYPES
-        and group_size in GPTQ_MARLIN_24_SUPPORTED_GROUP_SIZES
-    )
-    repack_supported = (
-        quant_type in GPTQ_MARLIN_24_SUPPORTED_QUANT_TYPES
-        and group_size in MARLIN_SUPPORTED_GROUP_SIZES
-    )
-    allspark_supported = (
-        quant_type in ALLSPARK_SUPPORTED_QUANT_TYPES
-        and group_size == -1
-        and not act_order
-        and is_k_full
-    )
-
-    def gen_marlin_params():
-        # Marlin quant
-        marlin_g_idx = marlin_sort_indices = marlin_zp = marlin_s2 = None
-        if quant_type == scalar_types.float4_e2m1f:
-            if group_size != 16 or act_order:
-                return
-            marlin_w_ref, marlin_q_w, marlin_s, marlin_s2 = rand_marlin_weight_fp4_like(
-                b.T, group_size
-            )
-        elif quant_type == scalar_types.float8_e4m3fn:
-            if group_size not in [-1, 128] or act_order:
-                return
-            marlin_w_ref, marlin_q_w, marlin_s = marlin_quant_fp8_torch(b.T, group_size)
-        elif group_size == 16:
-            return
-        elif has_zp:
-            marlin_w_ref, marlin_q_w, marlin_s, marlin_zp = awq_marlin_quantize(
-                b, quant_type, group_size
-            )
-        else:
-            marlin_w_ref, marlin_q_w, marlin_s, marlin_g_idx, marlin_sort_indices, _ = (
-                marlin_quantize(b, quant_type, group_size, act_order)
-            )
-        return (
-            marlin_w_ref,
-            marlin_q_w,
-            marlin_s,
-            marlin_s2,
-            marlin_zp,
-            marlin_g_idx,
-            marlin_sort_indices,
-        )
-
-    def gen_marlin_24_params():
-        marlin_24_w_ref = marlin_24_q_w_comp = marlin_24_meta = marlin_24_s = None
-        if marlin_24_supported:
-            (marlin_24_w_ref, marlin_24_q_w_comp, marlin_24_meta, marlin_24_s) = (
-                marlin_24_quantize(b, quant_type, group_size)
-            )
-        return (marlin_24_w_ref, marlin_24_q_w_comp, marlin_24_meta, marlin_24_s)
-
-    def gen_repack_params():
-        q_w_gptq = None
-        repack_sort_indices = None
-        if repack_supported:
-            (w_ref, q_w, s, g_idx, rand_perm) = gptq_quantize_weights(
-                b, quant_type, group_size, act_order
-            )
-            q_w_gptq = gptq_pack(q_w, quant_type.size_bits, size_k, size_n)
-
-            # For act_order, sort the "weights" and "g_idx"
-            # so that group ids are increasing
-            repack_sort_indices = torch.empty(0, dtype=torch.int, device=b.device)
-            if act_order:
-                (q_w, g_idx, repack_sort_indices) = sort_weights(q_w, g_idx)
-        return q_w_gptq, repack_sort_indices
-
-    def gen_allspark_params():
-        qw_reorder = s_reorder = zp_reorder = sm_count = sm_version = (
-            CUBLAS_M_THRESHOLD
-        ) = None
-        nonlocal allspark_supported
-        if allspark_supported:
-            properties = torch.cuda.get_device_properties(b.device.index)
-            sm_count = properties.multi_processor_count
-            sm_version = properties.major * 10 + properties.minor
-
-            supported_arch = sm_version >= 80 and sm_version < 90
-            allspark_supported = allspark_supported and supported_arch
-            if supported_arch:
-                w_ref, qw, s, zp = quantize_weights(b, quant_type, group_size, has_zp)
-                qw = qw.to(torch.uint8)
-
-                qw_reorder, s_reorder, zp_reorder = ops.allspark_repack_weight(
-                    qw, s, zp, has_zp
-                )
-                CUBLAS_M_THRESHOLD = ALLSPARK_AMPERE_M_CUBLAS_THRESHOLD
-        return (
-            qw_reorder,
-            s_reorder,
-            zp_reorder,
-            sm_count,
-            sm_version,
-            CUBLAS_M_THRESHOLD,
-        )
+    a_tmp = torch.zeros(size_m, size_k).to(torch.half).cuda()

+    # Marlin quant
    (
        marlin_w_ref,
        marlin_q_w,
        marlin_s,
-        marlin_s2,
-        marlin_zp,
        marlin_g_idx,
        marlin_sort_indices,
-    ) = gen_marlin_params()
-    marlin_24_w_ref, marlin_24_q_w_comp, marlin_24_meta, marlin_24_s = (
-        gen_marlin_24_params()
+        marlin_rand_perm,
+    ) = marlin_quantize(b, quant_type, group_size, act_order)
+
+    # Marlin_24 quant
+    (marlin_24_w_ref, marlin_24_q_w_comp, marlin_24_meta, marlin_24_s) = (
+        marlin_24_quantize(b, quant_type, group_size)
    )
-    q_w_gptq, repack_sort_indices = gen_repack_params()
-    qw_reorder, s_reorder, zp_reorder, sm_count, sm_version, CUBLAS_M_THRESHOLD = (
-        gen_allspark_params()
+
+    marlin_zp = torch.empty(0, dtype=torch.int, device=b.device)
+
+    # GPTQ quant
+    (w_ref, q_w, s, g_idx, rand_perm) = gptq_quantize_weights(
+        b, quant_type, group_size, act_order
    )
+    q_w_gptq = gptq_pack(q_w, quant_type.size_bits, size_k, size_n)
+
+    # For act_order, sort the "weights" and "g_idx"
+    # so that group ids are increasing
+    repack_sort_indices = torch.empty(0, dtype=torch.int, device=b.device)
+    if act_order:
+        (q_w, g_idx, repack_sort_indices) = sort_weights(q_w, g_idx)

    # Prepare
    marlin_workspace = MarlinWorkspace(
        size_n, GPTQ_MARLIN_MIN_THREAD_N, GPTQ_MARLIN_MAX_PARALLEL
    )
+
    marlin_24_workspace = MarlinWorkspace(
        size_n, GPTQ_MARLIN_24_MIN_THREAD_N, GPTQ_MARLIN_24_MAX_PARALLEL
    )
+    marlin_zp = torch.zeros_like(marlin_s, dtype=torch.int)
+
+    # AllSpark W8A16 quant
+    as_supported_case = (
+        quant_type in ALLSPARK_SUPPORTED_QUANT_TYPES
+        and group_size == -1
+        and not act_order
+        and is_k_full
+    )
+    if as_supported_case:
+        properties = torch.cuda.get_device_properties(b.device.index)
+        sm_count = properties.multi_processor_count
+        sm_version = properties.major * 10 + properties.minor
+
+        supported_arch = sm_version >= 80 and sm_version < 90
+        as_supported_case = as_supported_case and supported_arch
+        if supported_arch:
+            has_zp = False
+            w_ref, qw, s, zp = quantize_weights(b, quant_type, group_size, has_zp)
+            qw = qw.to(torch.uint8)
+
+            qw_reorder, s_reorder, zp_reorder = ops.allspark_repack_weight(
+                qw, s, zp, has_zp
+            )
+            CUBLAS_M_THRESHOLD = ALLSPARK_AMPERE_M_CUBLAS_THRESHOLD

