[RNG] Split up some tests to prevent out-of-memory and device_lost crashes #1923
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
egrabovskaya
requested review from
aelizaro,
andreyfe1,
ElenaTyuleneva and
paveldyakov
aelizaro
reviewed
Oct 29, 2024
test/xpu_api/random/statistics_tests/exponential_distribution_dp_test.pass.cpp
Outdated
Show resolved
Hide resolved
ElenaTyuleneva
previously approved these changes
Oct 30, 2024
egrabovskaya
force-pushed
the
dev/egrabovs/test_failures_fix
branch
from
34d761e
to
b7236d8
Compare
|
I removed extra code from geometric_distribution_test.pass as well. |
egrabovskaya
force-pushed
the
dev/egrabovs/test_failures_fix
branch
from
edb6faa
to
964641b
Compare
aelizaro
approved these changes
Nov 5, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Here is the clarification from Matthew and a driver team member:
"This issue showed up as a regression after correcting the thread-EU ratio reported for BMG & LNL GTs. It directly affects the code which is used to calculate the private memory size needed by the kernels.
The ratio was increased from 8 to 16 to fit actual hardware, and by doing so a few SYCL CTS test failures were fixed (since they now have enough private memory for their kernels to use)."
It looks like kernels are reserved more scratch memory with new drivers leading to out-of-memory crashes when a large number of kernels are compiled and cached on the device in the tests.