Add new prop to _XpuDevicePropertie for triton gemm optimization

[guangyey]

Stack from ghstack (oldest at bottom):

• -> Add new prop to _XpuDevicePropertie for triton gemm optimization #131738

Motivation

This PR aims to add new properties to _XpuDevicePropertie for triton gemm optimization.

Additional Context

ext_oneapi_supports_cl_extension is not a ABI-neutral API. It depends on compiler 2025.0. For more details, see intel/llvm#13212

cc @jgong5 @mingfeima @XiaobingSuper @sanchitintel @ashokei @jingxu10

[pytorch-bot]

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/131738

• 📄 Preview Python docs built from this PR
• 📄 Preview C++ docs built from this PR
• ❓ Need help or want to give feedback on the CI? Visit the bot commands wiki or our office hours

Note: Links to docs will display an error until the docs builds have been completed.

✅ You can merge normally! (1 Unrelated Failure)

As of commit bab9db0 with merge base 8ea5b57 ():

BROKEN TRUNK - The following job failed but were present on the merge base:

👉 Rebase onto the `viable/strict` branch to avoid these failures

• pull / linux-jammy-py3.8-gcc11 / test (distributed, 1, 2, amz2023.linux.2xlarge) (gh) (trunk failure)
  RuntimeError: distributed/flight_recorder/test_fr_analysis 1/1 failed!

This comment was automatically generated by Dr. CI and updates every 15 minutes.

[guangyey]

@pytorchbot rebase

[pytorchmergebot]

@pytorchbot started a rebase job onto refs/remotes/origin/viable/strict. Check the current status here

[pytorchmergebot]

Successfully rebased gh/guangyey/55/orig onto refs/remotes/origin/viable/strict, please pull locally before adding more changes (for example, via ghstack checkout https://github.com/pytorch/pytorch/pull/131738)

[ZzEeKkAa]

Suggested change

	device_prop->has_##member = \
	raw_device.ext_oneapi_supports_cl_extension("cl_intel_" #member);
	sycl::ext::oneapi::experimental::cl_version version_##member; \
	device_prop->has_##member = raw_device.ext_oneapi_supports_cl_extension( \
	"cl_intel_" #member, &version_##member);

The compilation does not fail, however I'm getting core dump without second argument (PTDB 0.5.3.27, python 3.9):

$ python -c "import torch; print(torch.xpu.get_device_capability())"
[1]    130676 segmentation fault (core dumped)  python -c "import torch; print(torch.xpu.get_device_capability())"

[ZzEeKkAa]

This is the backtrace of the problem:

Thread 1.1 "pt_main_thread" received signal SIGSEGV, Segmentation fault.
(gdb) bt
#0  0x00007fffd185d78d in sycl::_V1::ext::oneapi::experimental::detail::OpenCLC_Supports_Extension(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, sycl::_V1::ext::oneapi::experimental::cl_version*, unsigned int) () from /opt/intel/oneapi/pytorch-gpu-dev-0.5/lib/libsycl.so.7
#1  0x00007fffd1946fd7 in sycl::_V1::device::ext_oneapi_supports_cl_extension(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > const&, sycl::_V1::ext::oneapi::experimental::cl_version*) const () from /opt/intel/oneapi/pytorch-gpu-dev-0.5/lib/libsycl.so.7
#2  0x00007ffff72f5bdd in c10::xpu::(anonymous namespace)::initDeviceProperties (device_prop=0x9c476d0, device=0) at /home/yevhenii/Projects/pytorch/c10/xpu/XPUFunctions.cpp:115
#3  0x00007ffff72f609b in c10::xpu::get_device_properties (device_prop=0x9c476d0, device=0 '\000') at /home/yevhenii/Projects/pytorch/c10/xpu/XPUFunctions.cpp:158

[guangyey]

Thanks for your suggestion. I will check it.

[guangyey]

@ZzEeKkAa CI is green. CI use the following version, could you try again on your local machine?

[ZzEeKkAa]

I'm using PTDB 0.5.2.18 with updated ubuntu 22.04 and python 3.9 venv installed from deadsnakes/ppa. Just reproduced the issue.

icpx -v
Intel(R) oneAPI DPC++/C++ Compiler 2024.1.3 (2024.1.3.20240604)
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /opt/intel/oneapi/compiler/2024.1/bin/compiler
Configuration file: /opt/intel/oneapi/compiler/2024.1/bin/compiler/../icpx.cfg
Found candidate GCC installation: /usr/lib/gcc/x86_64-linux-gnu/11
Found candidate GCC installation: /usr/lib/gcc/x86_64-linux-gnu/12
Selected GCC installation: /usr/lib/gcc/x86_64-linux-gnu/11
Candidate multilib: .;@m64
Selected multilib: .;@m64

[guangyey]

OK, thanks. The compiler team has confirmed this is a bug.

[guangyey]

I will merge this PR once our CI upgrades to the appropriate compiler.

[ZzEeKkAa]

Thank you!

[guangyey]

No problem!

[guangyey]

@pytorchbot rebase -b main

[pytorchmergebot]

@pytorchbot started a rebase job onto refs/remotes/origin/main. Check the current status here

[pytorchmergebot]

Successfully rebased gh/guangyey/55/orig onto refs/remotes/origin/main, please pull locally before adding more changes (for example, via ghstack checkout https://github.com/pytorch/pytorch/pull/131738)

[guangyey]

@pytorchbot merge

[guangyey]

@ZzEeKkAa bundle 0.5.3 has been released. Let us merge this PR.

[pytorchmergebot]

Merge failed

Reason: This PR needs a release notes: label
If your changes are user facing and intended to be a part of release notes, please use a label starting with release notes:.

If not, please add the topic: not user facing label.

To add a label, you can comment to pytorchbot, for example
@pytorchbot label "topic: not user facing"

For more information, see
https://github.com/pytorch/pytorch/wiki/PyTorch-AutoLabel-Bot#why-categorize-for-release-notes-and-how-does-it-work.

Details for Dev Infra team

Raised by workflow job

[guangyey]

@pytorchbot merge

[pytorchmergebot]

Merge started

Your change will be merged once all checks pass (ETA 0-4 Hours).

Learn more about merging in the wiki.

Questions? Feedback? Please reach out to the PyTorch DevX Team

Advanced Debugging

Check the merge workflow status
here