llama.cpp/ggml
Ruben Ortlam 803dac2e48
vulkan: use vec dot for matrix matrix multiplications (#16056)
* vulkan: Change the mul_mm shared memory and register caching system to use vec2 instead of scalars, to enable using dot2 instructions

* use fma instead of dot to fix Nvidia and Apple performance issues
2025-09-20 10:42:56 +02:00
..
cmake ggml: Skip backend library linking code when GGML_BACKEND_DL=ON (#15094) 2025-08-07 13:45:41 +02:00
include metal : refactor + optimize v2 (#15995) 2025-09-17 20:38:12 +03:00
src vulkan: use vec dot for matrix matrix multiplications (#16056) 2025-09-20 10:42:56 +02:00
.gitignore vulkan : cmake integration (#8119) 2024-07-13 18:12:39 +02:00
CMakeLists.txt metal : refactor + optimize v2 (#15995) 2025-09-17 20:38:12 +03:00