AVO agents evolve attention kernels beating FlashAttention-4 by 10.5% on B200
bingxu_ · x · 2026-09-25
AVO, accepted at NeurIPS 2026, replaces fixed mutation/crossover heuristics in evolutionary search with autonomous coding agents that propose, repair, critique and verify kernel edits. After 7 days of autonomous evolution on Blackwell B200, AVO discovers attention kernels up to 3.5% faster than cuDNN and 10.5% faster than FlashAttention-4; transferring to grouped-query attention takes just 30 more minutes for up to 9.3% gains.
More from Infra
- Testing Flux 2 quants on a 9070 XT: BF16 beats FP8 in speed and quality — Sofa-Sleuth · 2026-09-25
- Geely unveils EV tech charging 10-70% in under 5 minutes — Polymarket · 2026-09-25
- Google Cloud launches PostgreSQL for agents in AlloyDB with isolated instances scaling to 1,000+ readers — usamawahabkhan · 2026-09-25
- Seroter's Reading List: Agentic DB Architecture and Agent Skills Guide — rseroter · 2026-09-25
- Huawei 960 racks hit 200kW per cabinet in 2027, still 4x behind Blackwell — teortaxesTex · 2026-09-25
- 10-year Treasury yield hits pre-GFC high as AI capex borrows at nation scale — tszzl · 2026-09-25