[Backend] Allow layout propagation through TransOp. #3316
Merged
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
|
can we add a simple lit test that does forward propagation: that should move the convert after |
|
Absolutely, I intended to add a test but wanted to check that the tests were passing first. Looks like they are, so yay. :) |
jlebar
force-pushed
the
dev-jlebar/trans-op
branch
from
b0be26e to
47d1dca
Compare
jlebar
force-pushed
the
dev-jlebar/trans-op
branch
from
47d1dca to
663662a
Compare
jlebar
force-pushed
the
dev-jlebar/propagate-dot-wait
branch
2 times, most recently
from
b2cae8a to
5fd0d0c
Compare
jlebar
force-pushed
the
dev-jlebar/trans-op
branch
2 times, most recently
from
ff2cbb6 to
4462f94
Compare
|
Added a test; phal, Thomas. |
ThomasRaoux
reviewed
Mar 8, 2024
This is a revival of #3097, which I now realize I need for some int4 matmul work I'm doing. Previously this caused performance problems / running out of shmem on A100, but it seems that #3261 (or perhaps something else) fixed it. This now has no performance delta on our internal performance tests on A100 or H100. GPC: trans-op
jlebar
force-pushed
the
dev-jlebar/trans-op
branch
from
4462f94 to
6aeae0a
Compare
|
Thanks, Thomas. PHAL. |
htyu
pushed a commit
to htyu/triton
that referenced
this pull request
Mar 20, 2024
[NVGPU] Allow layout propagation through TransOp. This is a revival of triton-lang#3097, which I now realize I need for some int4 matmul work I'm doing. Previously this caused performance problems / running out of shmem on A100, but it seems that triton-lang#3261 (or perhaps something else) fixed it. This now has no performance delta on our internal performance tests on A100 or H100.
karupayun
pushed a commit
to openxla/triton
that referenced
this pull request
Apr 3, 2024
[NVGPU] Allow layout propagation through TransOp. This is a revival of triton-lang#3097, which I now realize I need for some int4 matmul work I'm doing. Previously this caused performance problems / running out of shmem on A100, but it seems that triton-lang#3261 (or perhaps something else) fixed it. This now has no performance delta on our internal performance tests on A100 or H100.
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.
[NVGPU] Allow layout propagation through TransOp.
This is a revival of #3097, which I now
realize I need for some int4 matmul work I'm doing.
Previously this caused performance problems / running out of shmem on A100, but
it seems that #3261 (or perhaps something
else) fixed it. This now has no performance delta on our internal performance
tests on A100 or H100.
PR chain