Manually cache block attributes in PackBMatrix::pack_unpack_#702
Closed
swolchok wants to merge 1 commit into
Closed
Manually cache block attributes in PackBMatrix::pack_unpack_#702swolchok wants to merge 1 commit into
swolchok wants to merge 1 commit into
Conversation
Summary: I noticed from inspecting assembly that we were re-loading the 4 block attributes on each loop iteration. It initially surprised me that the compiler was not able to prove that the loop body doesn't mutate the block (I don't see any function calls and TBAA should mean that it can prove nothing aliases the block). On reflection, it is notable that `T` is `signed char` -- `char*` is the one exception to the strict aliasing rule (see https://en.cppreference.com/w/cpp/language/reinterpret_cast), so the compiler cannot prove that `unpack_buf` and `pack_buf` do not alias `block` and thus it has to re-load from `block` each time. Reviewed By: jspark1105 Differential Revision: D30976702 fbshipit-source-id: 3bae8716065db08b1d39b245f9e0757a57425be4
Contributor
|
This pull request was exported from Phabricator. Differential Revision: D30976702 |
Contributor
|
This pull request has been merged in dccc573. |
pytorch-bot Bot
pushed a commit
that referenced
this pull request
Feb 26, 2026
Summary: Pull Request resolved: #702 I noticed from inspecting assembly that we were re-loading the 4 block attributes on each loop iteration. It initially surprised me that the compiler was not able to prove that the loop body doesn't mutate the block (I don't see any function calls and TBAA should mean that it can prove nothing aliases the block). On reflection, it is notable that `T` is `signed char` -- `char*` is the one exception to the strict aliasing rule (see https://en.cppreference.com/w/cpp/language/reinterpret_cast), so the compiler cannot prove that `unpack_buf` and `pack_buf` do not alias `block` and thus it has to re-load from `block` each time. Reviewed By: jspark1105 Differential Revision: D30976702 fbshipit-source-id: 0cb47b30aae883aceface0e0a0708504e1254933
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
2 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.
Summary: I noticed from inspecting assembly that we were re-loading the 4 block attributes on each loop iteration. It initially surprised me that the compiler was not able to prove that the loop body doesn't mutate the block (I don't see any function calls and TBAA should mean that it can prove nothing aliases the block). On reflection, it is notable that
Tissigned char--char*is the one exception to the strict aliasing rule (see https://en.cppreference.com/w/cpp/language/reinterpret_cast), so the compiler cannot prove thatunpack_bufandpack_bufdo not aliasblockand thus it has to re-load fromblockeach time.Reviewed By: jspark1105
Differential Revision: D30976702