Xiao, Wang fd6a6658c3 Enable _int_mm on Intel GPU (#157769) ...

# Moativation

This PR is used to enable _int_mm on Intel GPU. And _int_mm is used by int8 quantization on torchao.

# Model Test Result:
We run meta-llama/Llama-3.1-8B-Instruct on Intel GPU and A100 using torchao int8-dynamic-quantization. The model configs as below:
Precision : torch.bfloat16
quantization configuration : Int8DynamicActivationInt8WeightConfig
dataset : wikitext

Result:
The perplexity values for Intel GPU and A100 are 9.582953453063965 and 9.57755184173584, respectively.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/157769
Approved by: https://github.com/EikanWang, https://github.com/desertfire

2025-08-02 05:16:01 +00:00

..

test_conv.py

Add device guard for xpu conv on multi device (#153067)

2025-05-09 09:41:51 +00:00

test_fusion.py

[Intel GPU] convolution fusion at XPU backend (#154202)

2025-05-28 03:14:18 +00:00

test_gemm.py

Enable _int_mm on Intel GPU (#157769)

2025-08-02 05:16:01 +00:00