vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

building PTX code for both of 60 and 61

Browse source 
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
This commit is contained in:
Howard Su 2023-07-12 16:41:56 +08:00 • committed by GitHub
parent 397da62002
commit a95e105acd
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 1 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

2
CMakeLists.txt
View file

@ -268,7 +268,7 @@ if (LLAMA_CUBLAS)
if (NOT DEFINED CMAKE_CUDA_ARCHITECTURES)
if (LLAMA_CUDA_DMMV_F16)
set(CMAKE_CUDA_ARCHITECTURES "60") # needed for f16 CUDA intrinsics
set(CMAKE_CUDA_ARCHITECTURES "60;61") # needed for f16 CUDA intrinsics
else()
set(CMAKE_CUDA_ARCHITECTURES "52;61") # lowest CUDA 12 standard + lowest for integer intrinsics
endif()