Dear all,
we compiled bedpostx_gpu for a MAC OS (El Captain) using CUDA 7.5 with a NVIDIA GT 750M (384 cores).
At a first try, we got an out of memory error. After specifying NJOBS=200, the processing finished with success.
Are there any guidelines to choose optimally the number of jobs depending on number of voxels, GPU cores, etc.?
Best Regards
Sebastian
|