[inductor] fix a device sync issue for benchmarking fusion #135531
Closed
Conversation
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
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/135531
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (5 Unrelated Failures)As of commit e5e1d1d with merge base cbc6b30 ( BROKEN TRUNK - The following jobs failed but were present on the merge base:👉 Rebase onto the `viable/strict` branch to avoid these failures
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
Fix #134768 . When we benchmark the latency for a fused node set, we do benchmarking twice: 1. benchmark the latency of the kernel including cloning mutated args 2. benchmark the latency of cloning mutated args without running the kernel We subtract result 2 from result 1 to get the latency of the kernel itself. But when the tensors are not on the cuda device 0, we get equal number for result 1 and result 2 no matter how much work the kernel does. The root cause is, in `triton.testing.do_bench` the `torch.cuda.synchronize` call sync the current cuda device (which is device 0 if it's not overriden). But since the tensors and kernels are located on another device, the sync actually does nothing (unless there happens to be other kernels on the device 0). The fix is to set the correct current device in our benchmarking code. cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng wenzhe-nrv jiayisunx ipiszy yf225 chenyang78 kadeng muchulee8 ColinPeppler amjames desertfire chauhang [ghstack-poisoned]
jansel
approved these changes
Sep 10, 2024
pytorchmergebot
pushed a commit
that referenced
this pull request
Sep 10, 2024
…135533) When a kernel does not have mutated args (this is quite common?), benchmarking the cost of cloning actually benchmarks a no-op. This still takes >100ms since triton.testing.do_bench will allocate 100 ms budget to run the kernel. Skipping this benchmarking can save quite some compilation time if the code path is hit multiple times. Let's say, if the code path is hit 100 times when the graph is large, we would save >10s. Pull Request resolved: #135533 Approved by: https://github.com/jansel ghstack dependencies: #135531
Chao1Han
pushed a commit
to Chao1Han/pytorch
that referenced
this pull request
Sep 20, 2024
…35531) Fix pytorch#134768 . When we benchmark the latency for a fused node set, we do benchmarking twice: 1. benchmark the latency of the kernel including cloning mutated args 2. benchmark the latency of cloning mutated args without running the kernel We subtract result 2 from result 1 to get the latency of the kernel itself. But when the tensors are not on the cuda device 0, we get equal number for result 1 and result 2 no matter how much work the kernel does. The root cause is, in `triton.testing.do_bench` the `torch.cuda.synchronize` call sync the current cuda device (which is device 0 if it's not overriden). But since the tensors and kernels are located on another device, the sync actually does nothing (unless there happens to be other kernels on the device 0). The fix is to set the correct current device in our benchmarking code. Pull Request resolved: pytorch#135531 Approved by: https://github.com/jansel
Chao1Han
pushed a commit
to Chao1Han/pytorch
that referenced
this pull request
Sep 20, 2024
…ytorch#135533) When a kernel does not have mutated args (this is quite common?), benchmarking the cost of cloning actually benchmarks a no-op. This still takes >100ms since triton.testing.do_bench will allocate 100 ms budget to run the kernel. Skipping this benchmarking can save quite some compilation time if the code path is hit multiple times. Let's say, if the code path is hit 100 times when the graph is large, we would save >10s. Pull Request resolved: pytorch#135533 Approved by: https://github.com/jansel ghstack dependencies: pytorch#135531
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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):
Fix #134768 .
When we benchmark the latency for a fused node set, we do benchmarking twice:
We subtract result 2 from result 1 to get the latency of the kernel itself.
But when the tensors are not on the cuda device 0, we get equal number for result 1 and result 2 no matter how much work the kernel does. The root cause is, in
triton.testing.do_benchthetorch.cuda.synchronizecall sync the current cuda device (which is device 0 if it's not overriden). But since the tensors and kernels are located on another device, the sync actually does nothing (unless there happens to be other kernels on the device 0).The fix is to set the correct current device in our benchmarking code.
cc @voznesenskym @penguinwu @EikanWang @jgong5 @Guobing-Chen @XiaobingSuper @zhuhaozhe @blzheng @wenzhe-nrv @jiayisunx @ipiszy @yf225 @chenyang78 @kadeng @muchulee8 @ColinPeppler @amjames @desertfire @chauhang