-
Notifications
You must be signed in to change notification settings - Fork 53
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix matmul incorrect results when k dim for CTA tile is a multiple of 16 #3616
Draft
rdspring1
wants to merge
10
commits into
main
Choose a base branch
from
hopper_matmul_cta_k_fix
base: main
Could not load branches
Branch not found: {{ refName }}
Loading
Could not load tags
Nothing to show
Loading
Are you sure you want to change the base?
Some commits from the old base branch may be removed from the timeline,
and old review comments may become outdated.
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
!test |
jacobhinkle
reviewed
Dec 19, 2024
tests/cpp/test_matmul.cpp
Outdated
Comment on lines
4036
to
4037
// NOTE Certain combinations of cta k dimension and circular buffer | ||
// prefetching can get incorrect results. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
👀
|
||
mparams.supported_vec_size = {8, 8, 4}; | ||
mparams.supported_vec_size = {8, 8, 8}; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Good catch, although I think it currently has no meaning until we start handling epilogue inputs with supported vec size.
!test |
rdspring1
force-pushed
the
hopper_matmul_cta_k_fix
branch
3 times, most recently
from
December 20, 2024 23:36
952cb0a
to
0e213d6
Compare
rdspring1
force-pushed
the
hopper_matmul_cta_k_fix
branch
from
December 20, 2024 23:56
d366352
to
0c784e0
Compare
!test |
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.
This PR fixes the incorrect results issue when k dimension for CTA tile is a multiple of
getK(mma_macro)
.Why?
scheduleMmaResults
, we need to split thek
reduction bygetK(mma_macro)
. A serial reduction will add the results fromwgmma
along k-dimension.Details
transformLikeMmaOutput
function to not be used inscheduleMmaResults
.HSH_TN_UseScheduler
testPerformance Results
HSH_TN_UseScheduler - 1.038 - Basically a tie
CTA tile is
M=64, K=256, K=32
HSH_NT_UseScheduler - 0.9155
CTA tile is
M=64, K=256, K=32