Better track output buffer lifetime and initialized state via types in GEMM internals #481
Conversation
This file contains 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
This has to use raw pointers because different parts of the output may be concurrently updated by multiple threads, but it can at least use lifetimes to tell Rust that it mutably borrows from the input.
If `beta` is zero then the output is potentially uninitialized. This was represented in the public API (`GemmExecutor::gemm_uninit`) but not internally as the output was always transmuted to `&mut [OutT]` before calling `gemm_impl`. Change the internals to use `MaybeUninit<OutT>` everywhere that the output is potentially uninitialized. This will help to avoid mistakes like the one fixed in 143f0a3.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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Outdated suggestions cannot be applied.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.
For performance the GEMM internals use raw pointers for output tiles, which may be uninitialized if
betais zero. Add lifetimes toOutputTilesand make more use ofMaybeUninit<OutT>to track the initialization status of each output tile through the implementation.