[inductor] Make mix-order-reduction split size not depends on split-reduction heuristics#166461
Closed
shunting314 wants to merge 7 commits intogh/shunting314/251/basefrom
Closed
[inductor] Make mix-order-reduction split size not depends on split-reduction heuristics#166461shunting314 wants to merge 7 commits intogh/shunting314/251/basefrom
shunting314 wants to merge 7 commits intogh/shunting314/251/basefrom
Conversation
…lit-reduction heuristics [ghstack-poisoned]
This was referenced Oct 28, 2025
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/166461
Note: Links to docs will display an error until the docs builds have been completed. ❗ 1 Active SEVsThere are 1 currently active SEVs. If your PR is affected, please view them below: ✅ No FailuresAs of commit 349c998 with merge base b2a0f90 ( This comment was automatically generated by Dr. CI and updates every 15 minutes. |
…pends on split-reduction heuristics" cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov coconutruben [ghstack-poisoned]
shunting314
changed the title
… on split-reduction heuristics" split size is critical for mix order reduction perf while the one picked by split reduction heuristics can be very bad for mix order reduction. <img width="1197" height="596" alt="Screenshot 2025-10-27 at 11 17 16 PM" src="https://github.com/user-attachments/assets/7faa11ad-3a7a-4b29-90ed-e85fc01077ea" /> For the first shape in the chart, split reduction picks a split-size around 2000 and results in poor perf. It important to allow mix-order reduction decides split size itself. (ss_8 in the chart means split-size == 8) cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov coconutruben [ghstack-poisoned]
… on split-reduction heuristics" split size is critical for mix order reduction perf while the one picked by split reduction heuristics can be very bad for mix order reduction. <img width="1197" height="596" alt="Screenshot 2025-10-27 at 11 17 16 PM" src="https://github.com/user-attachments/assets/7faa11ad-3a7a-4b29-90ed-e85fc01077ea" /> For the first shape in the chart, split reduction picks a split-size around 2000 and results in poor perf. It important to allow mix-order reduction decides split size itself. (ss_8 in the chart means split-size == 8) cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov coconutruben [ghstack-poisoned]
… on split-reduction heuristics" split size is critical for mix order reduction perf while the one picked by split reduction heuristics can be very bad for mix order reduction. <img width="1197" height="596" alt="Screenshot 2025-10-27 at 11 17 16 PM" src="https://github.com/user-attachments/assets/7faa11ad-3a7a-4b29-90ed-e85fc01077ea" /> For the first shape in the chart, split reduction picks a split-size around 2000 and results in poor perf. It important to allow mix-order reduction decides split size itself. (ss_8 in the chart means split-size == 8) cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov coconutruben [ghstack-poisoned]
… on split-reduction heuristics" split size is critical for mix order reduction perf while the one picked by split reduction heuristics can be very bad for mix order reduction. <img width="1197" height="596" alt="Screenshot 2025-10-27 at 11 17 16 PM" src="https://github.com/user-attachments/assets/7faa11ad-3a7a-4b29-90ed-e85fc01077ea" /> For the first shape in the chart, split reduction picks a split-size around 2000 and results in poor perf. It important to allow mix-order reduction decides split size itself. (ss_8 in the chart means split-size == 8) cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov coconutruben [ghstack-poisoned]
jansel
approved these changes
Oct 30, 2025
This was referenced Oct 30, 2025
v0i0
approved these changes
Oct 31, 2025
… on split-reduction heuristics" split size is critical for mix order reduction perf while the one picked by split reduction heuristics can be very bad for mix order reduction. <img width="1197" height="596" alt="Screenshot 2025-10-27 at 11 17 16 PM" src="https://github.com/user-attachments/assets/7faa11ad-3a7a-4b29-90ed-e85fc01077ea" /> For the first shape in the chart, split reduction picks a split-size around 2000 and results in poor perf. It important to allow mix-order reduction decides split size itself. (ss_8 in the chart means split-size == 8) cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy chenyang78 kadeng muchulee8 amjames chauhang aakhundov coconutruben [ghstack-poisoned]
Collaborator
|
Starting merge as part of PR stack under #166697 |
1 similar comment
Collaborator
|
Starting merge as part of PR stack under #166697 |
pytorchmergebot
pushed a commit
that referenced
this pull request
Nov 1, 2025
Pull Request resolved: #166585 Approved by: https://github.com/jansel, https://github.com/PaulZhang12 ghstack dependencies: #166053, #166382, #166461
pytorchmergebot
pushed a commit
