[ROCm] Enable Einsum for inferencing perf#12360
Merged
Merged
Conversation
added 2 commits
July 28, 2022 21:53
zhangyaobit
approved these changes
Jul 29, 2022
Contributor
Author
|
Thanks. |
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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: The existing Einsum kernel is disable for ROCm. This causes 1P model inferencing fall back to CPUExecutionProvider. In addition, additional Cast OPs and memcpy (D2H/H2D) are inserted, particularly in fp16 mode, before and after Einsum op. The whole e2e thus much slower than running in CUDA env.
Perf - before the change:
Latency(ms) Throughput(QPS)
49.47 646.86
After the change ~ 3x faster
Latency(ms) Throughput(QPS)
18.06 1772.09
Also verified using the rocm profiler - #10911