Arch Linux User Repository

This package builds without CUDA support which saves approx. 1.4GB compared to 'ethminer-cuda'.

AMD users, you may need to add -DUSE_SYS_OPENCL=ON to cmake and install opencl-amd. For me, this removed the CL compile error on launch and the "gave incorrect result. Lower overclocking.." message. I also needed to specify --cl-devices in my launch parameters. I seem to have 2 CL devices per video card. I think this is from opencl-amd. The SECOND duplicate seems to be the one I need to use. Hope this helps someone.

Also the boost lib fix seems totally unnecessary now. I can build and run without it.

This package builds without CUDA support which saves approx. 1.4GB compared to 'ethminer-cuda'.

Adopted, updated to v0.16.1.

@sarvex, it does that because that is the configuration supported by upstream, to prevent new bugs introduced by 'incorrect' versions of the dependencies.

The package does not depend on host libraries, it downloads boost and others again, even when the packages are already present on the system

WARNING: this package now builds WITHOUT cuda support. To avoid errors like codemonkey's, I've decided to split this package up into ethminer, and ethminer-cuda. If someone wants to build with cuda support, he/she can install ethminer-cuda and makepkg will take care of the dependencies.

EDIT: I installed cuda AFTER running makepkg, my issue below is incorrect.

For some reason this did not build with CUDA support for me, despite having a CUDA compatible card and cuda installed.

Manually compiling source with -DETHASHCUDA=ON did work, and nvcc does exist, so -f '/opt/cuda/bin/nvcc' should evaluate to true, but does not?

Why does this install the binary release without compiling the actual source and yet it requires all the dependencies for building?