Use 256-thread blocks for jagged dense-output kernel (#5848)#5848
Open
MericGit wants to merge 1 commit into
Open
Use 256-thread blocks for jagged dense-output kernel (#5848)#5848MericGit wants to merge 1 commit into
MericGit wants to merge 1 commit into
Conversation
Contributor
|
@MericGit has exported this pull request. If you are a Meta employee, you can view the originating Diff in D107571746. |
Summary: X-link: facebookresearch/FBGEMM#2766 Initial Rocm profiler thread tracing shows pretty poor utilization for jagged_1d_to_dense. Prev was setting it to use 16 threads only which is only 25% util of a wavefront when D=1 for 1d. For 2d (large D) values it seems OK to just consistently use 256. Alternative could be coding special path for just the D=1 (1d case). Open to suggestions. Generally this will shift them from using 1024 threads to 256. These changes impacts jagged_1d_to_dense, jagged_2d_to_dense, and jagged_to_padded_dense_forward. Mostly focused on jagged_1d_to_dense. Differential Revision: D107571746
meta-codesync
Bot
changed the title
MericGit
force-pushed
the
export-D107571746
branch
from
97c857b to
2256ee2
Compare
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary:
X-link: https://github.com/facebookresearch/FBGEMM/pull/2766
Initial Rocm profiler thread tracing shows pretty poor utilization for jagged_1d_to_dense.
Prev was setting it to use 16 threads only which is only 25% util of a wavefront when D=1 for 1d.
For 2d (large D) values it seems OK to just consistently use 256. Alternative could be coding special path for just the D=1 (1d case). Open to suggestions. Generally this will shift them from using 1024 threads to 256.
These changes impacts jagged_1d_to_dense, jagged_2d_to_dense, and jagged_to_padded_dense_forward. Mostly focused on jagged_1d_to_dense.
Differential Revision: D107571746