BLAST RADIUS

Open-weight & owned stack · Quantization efficiency

llama.cpp Adds AMD CDNA Fattn-MMA Kernel Support

Sep 27, 2026, first seen via Training & inference tooling releases

llama.cpp release b11214 enables the fattn-mma kernel on AMD CDNA for dkq above 256 and large batch sizes.

Read the original at github.comOpens the publisher's site in a new tab

More in Open-weight & owned stack