    globals = {
        # Gen params
@ -212,14 +140,15 @@ def bench_run(
        "size_n": size_n,
        "size_k": size_k,
        "a": a,
+        "a_tmp": a_tmp,
        # Marlin params
        "marlin_w_ref": marlin_w_ref,
        "marlin_q_w": marlin_q_w,
        "marlin_s": marlin_s,
-        "marlin_s2": marlin_s2,
        "marlin_zp": marlin_zp,
        "marlin_g_idx": marlin_g_idx,
        "marlin_sort_indices": marlin_sort_indices,
+        "marlin_rand_perm": marlin_rand_perm,
        "marlin_workspace": marlin_workspace,
        "is_k_full": is_k_full,
        # Marlin_24 params
@ -232,12 +161,12 @@ def bench_run(
        "q_w_gptq": q_w_gptq,
        "repack_sort_indices": repack_sort_indices,
        # AllSpark W8A16 params
-        "qw_reorder": qw_reorder,
-        "s_reorder": s_reorder,
-        "zp_reorder": zp_reorder,
-        "sm_count": sm_count,
-        "sm_version": sm_version,
-        "CUBLAS_M_THRESHOLD": CUBLAS_M_THRESHOLD,
+        "qw_reorder": qw_reorder if as_supported_case else None,
+        "s_reorder": s_reorder if as_supported_case else None,
+        "zp_reorder": zp_reorder if as_supported_case else None,
+        "sm_count": sm_count if as_supported_case else None,
+        "sm_version": sm_version if as_supported_case else None,
+        "CUBLAS_M_THRESHOLD": CUBLAS_M_THRESHOLD if as_supported_case else None,
        # Kernels
        "gptq_marlin_gemm": ops.gptq_marlin_gemm,
        "gptq_marlin_24_gemm": ops.gptq_marlin_24_gemm,
@ -248,7 +177,7 @@ def bench_run(
    min_run_time = 1

    # Warmup pytorch
-    for _ in range(5):
+    for i in range(5):
        torch.matmul(a, marlin_w_ref)

    results.append(
@ -263,17 +192,17 @@ def bench_run(

    results.append(
        benchmark.Timer(
-            stmt="output = gptq_marlin_gemm(a, None, marlin_q_w, marlin_s, marlin_s2, marlin_zp, marlin_g_idx, marlin_sort_indices, marlin_workspace.scratch, quant_type, size_m, size_n, size_k, is_k_full, False, False, False)",  # noqa: E501
+            stmt="output = gptq_marlin_gemm(a, marlin_q_w, marlin_s, marlin_zp, marlin_g_idx, marlin_sort_indices, marlin_workspace.scratch, quant_type, size_m, size_n, size_k, is_k_full, False, False, False)",  # noqa: E501
            globals=globals,
            label=label,
            sub_label=sub_label,
-            description="gptq_marlin_gemm",
+            description="gptq_marlin_gemm_fp16",
        ).blocked_autorange(min_run_time=min_run_time)
    )

    results.append(
        benchmark.Timer(
-            stmt="output = gptq_marlin_gemm(a, None, marlin_q_w, marlin_s, marlin_s2, marlin_zp, marlin_g_idx, marlin_sort_indices, marlin_workspace.scratch, quant_type, size_m, size_n, size_k, is_k_full, False, True, False)",  # noqa: E501
+            stmt="output = gptq_marlin_gemm(a, marlin_q_w, marlin_s, marlin_zp, marlin_g_idx, marlin_sort_indices, marlin_workspace.scratch, quant_type, size_m, size_n, size_k, is_k_full, False, True, False)",  # noqa: E501
            globals=globals,
            label=label,
            sub_label=sub_label,
@ -281,7 +210,10 @@ def bench_run(
        ).blocked_autorange(min_run_time=min_run_time)
    )

-    if marlin_24_supported:
+    if (
+        quant_type in GPTQ_MARLIN_24_SUPPORTED_QUANT_TYPES
+        and group_size in GPTQ_MARLIN_24_SUPPORTED_GROUP_SIZES
+    ):
        results.append(
            benchmark.Timer(
                stmt="output = gptq_marlin_24_gemm(a, marlin_24_q_w_comp, marlin_24_meta, marlin_24_s, marlin_24_workspace.scratch, quant_type, size_m, size_n, size_k)",  # noqa: E501
@ -292,18 +224,17 @@ def bench_run(
            ).blocked_autorange(min_run_time=min_run_time)
        )

