llama.cpp/ggml
Jeff Bolz 96452a3fa4
vulkan: Reuse conversion results in prealloc_y (#15410)
* vulkan: Reuse conversion results in prealloc_y

Cache the pipeline and tensor that were most recently used to fill prealloc_y,
and skip the conversion if the current pipeline/tensor match.

* don't use shared pointer for prealloc_y_last_pipeline_used
2025-08-21 16:55:00 +02:00
..
cmake ggml: Skip backend library linking code when GGML_BACKEND_DL=ON (#15094) 2025-08-07 13:45:41 +02:00
include musa: add GGML_UNUSED_VARS (#15446) 2025-08-21 11:06:05 +08:00
src vulkan: Reuse conversion results in prealloc_y (#15410) 2025-08-21 16:55:00 +02:00
.gitignore vulkan : cmake integration (#8119) 2024-07-13 18:12:39 +02:00
CMakeLists.txt CUDA: replace GGML_CUDA_F16 with CUDA arch checks (#15433) 2025-08-20 16:58:49 +02:00