-
Notifications
You must be signed in to change notification settings - Fork 22k
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
Implement operator for micro-pipelined all-gather -> _scaled_mm #129289
Conversation
[ghstack-poisoned]
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/129289
Note: Links to docs will display an error until the docs builds have been completed. ✅ You can merge normally! (1 Unrelated Failure)As of commit 4a81ea1 with merge base a21d436 (): UNSTABLE - The following job failed but was likely due to flakiness present on trunk and has been marked as unstable:
This comment was automatically generated by Dr. CI and updates every 15 minutes. |
ghstack-source-id: 268c246f430ed35d4186822e9a55320ca174f1c5 Pull Request resolved: #129289
fyi @drisspg on matmuls in async TP |
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.
Very cool it can all be done in python
…ed_mm" This PR implements `torch.ops.symm_mem.fused_all_gather_scaled_matmul`. It's similar to `torch.ops.symm_mem.fused_all_gather_matmul`, except that it takes scales and calls ` _scaled_mm`. [Profiling Trace vs. Baseline](https://interncache-all.fbcdn.net/manifold/perfetto-artifacts/tree/ui/index.html#!/?url=https://interncache-all.fbcdn.net/manifold/perfetto_internal_traces/tree/shared_trace/yifu_tmp0gmg1f2_) (FB internal only) cc mrshenli pritamdamania87 zhaojuanmao satgera gqchen aazzolini osalpekar jiayisuse H-Huang kwen2501 awgu penguinwu fegin XilunWu wanchaol fduwjj wz337 tianyu-l wconstab yf225 chauhang d4l3k [ghstack-poisoned]
…ed_mm" This PR implements `torch.ops.symm_mem.fused_all_gather_scaled_matmul`. It's similar to `torch.ops.symm_mem.fused_all_gather_matmul`, except that it takes scales and calls ` _scaled_mm`. [Profiling Trace vs. Baseline](https://interncache-all.fbcdn.net/manifold/perfetto-artifacts/tree/ui/index.html#!/?url=https://interncache-all.fbcdn.net/manifold/perfetto_internal_traces/tree/shared_trace/yifu_tmp0gmg1f2_) (FB internal only) cc mrshenli pritamdamania87 zhaojuanmao satgera gqchen aazzolini osalpekar jiayisuse H-Huang kwen2501 awgu penguinwu fegin XilunWu wanchaol fduwjj wz337 tianyu-l wconstab yf225 chauhang d4l3k [ghstack-poisoned]
ghstack-source-id: 08dec6ca9027f97e6252fb863442b21a45aa3552 Pull Request resolved: #129289
…ed_mm" This PR implements `torch.ops.symm_mem.fused_all_gather_scaled_matmul`. It's similar to `torch.ops.symm_mem.fused_all_gather_matmul`, except that it takes scales and calls ` _scaled_mm`. [Profiling Trace vs. Baseline](https://interncache-all.fbcdn.net/manifold/perfetto-artifacts/tree/ui/index.html#!/?url=https://interncache-all.fbcdn.net/manifold/perfetto_internal_traces/tree/shared_trace/yifu_tmp0gmg1f2_) (FB internal only) cc mrshenli pritamdamania87 zhaojuanmao satgera gqchen aazzolini osalpekar jiayisuse H-Huang kwen2501 awgu penguinwu fegin XilunWu wanchaol fduwjj wz337 tianyu-l wconstab yf225 chauhang d4l3k [ghstack-poisoned]
ghstack-source-id: d7e7bd2c6a386f0574e052130daa38ec0b55dc91 Pull Request resolved: #129289
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
The merge job was canceled or timed out. This most often happen if two merge requests were issued for the same PR, or if merge job was waiting for more than 6 hours for tests to finish. In later case, please do not hesitate to reissue the merge command |
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
The merge job was canceled or timed out. This most often happen if two merge requests were issued for the same PR, or if merge job was waiting for more than 6 hours for tests to finish. In later case, please do not hesitate to reissue the merge command |
…ed_mm" This PR implements `torch.ops.symm_mem.fused_all_gather_scaled_matmul`. It's similar to `torch.ops.symm_mem.fused_all_gather_matmul`, except that it takes scales and calls ` _scaled_mm`. [Profiling Trace vs. Baseline](https://interncache-all.fbcdn.net/manifold/perfetto-artifacts/tree/ui/index.html#!/?url=https://interncache-all.fbcdn.net/manifold/perfetto_internal_traces/tree/shared_trace/yifu_tmp0gmg1f2_) (FB internal only) cc mrshenli pritamdamania87 zhaojuanmao satgera gqchen aazzolini osalpekar jiayisuse H-Huang kwen2501 awgu penguinwu fegin XilunWu wanchaol fduwjj wz337 tianyu-l wconstab yf225 chauhang d4l3k [ghstack-poisoned]
ghstack-source-id: 189cbb29deeaadd1982a287147ff561926ae9268 Pull Request resolved: pytorch#129289
@pytorchbot merge |
Merge startedYour change will be merged once all checks pass (ETA 0-4 Hours). Learn more about merging in the wiki. Questions? Feedback? Please reach out to the PyTorch DevX Team |
…rch#129289) This PR implements `torch.ops.symm_mem.fused_all_gather_scaled_matmul`. It's similar to `torch.ops.symm_mem.fused_all_gather_matmul`, except that it takes scales and calls ` _scaled_mm`. [Profiling Trace vs. Baseline](https://interncache-all.fbcdn.net/manifold/perfetto-artifacts/tree/ui/index.html#!/?url=https://interncache-all.fbcdn.net/manifold/perfetto_internal_traces/tree/shared_trace/yifu_tmp0gmg1f2_) (FB internal only) Co-authored-by: Will Feng <[email protected]> Pull Request resolved: pytorch#129289 Approved by: https://github.com/Chillee, https://github.com/weifengpy, https://github.com/drisspg
Stack from ghstack (oldest at bottom):
This PR implements
torch.ops.symm_mem.fused_all_gather_scaled_matmul
. It's similar totorch.ops.symm_mem.fused_all_gather_matmul
, except that it takes scales and calls_scaled_mm
.Profiling Trace vs. Baseline (FB internal only)
cc @XilunWu @H-Huang @awgu @kwen2501 @wanchaol @fegin @fduwjj @wz337 @wconstab @d4l3k @c-p-i-o @mrshenli @pritamdamania87 @zhaojuanmao @satgera @gqchen @aazzolini @osalpekar @jiayisuse @penguinwu @tianyu-l @yf225 @chauhang