Implement a workaround to gemm issue in OneMKL #2082
Merged
Conversation
This file contains hidden or 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
|
View rendered docs @ https://intelpython.github.io/dpnp/index.html |
antonwolfy
force-pushed
the
impl-w/a-to-gemm-on-lnl-arl
branch
from
7311ae3 to
e59f9f7
Compare
Co-authored-by: Oleksandr Pavlyk <[email protected]>
…ture_is() isn't marked as const
antonwolfy
requested review from
npolina4,
vlad-perevezentsev and
vtavana
as code owners
oleksandr-pavlyk
approved these changes
Oct 2, 2024
vtavana
approved these changes
Oct 2, 2024
antonwolfy
added a commit
that referenced
this pull request
Oct 2, 2024
* Implement a workaround to gemm issue in OneMKL * Fix codespell issue * Enable w/a also for float dtype * Add Battlemage G21 arhitecture to w/a * Disable w/a for Arrow Lake * Remove Lunar Lake architecture from the w/a * Applied the pre-commit hooks * Update dpnp/backend/extensions/blas/gemm.hpp Co-authored-by: Oleksandr Pavlyk <[email protected]> * Applied pre-commit black hook * Add more clarification to the comment * Remove excess semicolon * Removed const keyword from review comment because ext_oneapi_architecture_is() isn't marked as const * Applied review comment * Updated the changelog --------- Co-authored-by: Oleksandr Pavlyk <[email protected]>
github-actions bot
added a commit
that referenced
this pull request
Oct 2, 2024
* Implement a workaround to gemm issue in OneMKL * Fix codespell issue * Enable w/a also for float dtype * Add Battlemage G21 arhitecture to w/a * Disable w/a for Arrow Lake * Remove Lunar Lake architecture from the w/a * Applied the pre-commit hooks * Update dpnp/backend/extensions/blas/gemm.hpp Co-authored-by: Oleksandr Pavlyk <[email protected]> * Applied pre-commit black hook * Add more clarification to the comment * Remove excess semicolon * Removed const keyword from review comment because ext_oneapi_architecture_is() isn't marked as const * Applied review comment * Updated the changelog --------- Co-authored-by: Oleksandr Pavlyk <[email protected]> 178342c
antonwolfy
added a commit
that referenced
this pull request
Oct 2, 2024
* Implement a workaround to gemm issue in OneMKL * Fix codespell issue * Enable w/a also for float dtype * Add Battlemage G21 arhitecture to w/a * Disable w/a for Arrow Lake * Remove Lunar Lake architecture from the w/a * Applied the pre-commit hooks * Update dpnp/backend/extensions/blas/gemm.hpp * Applied pre-commit black hook * Add more clarification to the comment * Remove excess semicolon * Removed const keyword from review comment because ext_oneapi_architecture_is() isn't marked as const * Applied review comment * Updated the changelog --------- Co-authored-by: Oleksandr Pavlyk <[email protected]>
6 tasks
antonwolfy
added a commit
that referenced
this pull request
Oct 11, 2024
This reverts commit 178342c.
6 tasks
antonwolfy
added a commit
that referenced
this pull request
Oct 14, 2024
antonwolfy
added a commit
that referenced
this pull request
Oct 14, 2024
antonwolfy
added a commit
that referenced
this pull request
Oct 14, 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.
The PR is about to implement a workaround to issue with incorrect result from
gemmfunction of OneMKL when running on Lunar Lake or Arrow Lake Battlemage graphics architectures.The PR proposes to enable the w/a for small size of input arrays with non-zero offset where the issue was initially identified by tests for eig/eigh functions.
It's totally unclear for what sizes of input array the w/a needs to be applied, so it proposes to set to close values used in failing tests.