Skip to content

[UR] Bump UR tag to 04799e7 to get OpenCL adapter fixes #11806

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged

Conversation

callumfare
Copy link
Contributor

@callumfare callumfare commented Nov 7, 2023

@callumfare callumfare force-pushed the callum/fix_opencl_mem_buffer_create branch 2 times, most recently from bef5b0d to d165d85 Compare November 7, 2023 16:26
Copy link
Contributor

@kbenzie kbenzie left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Now that oneapi-src/unified-runtime#1048 is merged, please pull in the latest sycl branch changes and update UNIFIED_RUNTIME_REPO & UNIFIED_RUNTIME_TAG as suggestion. Then make this ready for review.

@callumfare callumfare force-pushed the callum/fix_opencl_mem_buffer_create branch from 26d688a to 6e33e40 Compare November 20, 2023 10:36
@callumfare callumfare changed the title [UR] Update UR commit to fix OpenCL memory leak [UR] Bump UR tag to 04799e7 to get OpenCL adapter fixes Nov 20, 2023
@callumfare callumfare marked this pull request as ready for review November 20, 2023 10:38
@callumfare callumfare requested a review from a team as a code owner November 20, 2023 10:38
@kbenzie
Copy link
Contributor

kbenzie commented Nov 21, 2023

@intel/llvm-gatekeepers please merge.

@steffenlarsen steffenlarsen merged commit c67b6d7 into intel:sycl Nov 21, 2023
@againull
Copy link
Contributor

@callumfare SYCL :: Basic/get_backend.cpp test is failing in post-commit after merging this PR. Could you please take a look?
Details:
https://github.com/intel/llvm/actions/runs/6943189569/job/18888237029


Failed Tests (1):
SYCL :: Basic/get_backend.cpp

2023-11-21T11:51:16.5243903Z FAIL: SYCL :: Basic/get_backend.cpp (191 of 1751)
2023-11-21T11:51:16.5245036Z ******************** TEST 'SYCL :: Basic/get_backend.cpp' FAILED ********************
2023-11-21T11:51:16.5245705Z Exit Code: -6
2023-11-21T11:51:16.5280748Z
2023-11-21T11:51:16.5281189Z Command Output (stdout):
2023-11-21T11:51:16.5281925Z --
2023-11-21T11:51:16.5282275Z # RUN: at line 1
2023-11-21T11:51:16.5284128Z /__w/llvm/llvm/toolchain/bin//clang++ -fsycl -fsycl-targets=spir64 /__w/llvm/llvm/llvm/sycl/test-e2e/Basic/get_backend.cpp -o /__w/llvm/llvm/build-e2e/Basic/Output/get_backend.cpp.tmp.out
2023-11-21T11:51:16.5287393Z # executed command: /__w/llvm/llvm/toolchain/bin//clang++ -fsycl -fsycl-targets=spir64 /__w/llvm/llvm/llvm/sycl/test-e2e/Basic/get_backend.cpp -o /__w/llvm/llvm/build-e2e/Basic/Output/get_backend.cpp.tmp.out
2023-11-21T11:51:16.5289341Z # note: command had no output on stdout or stderr
2023-11-21T11:51:16.5289976Z # RUN: at line 2
2023-11-21T11:51:16.5680409Z /__w/llvm/llvm/build-e2e/Basic/Output/get_backend.cpp.tmp.out
2023-11-21T11:51:16.5681568Z # executed command: /__w/llvm/llvm/build-e2e/Basic/Output/get_backend.cpp.tmp.out
2023-11-21T11:51:16.5682433Z # .---command stderr------------
2023-11-21T11:51:16.5683916Z # | terminate called after throwing an instance of 'sycl::_V1::runtime_error'
2023-11-21T11:51:16.5685771Z # | what(): Native API failed. Native API returns: -999 (Unknown PI error) -999 (Unknown PI error)
2023-11-21T11:51:16.5686858Z # `-----------------------------
2023-11-21T11:51:16.5880114Z # error: command failed with exit status: -6
2023-11-21T11:51:16.5960188Z
2023-11-21T11:51:16.6003733Z --
2023-11-21T11:51:16.6078884Z

@kbenzie
Copy link
Contributor

kbenzie commented Nov 21, 2023

@againull @callumfare has just finished for the day, we can look into this tomorrow.

@kbenzie
Copy link
Contributor

kbenzie commented Nov 21, 2023

I notice that the failing test is running on Level Zero not OpenCL. The UR changes this PR introduces are limited to only the OpenCL adapter as can be seen here.

I checked back 4 days of SYCL Post Commit runs and it is odd that Basic/get_backend.cpp never showed up as a failure until this PR but the e2e-lin Intel Arc A-Series Graphics with Level Zero workflow has been unstable that whole time.

I've been unable to reproduce this locally although I have an 8th gen iGPU rather than an Arc card. We can try to access another system to debug further tomorrow.

@callumfare
Copy link
Contributor Author

callumfare commented Nov 22, 2023

I've not been able to reproduce the failure on an Arc A770, although I did see other E2E tests failing:

Failed Tests (11):
  SYCL :: DeviceLib/imf_fp32_rounding_test.cpp
  SYCL :: DeviceLib/string_test.cpp
  SYCL :: ESIMD/preemption.cpp
  SYCL :: KernelCompiler/kernel_compiler.cpp
  SYCL :: KernelCompiler/opencl_capabilities.cpp
  SYCL :: Plugin/level_zero_usm_residency.cpp
  SYCL :: Tracing/usm/queue_copy_released_pointer.cpp
  SYCL :: Tracing/usm/queue_single_task_nullptr.cpp
  SYCL :: Tracing/usm/queue_single_task_released_pointer.cpp
  SYCL :: USM/usm_leak_check.cpp
  SYCL :: XPTI/mem_alloc_events_linux.cpp

I suspect those might be related to the driver version.

callumfare added a commit to kbenzie/intel-llvm that referenced this pull request Dec 18, 2023
Includes UR changes from [ [OpenCL] Fix memory leak and coverity issue
with struct-to-array casts intel#1048
](oneapi-src/unified-runtime#1048).
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants