@ekzhang1: me looking at people like this guy who write real gpu kernels :)

X AI KOLs Timeline News

Summary

AI model Claude was used to write a FlashAttention forward kernel using the pyptx DSL, achieving near-parity performance with hand-tuned FlashAttention-4 on NVIDIA B200 hardware.

👀 me looking at people like this guy who write real gpu kernels :)
Original Article
View Cached Full Text

Cached at: 07/08/26, 03:41 AM

👀 me looking at people like this guy who write real gpu kernels :)

Patrick C Toulme (@PatrickToulme): Claude Fable wrote a FlashAttention forward kernel in pyptx DSL for the NVIDIA B200 (Blackwell) that runs at 0.92–0.99× of FlashAttention-4, the hand-tuned CUTLASS kernel — parity on two sequence lengths, ~1350 TFLOPS bf16.

It’s written in pyptx, my Python DSL that emits raw

Similar Articles