[ESIMD] Fix atomic_update() implementation for N=16 and N=32 on Gen12 #12722
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
v-klochkov
force-pushed
the
esimd_fix_atomic_update_gen12
branch
from
a626b8e to
9c3515a
Compare
v-klochkov
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
v-klochkov
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
atomic_update() for USM and ACC were lowered to SVM/DWORD atomic intrinsics even though the HW instructions on Gen12 supported only N up to 8 for USM and up to 16 for ACC. GPU had legalization pass for N that split longer vectors to smaller and available in HW. That GPU optimization/legalization workes incorrectly for USM as it splits longer vectors assuming instruction is available for N=16 in case of USM, which is not correct. The patch here implements splitting of N=16 and N=32 cases for atomic_update(usm, ...) to N=8 vectors until GPU fixes the legalization for USM atomic_update. Signed-off-by: Klochkov, Vyacheslav N <[email protected]>
v-klochkov
force-pushed
the
esimd_fix_atomic_update_gen12
branch
from
9c3515a to
732d63e
Compare
v-klochkov
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
v-klochkov
temporarily deployed
to
WindowsCILock
GitHub Actions
Inactive
fineg74
approved these changes
Feb 15, 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Outdated suggestions cannot be applied.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.
atomic_update() for USM and ACC N=16,32 were lowered to SVM/DWORD atomic
intrinsics even though the HW instructions on Gen12 supported only
N up to 8 for USM and up to 16 for ACC.
GPU had legalization pass for N that split longer vectors to smaller and available in HW.
That GPU optimization/legalization workes incorrectly for USM as it
splits longer vectors assuming instruction is available for N=16 in case
of USM, which is not correct.
The patch here implements splitting of N=16 and N=32 cases for
atomic_update(usm, ...) to N=8 vectors until GPU fixes the legalization
for USM atomic_update.