mirror of
https://github.com/vllm-project/vllm.git
synced 2025-10-20 14:53:52 +08:00
Build CUDA11.8 wheels for release (#1596)
This commit is contained in:
2
.github/workflows/publish.yml
vendored
2
.github/workflows/publish.yml
vendored
@ -50,7 +50,7 @@ jobs:
|
||||
os: ['ubuntu-20.04']
|
||||
python-version: ['3.8', '3.9', '3.10', '3.11']
|
||||
pytorch-version: ['2.1.0']
|
||||
cuda-version: ['12.1']
|
||||
cuda-version: ['11.8', '12.1']
|
||||
|
||||
steps:
|
||||
- name: Checkout
|
||||
|
15
setup.py
15
setup.py
@ -12,6 +12,8 @@ from torch.utils.cpp_extension import BuildExtension, CUDAExtension, CUDA_HOME
|
||||
|
||||
ROOT_DIR = os.path.dirname(__file__)
|
||||
|
||||
MAIN_CUDA_VERSION = "12.1"
|
||||
|
||||
# Supported NVIDIA GPU architectures.
|
||||
SUPPORTED_ARCHS = {"7.0", "7.5", "8.0", "8.6", "8.9", "9.0"}
|
||||
|
||||
@ -225,7 +227,7 @@ def get_path(*filepath) -> str:
|
||||
return os.path.join(ROOT_DIR, *filepath)
|
||||
|
||||
|
||||
def find_version(filepath: str):
|
||||
def find_version(filepath: str) -> str:
|
||||
"""Extract version information from the given filepath.
|
||||
|
||||
Adapted from https://github.com/ray-project/ray/blob/0b190ee1160eeca9796bc091e07eaebf4c85b511/python/setup.py
|
||||
@ -238,6 +240,15 @@ def find_version(filepath: str):
|
||||
raise RuntimeError("Unable to find version string.")
|
||||
|
||||
|
||||
def get_vllm_version() -> str:
|
||||
version = find_version(get_path("vllm", "__init__.py"))
|
||||
cuda_version = str(nvcc_cuda_version)
|
||||
if cuda_version != MAIN_CUDA_VERSION:
|
||||
cuda_version_str = cuda_version.replace(".", "")[:3]
|
||||
version += f"+cu{cuda_version_str}"
|
||||
return version
|
||||
|
||||
|
||||
def read_readme() -> str:
|
||||
"""Read the README file if present."""
|
||||
p = get_path("README.md")
|
||||
@ -256,7 +267,7 @@ def get_requirements() -> List[str]:
|
||||
|
||||
setuptools.setup(
|
||||
name="vllm",
|
||||
version=find_version(get_path("vllm", "__init__.py")),
|
||||
version=get_vllm_version(),
|
||||
author="vLLM Team",
|
||||
license="Apache 2.0",
|
||||
description=("A high-throughput and memory-efficient inference and "
|
||||
|
Reference in New Issue
Block a user