cpu: aarch64: batch_normalization : Expand ARM SVE support in jit_uni_batch_normalization #1918
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
cpu: aarch64: Expand ARM SVE support in jit_uni_batch_normalization Added sve_256 in the implementation list Update jit_uni_batch_normalization.cpp Updated the block size definition to accommodate different ISAs. Added 'OR' conditions to extend support for additonal block_size. Predicate registers are set according to isa vector length. ldr and str instruction changed to ldw1 and stw1 respectively. To support load and store operations as per ISA.
mgouicem
added
the
platform:cpu-aarch64
jondea
reviewed
May 17, 2024
There was a problem hiding this comment.
Thank you for the contribution, we really appreciate your work on vlen 256. I just have a few comments about whether it could be more general. If it works for an SVE vlen of 256 and 512, why not any length? Then you could remove any hard coded vlen variables, and there could be a single implementation.
jondea
approved these changes
May 20, 2024
|
@vpirogov , Hello |
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Outdated suggestions cannot be applied.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.
Description
This commit enhances the existing ARM SVE support in jit_uni_batch_normalization to include additional vector length compatibility. The changes made are for implementation of different ARM SVE vector length.
Major Code changes:
Updated the block size definition to accommodate other SVE vector length.
Added 'OR' and 'AND' conditions to extend support for other SVE vector.
Predicate registers are set according to isa vector length.
ldr and str instructions are replaced by ld1w and st1w respectively. For appropriate load and strore operation considering isa vector length.
Checklist
General
[✓] Do all unit and benchdnn tests (make test and make test_benchdnn_*) pass locally for each commit? Yes
Test output is same with and without this commit.
make test output:
99% tests passed, 1 tests failed out of 103
Total Test time (real) = 36.37 sec
The following tests FAILED:
102 - test_graph_unit_cpu (Failed)
Errors while running CTest
Output from these tests are in: /home/nikhil/TEST/oneDNN/build/Testing/Temporary/LastTest.log
Use "--rerun-failed --output-on-failure" to re-run the failed cases verbosely.
make: *** [Makefile:71: test] Error 8
make test_benchdnn_* output:
make: *** No rule to make target 'test_benchdnn_*'. Stop.
make test_benchdnn_bnorm_ci_cpu/fast output.
tests:4445 passed:1232 skipped:2989 mistrusted:224 unimplemented:0 invalid_arguments:0 failed:0 listed:0