Intel XPU CI Enabling

[liangan1]

Motivation

To improve quality on Intel XPU device, we plan to enable the CI/CD process on Intel XPUs. The CI is based on the docker env and we will firstly enable the CI infra and then enable the UTs feature by feature.

Attention: XPU CI is only triggered by "ciflow/xpu" label and the test folders or test_xxx file should be included by the .github/scripts/ci_test_xpu.sh

Status

The common dependency, e.g., CI infra(#3027) and test folders(#3335) has been landed. The feature will be enabled one by one. Community owner suggests us to make the PR as small as possible(Not change more than 2 files per PR) to reduce the review effort.

Task List

@zxd1997066 is the owner of this task list.

• [CI]Enable CI for TorchAO on XPU Using Docker #3027
• [Intel XPU] Add common code for UTs on XPU #3335
• [xpu][test] Port 2 test/quantization_{gptq, quant_primitive} UT files to intel XPU #3350
• [xpu][test] Port 2 test/quantization_{qat, quant_api} UT files to intel XPU  #3351
• [Intel XPU] Enable test/quantization UTs on XPU  #3249
• [xpu][test] Make tests in test_float8_tensor.py device agnostic #3332
• [Intel GPU] Extend TestQAT module with xpu testcases #3177
• Xpu ut/dtypes #2797
• Support mx_tensor and enable it's test on Intel GPU #3248
• Xpu ut/float8 #2799
• xpu ut enbaling: test_integration.py #2800
• Xpu ut/prototype #2801
• xpu ut enablinng: moe_training #2802
• xpu ut enabling: test_ao_models #2803