-
Notifications
You must be signed in to change notification settings - Fork 21.4k
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
[Quant][PT2E]Make _fuse_conv_bn_ support graph capture by torch._dynamo.export #107951
[Quant][PT2E]Make _fuse_conv_bn_ support graph capture by torch._dynamo.export #107951
Conversation
🔗 Helpful Links🧪 See artifacts and rendered test results at hud.pytorch.org/pr/107951
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 8c28215 with merge base 2bddfb0 ( 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: 8e25878b43a189b5eecb3c5c25fb580f8025c742 Pull Request resolved: #107951
[ghstack-poisoned]
ghstack-source-id: 4e39262f14b5d3b08a7adc16824610d4c25b58a1 Pull Request resolved: #107951
…torch._dynamo.export" [ghstack-poisoned]
@jgong5 @jerryzh168 Please kindly help to take a look of this PR. |
if conv_node.target == torch.ops.aten.convolution.default: | ||
assert type(conv_node.args[6]) is bool | ||
transpose = conv_node.args[6] |
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.
could you add a TODO to remove these, same for others
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.
Thanks for the comments, TODO added here and others in this diff.
ghstack-source-id: 78e6d6f36dd6355b450d40507811564a4e381b25 Pull Request resolved: #107951
…torch._dynamo.export" **Summary** The latest check-in a0cfaf0 for the conv-bn folding assumes the graph is captured by the new graph capture API `torch._export.capture_pre_autograd_graph`. Since we still need to use the original graph capture API `torch._dynamo_export` in 2.1 release. So, this check-in made negative impact to workloads' performance heavily. Made this PR to fix this issue by trying to make the conv-bn folding function workable with both new and original graph capture API. [ghstack-poisoned]
ghstack-source-id: fce277b4d8e5db5429f7b3ddc1bea1fc04466a91 Pull Request resolved: #107951
…torch._dynamo.export" **Summary** The latest check-in a0cfaf0 for the conv-bn folding assumes the graph is captured by the new graph capture API `torch._export.capture_pre_autograd_graph`. Since we still need to use the original graph capture API `torch._dynamo_export` in 2.1 release. So, this check-in made negative impact to workloads' performance heavily. Made this PR to fix this issue by trying to make the conv-bn folding function workable with both new and original graph capture API. [ghstack-poisoned]
ghstack-source-id: fb9124b4c4cfb736844c0f4e75e11fa729eb6713 Pull Request resolved: #107951
…torch._dynamo.export" **Summary** The latest check-in a0cfaf0 for the conv-bn folding assumes the graph is captured by the new graph capture API `torch._export.capture_pre_autograd_graph`. Since we still need to use the original graph capture API `torch._dynamo_export` in 2.1 release. So, this check-in made negative impact to workloads' performance heavily. Made this PR to fix this issue by trying to make the conv-bn folding function workable with both new and original graph capture API. [ghstack-poisoned]
@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 |
…mo.export (#107951) **Summary** The latest check-in a0cfaf0 for the conv-bn folding assumes the graph is captured by the new graph capture API `torch._export.capture_pre_autograd_graph`. Since we still need to use the original graph capture API `torch._dynamo_export` in 2.1 release. So, this check-in made negative impact to workloads' performance heavily. Made this PR to fix this issue by trying to make the conv-bn folding function workable with both new and original graph capture API. Pull Request resolved: #107951 Approved by: https://github.com/jgong5, https://github.com/jerryzh168 ghstack dependencies: #106836, #106838, #106958
… API" **TODO** - Enable Dynamic Shape - Revert the changes in #107951 cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx peterbell10 ipiszy ngimel yf225 chenyang78 kadeng muchulee8 aakhundov [ghstack-poisoned]
…ew graph capture API" **Summary** - **TODO** Revert the changes in #107951 cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx peterbell10 ipiszy ngimel yf225 chenyang78 kadeng muchulee8 aakhundov [ghstack-poisoned]
… API" **Summary** - **TODO** Revert the changes in #107951 cc voznesenskym penguinwu EikanWang jgong5 Guobing-Chen XiaobingSuper zhuhaozhe blzheng Xia-Weiwen wenzhe-nrv jiayisunx peterbell10 ipiszy ngimel yf225 chenyang78 kadeng muchulee8 aakhundov [ghstack-poisoned]
#108317) **Summary** Revert the changes in #107951 to make the utils function only support graph captured by `capture_pre_autograd_graph`. Pull Request resolved: #108317 Approved by: https://github.com/jgong5, https://github.com/jerryzh168 ghstack dependencies: #108214
Stack from ghstack (oldest at bottom):
Summary
The latest check-in a0cfaf0 for the conv-bn folding assumes the graph is captured by the new graph capture API
torch._export.capture_pre_autograd_graph
. Since we still need to use the original graph capture APItorch._dynamo_export
in 2.1 release. So, this check-in made negative impact to workloads' performance heavily. Made this PR to fix this issue by trying to make the conv-bn folding function workable with both new and original graph capture API.