Skip to content
This repository was archived by the owner on Apr 24, 2022. It is now read-only.

Commit 8f48e43

Browse files
committed
Support Ampere graphic cards using Cuda 11.0/11.1
Ampere graphic cards require compute capability 8.0 and 8.6. Without this, trying to run this on my RTX 3090 fails. See https://arnon.dk/matching-sm-architectures-arch-and-gencode-for-various-nvidia-cards/
1 parent d5f3a97 commit 8f48e43

1 file changed

Lines changed: 8 additions & 0 deletions

File tree

libethash-cuda/CMakeLists.txt

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -28,6 +28,14 @@ else()
2828
if(NOT CUDA_VERSION VERSION_LESS 10.0)
2929
list(APPEND CUDA_NVCC_FLAGS "-gencode arch=compute_75,code=sm_75")
3030
endif()
31+
if(NOT CUDA_VERSION VERSION_LESS 11.0)
32+
# NVIDIA A100 and NVIDIA DGX-A100
33+
list(APPEND CUDA_NVCC_FLAGS "-gencode arch=compute_80,code=sm_80")
34+
endif()
35+
if(NOT CUDA_VERSION VERSION_LESS 11.1)
36+
# Tesla GA10x cards, RTX Ampere – RTX 3080/3090, RTX A6000, RTX A40
37+
list(APPEND CUDA_NVCC_FLAGS "-gencode arch=compute_86,code=sm_86")
38+
endif()
3139
endif()
3240

3341
file(GLOB sources "*.cpp" "*.cu")

0 commit comments

Comments
 (0)