What steps will reproduce the problem?
1. Upgrade CUDA from version 5.5.22-1 to 6.0.37-1
2. Link my code against gpuocelot
My code linked succeffsully against CUDA 5.5, but fails against CUDA 6.0:
function __nv_init_managed_rt_with_module(void**): error: undefined reference
to '__cudaInitModule'
I'm using gpuocelot-2255-4-x86_64 downloaded and compiled from the SVN
repository as part of an arch user repository (AUR) package build
Hope this can be fixed so that CUDA 6.0 features will be supported
Keep the good work going!
Original issue reported on code.google.com by
a.bianci...@gmail.comon 26 Apr 2014 at 5:21