# Zyphra introduces ZYFA, an FP8 attention kernel for AMD MI355X

> Original title: Introducing Zyphra FlashAttention (ZYFA): faster, more accurate 8-bit floating-point (FP8) attention on AMD MI355X.

- Source: Zyphra
- Published: 2026-11-06T02:36:46.000Z
- Bad Influence: https://badinfluence.news/items/nl3lqp0dr9lbja99eulv3gcfg
- Original: https://x.com/ZyphraAI/status/2118533018558308772

## Summary

Zyphra introduces Zyphra FlashAttention (ZYFA), an 8-bit floating-point attention kernel for AMD MI355X GPUs. Its prefill kernels deliver 1.16–1.76x the throughput of AMD's AITER inference kernel library on real model inputs at 16K tokens under sustained load.

## Post

Introducing Zyphra FlashAttention (ZYFA): faster, more accurate 8-bit floating-point (FP8) attention on AMD MI355X.

Our prefill kernels deliver 1.16–1.76x the throughput of AMD’s inference kernel library, AITER, on real model inputs at 16K tokens under sustained load. 🧵

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHWaKJtMbgAAecWD.jpg&mode=thumb&exp=1794182400&sig=7100d971a5d8f052)