UltraQuant on AMD Instinct: More Efficient Agentic Serving for Qwen3.8-MXFP4 — ROCm Blogs.A native 4-bit MXFP4 KV cache | Hanami
UltraQuant on AMD Instinct: More Efficient Agentic Serving for Qwen3.8-MXFP4 — ROCm Blogs. A native 4-bit MXFP4 KV cache that speeds up Qwen3.8 decode over 8-bit KV on AMD Instinct MI355X, with no accuracy loss.