CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-31 08:51:55 +00:00
Code Issues Packages Projects Releases Wiki Activity

cuda : only use native when supported by cmake (#10389)

Browse Source
This commit is contained in:
Diego Devesa
2024-11-18 18:43:40 +01:00
committed by GitHub
parent 531cb1c233
commit d3481e6316
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
ggml/src/ggml-cuda/CMakeLists.txt
View File

@@ -12,7 +12,7 @@ if (CUDAToolkit_FOUND)
# 61 == Pascal, __dp4a instruction (per-byte integer dot product) # 61 == Pascal, __dp4a instruction (per-byte integer dot product)
# 70 == V100, FP16 tensor cores # 70 == V100, FP16 tensor cores
# 75 == Turing, int8 tensor cores # 75 == Turing, int8 tensor cores
if (GGML_NATIVE AND CUDAToolkit_VERSION VERSION_GREATER_EQUAL "11.6") if (GGML_NATIVE AND CUDAToolkit_VERSION VERSION_GREATER_EQUAL "11.6" AND CMAKE_VERSION VERSION_GREATER_EQUAL "3.24")
set(CMAKE_CUDA_ARCHITECTURES "native") set(CMAKE_CUDA_ARCHITECTURES "native")
elseif(GGML_CUDA_F16 OR GGML_CUDA_DMMV_F16) elseif(GGML_CUDA_F16 OR GGML_CUDA_DMMV_F16)
set(CMAKE_CUDA_ARCHITECTURES "60;61;70;75") set(CMAKE_CUDA_ARCHITECTURES "60;61;70;75")