Genoil, I tried the new release, checkout the 110 branch.
Now I get a new error: [ 61%] Building NVCC (Device) object libethash-cuda/CMakeFiles/ethash-cuda.dir/ethash-cuda_generated_ethash_cuda_miner_kernel.cu.o
nvcc fatal : Unsupported gpu architecture 'compute_61'
I'm getting this error too when trying to make the 110 branch with -DBUNDLE=cudaminer ..
Running on Mint 17.3, all other versions worked prior.
This is because the default list of supported compute platforms now includes Compute 6.1 for the 10x0 series. It requires CUDA 8 RC to build. If you don't want to install CUDA 8, you can add -DCOMPUTE=52 to the cmake line to support only Compute 5.2 devices, -DCOMPUTE=30 to support only Compute 3.0 devices, and so on.
Ahh, thanks, makes sense now. Is there a way to support 3.5 and 5.0 on the -DCOMPUTE? For example if I have:
Listing CUDA devices.
FORMAT: [deviceID] deviceName
Compute version: 3.5
cudaDeviceProp::totalGlobalMem: 3220504576
[1] GeForce GTX 750 Ti
Compute version: 5.0
cudaDeviceProp::totalGlobalMem: 2147287040
(I would test, but I'm away from the rigs atm) Thanks.
Got in remote. -DCOMPUTE=52 seemed to work for both cards. (3.5 & 5.0)