[Flow] Enable softmax-like fusion under aggressive fusion. #17747
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
MaheshRavishankar
added
benchmarks:cuda
|
Oops. Sorry push the branch to upstream repo instead of my fork and broke the naming convention. Will delete branch after it lands. |
Under aggressive fusion, drop the restriction of consumer iteration space being same dimensionality as the producer iteration space. Typically this can lead to large vectors if not handled properly. So this is guarded under `--iree-flow-enable-aggressive-fusion` flag. Signed-off-by: MaheshRavishankar <[email protected]>
MaheshRavishankar
force-pushed
the
sdxl_quantized_enable_softmax_fusion
branch
from
10c3216 to
a62dd5a
Compare
Abbreviated Benchmark Summary@ commit b8a2701f5d91366e7318dcfdb76cdba464bab8d3 (vs. base 4294a5b0ebaec6dcca483bf16f5918108b09ea0a) Data-Tiling Comparison TableClick to show
Regressed Latencies 🚩
Improved Latencies 🎉
[Top 3 out of 7 results showed] No improved or regressed compilation metrics 🏖️ For more information: |
hanhanW
approved these changes
Jun 28, 2024
| @@ -1,4 +1,4 @@ | |||
| // RUN: iree-opt --pass-pipeline="builtin.module(util.func(iree-flow-form-dispatch-regions))" --split-input-file %s | FileCheck %s | |||
| // RUN: iree-opt --pass-pipeline="builtin.module(util.func(iree-flow-form-dispatch-regions{aggressive-fusion=true}))" --split-input-file %s | FileCheck %s | |||
There was a problem hiding this comment.
Are we going to miss the tests for default path? Do we add a new mlir file to test aggressive-fusion=true? I don't suggest to add an additional FileCheck in this file because some of tests seem redundant to the aggressive fusion.
|
oh, my bad.. I did not notice that there is auto-merge.. |
LLITCHEV
pushed a commit
to LLITCHEV/iree
that referenced
this pull request
Jul 30, 2024
…17747) Under aggressive fusion, drop the restriction of consumer iteration space being same dimensionality as the producer iteration space. Typically this can lead to large vectors if not handled properly. So this is guarded under `--iree-flow-enable-aggressive-fusion` flag. Fixes nod-ai/SHARK-ModelDev#749 Signed-off-by: MaheshRavishankar <[email protected]> Signed-off-by: Lubo Litchev <[email protected]>
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.
Under aggressive fusion, drop the restriction of consumer iteration space being same dimensionality as the producer iteration space. Typically this can lead to large vectors if not handled properly. So this is guarded under
--iree-flow-enable-aggressive-fusionflag.Fixes nod-ai/SHARK-ModelDev#749