Back to feed

b7707

Jan 11, 2026
Meta/llama.cppCLIvb7707

Vulkan: Optimize Matmul parameters for AMD GPUs with Coopmat support (#18749)

  • vulkan: Enable and optimize large matmul parameter combination for AMD

  • limit tuning to AMD GPUs with coopmat support

  • use tx_m values instead of _l

macOS/iOS:

Linux:

Windows:

openEuler: