Hi, I compiled the new cuda-9 branch and everything compiled without problem.
Makefile.am was adjusted to nvcc_ARCH = -gencode=arch=compute_35,code="sm_37,compute_35" for my K80-card
Then I wanted to run the benchmark to test my setup. Started fine and then failed

./ccminer --benchmark

...


[2017-11-25 15:14:33] Benchmark algo equihash...
[2017-11-25 15:14:33] GPU #0: solver: CUDA error 'an illegal memory access was encountered' in func 'solve' line 2052
[2017-11-25 15:14:34] GPU #0: an illegal memory access was encountered
[2017-11-25 15:14:34] GPU #0: scanhash_equihash:192 all CUDA-capable devices are busy or unavailable
[2017-11-25 15:14:34] GPU #0: init: CUDA: failed to alloc memory

what could be the reason or the solution?

Thx