-    if repack_supported:
-        results.append(
-            benchmark.Timer(
-                stmt="q_res = gptq_marlin_repack(q_w_gptq, repack_sort_indices, size_k, size_n, quant_type.size_bits)",  # noqa: E501
-                globals=globals,
-                label=label,
-                sub_label=sub_label,
-                description="gptq_marlin_repack",
-            ).blocked_autorange(min_run_time=min_run_time)
-        )
+    results.append(
+        benchmark.Timer(
+            stmt="q_res = gptq_marlin_repack(q_w_gptq, repack_sort_indices, size_k, size_n, quant_type.size_bits)",  # noqa: E501
+            globals=globals,
+            label=label,
+            sub_label=sub_label,
+            description="gptq_marlin_repack",
+        ).blocked_autorange(min_run_time=min_run_time)
+    )

-    if allspark_supported:
+    if as_supported_case:
        results.append(
            benchmark.Timer(
                stmt="output = allspark_w8a16_gemm(a, qw_reorder, s_reorder, zp_reorder, size_n, group_size, sm_count, sm_version, CUBLAS_M_THRESHOLD, False, True)",  # noqa: E501
@ -319,6 +250,7 @@ def main(args):
    print("Benchmarking models:")
    for i, model in enumerate(args.models):
        print(f"[{i}]  {model}")
+
    results: list[benchmark.Measurement] = []

    for model in args.models:
@ -346,17 +278,14 @@ def main(args):
                    ):
                        continue

-                    for quant_type in query_marlin_supported_quant_types():
+                    for quant_type in query_marlin_supported_quant_types(False):
                        if (
                            len(args.limit_num_bits) > 0
                            and quant_type.size_bits not in args.limit_num_bits
                        ):
                            continue

-                        for group_size in (
-                            MARLIN_SUPPORTED_GROUP_SIZES
-                            + FP4_MARLIN_SUPPORTED_GROUP_SIZES
-                        ):
+                        for group_size in MARLIN_SUPPORTED_GROUP_SIZES:
                            if (
                                len(args.limit_group_size) > 0
                                and group_size not in args.limit_group_size
--- a/benchmarks/kernels/benchmark_moe.py
+++ b/benchmarks/kernels/benchmark_moe.py
@ -3,11 +3,11 @@

 import argparse
 import json
-import os
 import time
 from contextlib import nullcontext
 from datetime import datetime
 from itertools import product
+from types import SimpleNamespace
 from typing import Any, TypedDict

 import ray
@ -23,13 +23,6 @@ from vllm.utils import FlexibleArgumentParser
 FP8_DTYPE = current_platform.fp8_dtype()


-def ensure_divisibility(numerator, denominator, text):
-    """Ensure that numerator is divisible by the denominator."""
-    assert numerator % denominator == 0, "{} {} is not divisible by tp {}.".format(
-        text, numerator, denominator
-    )
-
-
 class BenchmarkConfig(TypedDict):
    BLOCK_SIZE_M: int
    BLOCK_SIZE_N: int
@ -50,7 +43,7 @@ def benchmark_config(
    use_fp8_w8a8: bool,
    use_int8_w8a16: bool,
    num_iters: int = 100,
-    block_quant_shape: list[int] = None,
+    block_quant_shape: List[int] = None,
    use_deep_gemm: bool = False,
 ) -> float:
    init_dtype = torch.float16 if use_fp8_w8a8 else dtype
@ -94,9 +87,6 @@ def benchmark_config(
            (num_experts, 2 * shard_intermediate_size), dtype=torch.float32
        )
        w2_scale = torch.randn((hidden_size, num_experts), dtype=torch.float32)
-    if use_deep_gemm:
-        # we use the default block shape for deepgemm
-        block_quant_shape = [128, 128]
    if use_fp8_w8a8:
        if block_quant_shape:
            block_n, block_k = block_quant_shape[0], block_quant_shape[1]
@ -410,7 +400,7 @@ class BenchmarkWorker:
        dtype: torch.dtype,
        use_fp8_w8a8: bool,
        use_int8_w8a16: bool,
-        block_quant_shape: list[int] = None,
+        block_quant_shape: List[int] = None,
        use_deep_gemm: bool = False,
    ) -> tuple[dict[str, int], float]:
        current_platform.seed_everything(self.seed)
@ -430,6 +420,7 @@ class BenchmarkWorker:
                hidden_size,
                topk,
                dtype_str,
+                is_marlin=False,
            )
        else:
            config = op_config[min(op_config.keys(), key=lambda x: abs(x - num_tokens))]
@ -541,8 +532,7 @@ def save_configs(
    dtype: torch.dtype,
    use_fp8_w8a8: bool,
    use_int8_w8a16: bool,
-    block_quant_shape: list[int],
-    save_dir: str,
+    block_quant_shape: List[int],
 ) -> None:
    dtype_str = get_config_dtype_str(
        dtype, use_int8_w8a16=use_int8_w8a16, use_fp8_w8a8=use_fp8_w8a8
@ -553,8 +543,7 @@ def save_configs(
    filename = get_config_file_name(
        num_experts, shard_intermediate_size // 2, dtype_str, block_quant_shape
    )
-    os.makedirs(save_dir, exist_ok=True)
-    filename = os.path.join(save_dir, filename)
+
    print(f"Writing best config to {filename}...")
    with open(filename, "w") as f:
        json.dump(configs, f, indent=4)
@ -574,31 +563,28 @@ def main(args: argparse.Namespace):
    config = get_config(model=args.model, trust_remote_code=args.trust_remote_code)
    if args.model_prefix:
        config = getattr(config, args.model_prefix)
+    config = SimpleNamespace(**config)

