frodo_buggins c30fc1cbd7 Add engine building selection to README | il y a 3 ans | |
---|---|---|
src | il y a 3 ans | |
test | il y a 4 ans | |
third_party | il y a 4 ans | |
tools | il y a 4 ans | |
.gitignore | il y a 4 ans | |
ACKNOWLEDGEMENTS | il y a 4 ans | |
CMakeLists.txt | il y a 3 ans | |
LICENSE | il y a 4 ans | |
README.md | il y a 3 ans | |
main.cpp | il y a 4 ans |
A CUDA miner for yggdrasil network addresses
OpenCL engine requires OpenCL C++ headers (opencl-clhpp-headers
on Ubuntu, opencl-clhpp
on Archlinux, dev-libs/clhpp
on Gentoo, etc.).
The GPU's compute capability should be specified for optimal performance. It can be obtained from CUDA capabilities.
cmake -H. -Bbuild -DCMAKE_CUDA_ARCHITECTURES=<your CUDA capability, i.e 61 for Pascal 1060 GTX>
cmake --build build
Currently there are two engines: CUDA and OpenCL. By default both are built, but it's possible to select only one via YGG_BRUTE_ENGINES
variable:
# Select only the CUDA engine
cmake -H. -Bbuild -DYGG_BRUTE_ENGINES=cuda
# Select only the OpenCl engine
cmake -H. -Bbuild -DYGG_BRUTE_ENGINES=opencl
ygg-brute --help