llama.cpp releasesTools
b12285: cuda : fattn mma head sizes 40 72 (#30285)
cuda : add MMA FlashAttention support for head sizes 40 and 72 Head sizes 40 and 72 (used e.g. in vision towers of Gemma 3/4 and Qwen3-VL) were excluded from every tensor-core dispatch branch and…
Read at llama.cpp releases ↗Related
Toolsb12263 llama.cpp releases

AsiaNvidia CEO Jensen Huang’s affair: everything you need to know SCMP Tech

ValleyNvidia’s crisis has corporate governance experts reaching for the popcorn TechCrunch AI

AsiaWhat Taiwanese chip giant TSMC gains from taking stakes in OpenAI and Anthropic’s AI race SCMP Tech

AsiaAMD expands EPYC server chip lineup for AI infrastructure Tech in Asia
