Skip to content

[SYCL][NFC] Optimize int-convert.cpp test #12043

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

Conversation

AlexeySachkov
Copy link
Contributor

The test sporadically timed out on Windows because default implicit context is disabled in there.

Due to the fact that separate queue object was create for every kernel, it caused full re-JIT of the whole device code.

Refactored the test to use a single queue object explicitly to have context and all associated caches to be shared between all kernels in the test.

This resolves #12011

The test sporadically timed out on Windows because default implicit
context is disabled in there.

Due to the fact that separate `queue` object was create for every
kernel, it caused full re-JIT of the whole device code.

Refactored the test to use a single queue object explicitly to have
context and all associated caches to be shared between all kernels in
the test.

This resolves intel#12011
@AlexeySachkov AlexeySachkov requested a review from a team as a code owner November 30, 2023 15:39
@AlexeySachkov AlexeySachkov changed the title [SYCL] Optimize int-convert.cpp test [SYCL][NFC] Optimize int-convert.cpp test Nov 30, 2023
@AlexeySachkov
Copy link
Contributor Author

CI on CUDA failed because machine lost connection to the server - I will proceed with the merge, because the change is trivial enough for that.

As an indication that the test is actually quick on Windows now you can see logs from e2e tests:

Slowest Tests:
--------------------------------------------------------------------------
171.28s: SYCL :: ESIMD/unified_memory_api/atomic_update_acc.cpp
151.17s: SYCL :: ESIMD/api/functional/ctors/ctor_load_acc_core.cpp
144.70s: SYCL :: ESIMD/regression/tanh_fix_test.cpp

Comparing to another (randomly chosen) run of Windows e2e tests in pre-commit:

Slowest Tests:
--------------------------------------------------------------------------
780.84s: SYCL :: Basic/vector/int-convert.cpp
174.57s: SYCL :: ESIMD/unified_memory_api/atomic_update_acc.cpp
143.07s: SYCL :: Basic/sycl_2020_images/host_task_sampled_image_read_linear.c

@AlexeySachkov AlexeySachkov merged commit a37c661 into intel:sycl Nov 30, 2023
@AlexeySachkov AlexeySachkov deleted the private/asachkov/optimize-int-convert-test branch December 1, 2023 13:45
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.

Basic/vector/int-convert.cpp' FAILED on unrelated changes
2 participants