Yhn SportBaseball first seen 1 d ago, last 10 min ago, peak #3
FlashAttention kernel auto-derived from 25 lines of SQL
Original: Show HN: Automatic deduction of FlashAttention kernel from ~25 lines of SQL
A developer has shown a tool that automatically deduces a FlashAttention GPU kernel from roughly 25 lines of SQL, shared as a benchmark on a code repository. The claim has drawn attention from programmers discussing whether a declarative query language can really generate an optimized attention kernel, a core building block of modern AI models, without hand-written CUDA code.
Why now: It combines SQL, compiler-style automatic code generation and FlashAttention, a hot topic in AI performance engineering.
FlashAttentionxarray-sqlGitHub
Rank over time, top of the chart is #1. 60 snapshots from 23 h ago to 10 min ago.
Evidence
API: https://socialmediatrends-api.osmike.com/v1/trends/4283000