Zyphra · FP8 Attention Kernels
ZYFA
Zyphra has released ZYFA, open-source FP8 and BF16 prefill attention kernels for AMD Instinct MI355X accelerators. On real model activations, its FP8 kernels run 1.16-1.76x faster than AMD's AITER library, with 62-78% lower error at head dimension 128. The page does not state a licence.
Open the source ↗Similar Developer Tools releases
- Together Compiler (TKCC)
Together AI · Continual Inference Compiler · Week 46, 2026
- CUDA Toolkit 13.5
NVIDIA · GPU Toolkit Release · Week 46, 2026
- DeepInfer
DeepSeek · Inference Engine · Week 44, 2026
- Toward Recursive Self-Improvement
Z.ai · Self-Built Inference Stack · Week 39, 2026
- A Hardware-Level Tour of LLM Inference
Dr Ashish Bamania · LLM Inference Internals · Week 36, 2026
- Kie.ai
Kie.ai · AI Infrastructure · Week 31, 2026
More from Zyphra
- Zyphra
Zyphra · Survey & Overview · Week 19, 2026
- ZAYA1 8B
Zyphra · Small Language Models · Week 19, 2026
- ZUNA
Zyphra · Research Paper · Week 8, 2026
- Zonos-v0.1-hybrid
Zyphra · MoE Architecture · Week 2, 2025
Published by Quanta AI Labs, an AI agency in India. Browse every week in AI News Weekly.