    if config.architectures[0] == "DbrxForCausalLM":
        E = config.ffn_config.moe_num_experts
        topk = config.ffn_config.moe_top_k
        intermediate_size = config.ffn_config.ffn_hidden_size
+        shard_intermediate_size = 2 * intermediate_size // args.tp_size
    elif config.architectures[0] == "JambaForCausalLM":
        E = config.num_experts
        topk = config.num_experts_per_tok
        intermediate_size = config.intermediate_size
-    elif config.architectures[0] in (
-        "DeepseekV3ForCausalLM",
-        "DeepseekV2ForCausalLM",
-        "Glm4MoeForCausalLM",
-    ):
+        shard_intermediate_size = 2 * intermediate_size // args.tp_size
+    elif config.architectures[0] in ("DeepseekV3ForCausalLM", "DeepseekV2ForCausalLM"):
        E = config.n_routed_experts
        topk = config.num_experts_per_tok
        intermediate_size = config.moe_intermediate_size
+        shard_intermediate_size = 2 * intermediate_size // args.tp_size
    elif config.architectures[0] in ("Qwen2MoeForCausalLM", "Qwen3MoeForCausalLM"):
        E = config.num_experts
        topk = config.num_experts_per_tok
        intermediate_size = config.moe_intermediate_size
-    elif config.architectures[0] in ("HunYuanMoEV1ForCausalLM"):
-        E = config.num_experts
-        topk = config.moe_topk[0]
-        intermediate_size = config.moe_intermediate_size[0]
+        shard_intermediate_size = 2 * intermediate_size // args.tp_size
    else:
        # Support for llama4
        config = config.get_text_config()
@ -606,16 +592,14 @@ def main(args: argparse.Namespace):
        E = config.num_local_experts
        topk = config.num_experts_per_tok
        intermediate_size = config.intermediate_size
-    enable_ep = bool(args.enable_expert_parallel)
-    if enable_ep:
-        ensure_divisibility(E, args.tp_size, "Number of experts")
-        E = E // args.tp_size
-        shard_intermediate_size = 2 * intermediate_size
-    else:
-        ensure_divisibility(intermediate_size, args.tp_size, "intermediate_size")
        shard_intermediate_size = 2 * intermediate_size // args.tp_size
+
    hidden_size = config.hidden_size
-    dtype = torch.float16 if current_platform.is_rocm() else config.torch_dtype
+    dtype = (
+        torch.float16
+        if current_platform.is_rocm()
+        else getattr(torch, config.torch_dtype)
+    )
    use_fp8_w8a8 = args.dtype == "fp8_w8a8"
    use_int8_w8a16 = args.dtype == "int8_w8a16"
    block_quant_shape = get_weight_block_size_safety(config)
@ -642,7 +626,7 @@ def main(args: argparse.Namespace):
            4096,
        ]
    else:
-        batch_sizes = args.batch_size
+        batch_sizes = [args.batch_size]

    use_deep_gemm = bool(args.use_deep_gemm)

