Jeff Bolz c44d449635 vulkan: fuse adds (llama/15252) ...

* vulkan: fuse adds

Fuse adds that have the same shape, which are common in MoE models.
It will currently fuse up to 6 adds, because we assume no more than
8 descriptors per dispatch. But this could be changed.

* check runtimeDescriptorArray feature

* disable multi_add for Intel due to likely driver bug

2025-08-18 20:30:45 +03:00

..

cmake

ggml: Skip backend library linking code when GGML_BACKEND_DL=ON (llama/15094)

2025-08-18 20:30:45 +03:00

include

ggml: initial IBM zDNN backend (llama/14975)

2025-08-18 20:30:45 +03:00

src

vulkan: fuse adds (llama/15252)

2025-08-18 20:30:45 +03:00

.gitignore

…

CMakeLists.txt

ggml: initial IBM zDNN backend (llama/14975)

2025-08-18 20:30:45 +03:00