diff options
| author | Johannes Gäßler <johannesg@5d6.de> | 2023-07-31 19:52:22 +0200 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2023-07-31 19:52:22 +0200 |
| commit | b772bba42e3bbca3cdab224456f8ff2ce427fd0b (patch) | |
| tree | 97b56e2c5af55b00eb46b4f038f9eed9120dde66 | |
| parent | 0728c5a8b9569183ffca0399caac099afef87595 (diff) | |
CUDA: fixed cmake F16 option (#2471)
| -rw-r--r-- | CMakeLists.txt | 6 |
1 files changed, 3 insertions, 3 deletions
diff --git a/CMakeLists.txt b/CMakeLists.txt index 4ecb3d5..1d4e63f 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -73,7 +73,7 @@ set(LLAMA_CUDA_MMQ_Y "64" CACHE STRING "llama: y tile size for mmq CUDA ke option(LLAMA_CUDA_FORCE_DMMV "llama: use dmmv instead of mmvq CUDA kernels" OFF) set(LLAMA_CUDA_DMMV_X "32" CACHE STRING "llama: x stride for dmmv CUDA kernels") set(LLAMA_CUDA_MMV_Y "1" CACHE STRING "llama: y block size for mmv CUDA kernels") -option(LLAMA_CUDA_DMMV_F16 "llama: use 16 bit floats for dmmv CUDA kernels" OFF) +option(LLAMA_CUDA_F16 "llama: use 16 bit floats for some calculations" OFF) set(LLAMA_CUDA_KQUANTS_ITER "2" CACHE STRING "llama: iters./thread per block for Q2_K/Q6_K") option(LLAMA_CLBLAST "llama: use CLBlast" OFF) option(LLAMA_METAL "llama: use Metal" OFF) @@ -265,8 +265,8 @@ if (LLAMA_CUBLAS) if (DEFINED LLAMA_CUDA_DMMV_Y) add_compile_definitions(GGML_CUDA_MMV_Y=${LLAMA_CUDA_DMMV_Y}) # for backwards compatibility endif() - if (LLAMA_CUDA_DMMV_F16) - add_compile_definitions(GGML_CUDA_DMMV_F16) + if (LLAMA_CUDA_F16 OR LLAMA_CUDA_DMMV_F16) + add_compile_definitions(GGML_CUDA_F16) endif() add_compile_definitions(K_QUANTS_PER_ITERATION=${LLAMA_CUDA_KQUANTS_ITER}) |