@ -709,7 +693,6 @@ def main(args: argparse.Namespace):
            use_fp8_w8a8,
            use_int8_w8a16,
            block_quant_shape,
-            args.save_dir,
        )
        end = time.time()
        print(f"Tuning took {end - start:.2f} seconds")
@ -746,16 +729,12 @@ if __name__ == "__main__":
    parser.add_argument(
        "--tp-size", "-tp", "--tensor-parallel-size", type=int, default=2
    )
-    parser.add_argument("--enable-expert-parallel", "-enable-ep", action="store_true")
    parser.add_argument(
        "--dtype", type=str, choices=["auto", "fp8_w8a8", "int8_w8a16"], default="auto"
    )
    parser.add_argument("--use-deep-gemm", action="store_true")
-    parser.add_argument(
-        "--save-dir", type=str, default="./", help="Directory to save tuned results"
-    )
    parser.add_argument("--seed", type=int, default=0)
-    parser.add_argument("--batch-size", type=int, nargs="+", required=False)
+    parser.add_argument("--batch-size", type=int, required=False)
    parser.add_argument("--tune", action="store_true")
    parser.add_argument("--trust-remote-code", action="store_true")
    parser.add_argument("--model-prefix", type=str, required=False)
--- a/benchmarks/kernels/benchmark_moe_align_block_size.py
+++ b/benchmarks/kernels/benchmark_moe_align_block_size.py
@ -1,74 +0,0 @@
-# SPDX-License-Identifier: Apache-2.0
-# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
-import argparse
-import itertools
-
-import torch
-
-from vllm.model_executor.layers.fused_moe.moe_align_block_size import (
-    moe_align_block_size,
-)
-from vllm.triton_utils import triton
-
-
-def get_topk_ids(num_tokens: int, num_experts: int, topk: int) -> torch.Tensor:
-    return torch.stack(
-        [
-            torch.randperm(num_experts, dtype=torch.int32, device="cuda")[:topk]
-            for _ in range(num_tokens)
-        ]
-    )
-
-
-# test configurations
-num_tokens_range = [1, 16, 256, 4096]
-num_experts_range = [16, 64, 224, 256, 280, 512]
-topk_range = [1, 2, 8]
-configs = list(itertools.product(num_tokens_range, num_experts_range, topk_range))
-
-
-@triton.testing.perf_report(
-    triton.testing.Benchmark(
-        x_names=["num_tokens", "num_experts", "topk"],
-        x_vals=configs,
-        line_arg="provider",
-        line_vals=["vllm"],
-        line_names=["vLLM"],
-        plot_name="moe-align-block-size-performance",
-        args={},
-    )
-)
-def benchmark(num_tokens, num_experts, topk, provider):
-    """Benchmark function for Triton."""
-    block_size = 256
-    topk_ids = get_topk_ids(num_tokens, num_experts, topk)
-
-    quantiles = [0.5, 0.2, 0.8]
-
-    if provider == "vllm":
-        ms, min_ms, max_ms = triton.testing.do_bench(
-            lambda: moe_align_block_size(topk_ids, block_size, num_experts),
-            quantiles=quantiles,
-        )
-
-    return 1000 * ms, 1000 * max_ms, 1000 * min_ms
-
-
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser()
-    parser.add_argument(
-        "--num_experts",
-        type=int,
-        default=64,
-        choices=[8, 16, 32, 64, 128, 256],
-    )
-    parser.add_argument(
-        "--topk",
-        type=int,
-        default=8,
-        choices=[2, 4, 8],
-        help="Top-k value for correctness check.",
-    )
-    args = parser.parse_args()
-
-    benchmark.run(print_data=True, show_plots=True)
--- a/benchmarks/kernels/benchmark_moe_permute_unpermute.py
+++ b/benchmarks/kernels/benchmark_moe_permute_unpermute.py
@ -8,13 +8,12 @@ import ray
 import torch
 from transformers import AutoConfig

-from vllm.model_executor.layers.fused_moe.fused_moe import *
-from vllm.model_executor.layers.fused_moe.moe_permute_unpermute import (
+from vllm.model_executor.layers.fused_moe.deep_gemm_moe import (
    _moe_permute,
    _moe_unpermute_and_reduce,
-    moe_permute,
-    moe_unpermute,
 )
+from vllm.model_executor.layers.fused_moe.fused_moe import *
+from vllm.model_executor.layers.fused_moe.moe_permute_unpermute import *
 from vllm.model_executor.layers.fused_moe.utils import _fp8_quantize
 from vllm.platforms import current_platform
 from vllm.utils import FlexibleArgumentParser
@ -64,19 +63,18 @@ def benchmark_permute(

    def run():
        if use_customized_permute:
-            (
-                permuted_hidden_states,
-                a1q_scale,
-                first_token_off,
-                inv_perm_idx,
-                m_indices,
-            ) = moe_permute(
-                qhidden_states,
-                a1q_scale=None,
-                topk_ids=topk_ids,
-                n_expert=num_experts,
-                expert_map=None,
-                align_block_size=align_block_size,
+            (permuted_hidden_states, first_token_off, inv_perm_idx, m_indices) = (
+                moe_permute(
+                    qhidden_states,
+                    topk_weights=topk_weights,
+                    topk_ids=topk_ids,
+                    token_expert_indices=token_expert_indices,
+                    topk=topk,
+                    n_expert=num_experts,
+                    n_local_expert=num_experts,
+                    expert_map=None,
+                    align_block_size=align_block_size,
+                )
            )
        else:
            (
@ -152,19 +150,18 @@ def benchmark_unpermute(

    def prepare():
        if use_customized_permute:
-            (
-                permuted_hidden_states,
-                a1q_scale,
-                first_token_off,
-                inv_perm_idx,
-                m_indices,
-            ) = moe_permute(
-                qhidden_states,
-                a1q_scale=None,
-                topk_ids=topk_ids,
-                n_expert=num_experts,
-                expert_map=None,
-                align_block_size=align_block_size,
+            (permuted_hidden_states, first_token_off, inv_perm_idx, m_indices) = (
+                moe_permute(
+                    qhidden_states,
+                    topk_weights=topk_weights,
+                    topk_ids=topk_ids,
+                    token_expert_indices=token_expert_indices,
+                    topk=topk,
+                    n_expert=num_experts,
+                    n_local_expert=num_experts,
+                    expert_map=None,
+                    align_block_size=align_block_size,
+                )
            )
            # convert to fp16/bf16 as gemm output
            return (
@ -194,19 +191,16 @@ def benchmark_unpermute(

    def run(input: tuple):
        if use_customized_permute:
-            (
-                permuted_hidden_states,
-                first_token_off,
-                inv_perm_idx,
-                m_indices,
-            ) = input
-            output = torch.empty_like(hidden_states)
+            (permuted_hidden_states, first_token_off, inv_perm_idx, m_indices) = input
            moe_unpermute(
-                output,
                permuted_hidden_states,
                topk_weights,
+                topk_ids,
                inv_perm_idx,
                first_token_off,
+                topk,
+                num_experts,
+                num_experts,
            )
        else:
            (
@ -217,11 +211,7 @@ def benchmark_unpermute(
                inv_perm,
            ) = input
            _moe_unpermute_and_reduce(
-                output_hidden_states,
-                permuted_hidden_states,
-                inv_perm,
-                topk_weights,
-                True,
+                output_hidden_states, permuted_hidden_states, inv_perm, topk_weights
            )

    # JIT compilation & warmup
@ -328,7 +318,6 @@ def main(args: argparse.Namespace):
    elif (
        config.architectures[0] == "DeepseekV3ForCausalLM"
        or config.architectures[0] == "DeepseekV2ForCausalLM"
-        or config.architectures[0] == "Glm4MoeForCausalLM"
    ):
        E = config.n_routed_experts
        topk = config.num_experts_per_tok
--- a/Show More
+++ b/Show More