Optimize work_group reduce for GroupNorm Forward #1688
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
xytintel
reviewed
May 27, 2025
|
|
||
| [[intel::reqd_sub_group_size(SIMD)]] void operator()( | ||
| sycl::nd_item<1> item) const { | ||
| const int64_t i = item.get_group(0); | ||
| WelfordOp welford_op = {/*correction=*/0, /*take_sqrt=*/false, item}; | ||
| WelfordOp welford_op = {/*correction=*/0, /*take_sqrt=*/false}; |
There was a problem hiding this comment.
Where is the nd_item object?
There was a problem hiding this comment.
WelfordOpXPU is replaced by WelfordOp, which does not need item as a pamameter.
xytintel
approved these changes
May 30, 2025
|
@xytintel , @yucai-intel , may I know which workload drives such an optimization? |
|
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Outdated suggestions cannot be applied.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.
Optimized the calculation of data on shared memory in GroupReduceWithoutBroadcast, so that the performance of GroupNorm Forward when shape=(1024, 128, 32, 32) is improved by 20+% (latency 2700+ms->~2000ms).