that referenced
this pull request
Nov 1, 2025
…ipts (#166697) It's nice to add a curve with a customized compilation options so that we can compare side-by-side the perf improvement of new features. E.g. for mix-order-reduction, by running the following command ``` python benchmarks/dynamo/genai_layers/benchmark.py --tolerance=1e-2 --exit-on-accuracy-failure --visualize rmsnorm_backward --custom-compile-name="compiled-no-fusion" --custom-compile-options='{"triton.mix_order_reduction":false}' ``` I get following output: ``` Geomean speedup for benchmark RMSNormBackward eager 11 data points compiled 11 data points, 15.82x speedup quack 11 data points, 15.45x speedup liger 11 data points, 14.06x speedup compiled-no-fusion 11 data points, 10.26x speedup ``` The output shows that the feature on average improve perf by `15.82 / 10.26 = 1.54x` for all the shapes tested. (I remove a shape (32768, 32768) whose rnumel is too large and not representative). The new curve also shows up in the figure: <img width="3564" height="2368" alt="RMSNormBackward_bench" src="https://github.com/user-attachments/assets/1ffac2bc-e726-4f1e-806d-e9e5de711492" /> Pull Request resolved: #166697 Approved by: https://github.com/BoyuanFeng ghstack dependencies: #166053, #166382, #166461, #166585, #166675
etaf
pushed a commit
to etaf/pytorch-inductor-xpu
that referenced
this pull request
Nov 4, 2025
…eduction heuristics (pytorch#166461) split size is critical for mix order reduction perf while the one picked by split reduction heuristics can be very bad for mix order reduction. <img width="1197" height="596" alt="Screenshot 2025-10-27 at 11 17 16 PM" src="https://github.com/user-attachments/assets/7faa11ad-3a7a-4b29-90ed-e85fc01077ea" /> For the first shape in the chart, split reduction picks a split-size around 2000 and results in poor perf. It important to allow mix-order reduction decides split size itself. (ss_8 in the chart means split-size == 8) Pull Request resolved: pytorch#166461 Approved by: https://github.com/jansel, https://github.com/v0i0 ghstack dependencies: pytorch#166053, pytorch#166382
etaf
pushed a commit
to etaf/pytorch-inductor-xpu
that referenced
this pull request
Nov 4, 2025
Pull Request resolved: pytorch#166585 Approved by: https://github.com/jansel, https://github.com/PaulZhang12 ghstack dependencies: pytorch#166053, pytorch#166382, pytorch#166461
etaf
pushed a commit
to etaf/pytorch-inductor-xpu
that referenced
this pull request
Nov 4, 2025
Pull Request resolved: pytorch#166675 Approved by: https://github.com/BoyuanFeng ghstack dependencies: pytorch#166053, pytorch#166382, pytorch#166461, pytorch#166585
etaf
pushed a commit
to etaf/pytorch-inductor-xpu
that referenced
this pull request
Nov 4, 2025
…ipts (pytorch#166697) It's nice to add a curve with a customized compilation options so that we can compare side-by-side the perf improvement of new features. E.g. for mix-order-reduction, by running the following command ``` python benchmarks/dynamo/genai_layers/benchmark.py --tolerance=1e-2 --exit-on-accuracy-failure --visualize rmsnorm_backward --custom-compile-name="compiled-no-fusion" --custom-compile-options='{"triton.mix_order_reduction":false}' ``` I get following output: ``` Geomean speedup for benchmark RMSNormBackward eager 11 data points compiled 11 data points, 15.82x speedup quack 11 data points, 15.45x speedup liger 11 data points, 14.06x speedup compiled-no-fusion 11 data points, 10.26x speedup ``` The output shows that the feature on average improve perf by `15.82 / 10.26 = 1.54x` for all the shapes tested. (I remove a shape (32768, 32768) whose rnumel is too large and not representative). The new curve also shows up in the figure: <img width="3564" height="2368" alt="RMSNormBackward_bench" src="https://github.com/user-attachments/assets/1ffac2bc-e726-4f1e-806d-e9e5de711492" /> Pull Request resolved: pytorch#166697 Approved by: https://github.com/BoyuanFeng ghstack dependencies: pytorch#166053, pytorch#166382, pytorch#166461, pytorch#166585, pytorch#166675
Khanaksahu
pushed a commit
to Khanaksahu/pytorch
that referenced
this pull request
Nov 17, 2025
…eduction heuristics ghstack-source-id: 57d7f0e Pull Request resolved: pytorch/pytorch#166461
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
4 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.
Stack from ghstack (oldest at bottom):
split size is critical for mix order reduction perf while the one picked by split reduction heuristics can be very bad for mix order reduction.
For the first shape in the chart, split reduction picks a split-size around 2000 and results in poor perf. It important to allow mix-order reduction decides split size itself. (ss_8 in the chart means split-size == 8)
cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @chenyang78 @kadeng @muchulee8 @amjames @chauhang @aakhundov @coconutruben