AWhat changed
Reorganized exact-attention work across GPU blocks and warps and reduced non-matrix-multiplication operations to improve accelerator utilization.
Scientific scope
Reorganized exact-attention work across GPU blocks and warps and reduced non-matrix-multiplication operations to improve accelerator utilization. Year follows the 2023 preprint; later conference publication is not a second result. This is a distinct work-partitioning refinement of FlashAttention, with measured gains tied to the tested GPU.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Year follows the 2023 preprint; later conference publication is not a second result. This is a distinct work-partitioning refinement of FlashAttention, with measured gains tied to the tested GPU.
Contributors
- Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
- FlashAttention-2: Faster Attention with Better Parallelism and Work Partitioning
- Tri Dao: original FlashAttention-2 paper
Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.
Full scientific record · Browse contribution grades · Request a correction