Refactors variable declarations for better readability #68
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
Reorders tensor variable declarations to group related functionality together and improves code formatting consistency. Moves tensor fragment declarations closer to their logical usage groups and reformats conditional statements with proper indentation for enhanced code maintainability.
Contributor
There was a problem hiding this comment.
Pull Request Overview
Refactors variable declarations in flash_fwd_kernel.h to improve readability by grouping related tensor fragments and shared-memory copy operations, and reformats a multi-line call for consistent indentation.
- Rewrites the
std::mincall within theIs_causalblock into a properly indented multi-line expression. - Moves the
tOrVttensor fragment declaration to sit alongside other partition fragments. - Relocates the SMEM V copy declarations (
smem_tiled_copy_V,smem_thr_copy_V,tOsVt) to group them with the K copy logic.
Simplifies the flash forward splitkv kernel by removing the Append_KV template parameter and associated logic. This reduces template instantiation complexity and removes one level of nested switch statements. Also updates debug printf statements to include function names for better debugging clarity and adjusts block size constants for improved performance characteristics.
Updates block size calculation to use smaller values for better memory efficiency. Comments out split-KV functionality to avoid extra memory overhead as mentioned in the existing comment about always setting num_splits back to 1.
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Outdated suggestions cannot be applied.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.
Reorders tensor variable declarations to group related functionality together and improves code formatting consistency.
Moves tensor fragment declarations closer to their logical usage groups and reformats conditional statements with proper indentation for enhanced code maintainability.