AWhat changed
Computed exact attention with tiled GPU operations and recomputation, reducing transfers between high-bandwidth memory and on-chip storage without materializing the full attention matrix.
Scientific scope
Computed exact attention with tiled GPU operations and recomputation, reducing transfers between high-bandwidth memory and on-chip storage without materializing the full attention matrix. The dense algorithm changes memory access, not quadratic arithmetic complexity. IO bounds have an explicit memory model; speedups are hardware-specific. Sparse extensions are approximate and remain distinct.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. The dense algorithm changes memory access, not quadratic arithmetic complexity. IO bounds have an explicit memory model; speedups are hardware-specific. Sparse extensions are approximate and remain distinct.
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: Fast and Memory-Efficient Exact Attention with IO-Awareness
- Stanford HAI: FlashAttention research record
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