Zyphra Releases Faster AMD MI355X Kernels

Zyphra released ZYFA, an open-source kernel library for AMD Instinct MI355X GPUs, offering FP8 and BF16 prefill attention. The code runs 1.16-1.76x faster than AMD's default AITER library. At head dimension 128, ZYFA's FP8 error is 62-78% lower than AITER's. The library supports causal forward attention and runs exclusively on CDNA4 architecture.