#Recompile with cmake driver

$ export DYLD_LIBRARY_PATH=/Developer/NVIDIA/CUDA-7.0/lib:$DYLD_LIBRARY_PATH $ export PATH=/Developer/NVIDIA/CUDA-7.0/bin:$PATH In any case, I am stuck and any pointers would be appreciatedĪll right, so it appears I had skipped a step when installing CUDA support, namely those listed in the doc: I guess the MD_CUDABackend,h is still used nonetheless? This wouldn't annoy me too much if I had not specifically NOT used the CUDA option when using CMake. Still, after this and cleaning up the previous make files and rebuilding them, make -j4 still stalls at cuda.h which it can't find. Which is where I found out cuda.h resides.

I looked up some refresher about setting up some new default path on MacOS and the result of echo $PATH is now containing: OK, so the path to the CUDA library is not defined. Users/./oxDNA/src/Backends/./CUDA/Backends/MD_CUDABackend.h:12:10: fatal error: 'cuda.h' file not found In file included from /Users/./oxDNA/src/Backends/BackendFactory.cpp:17: There appears to be a typo there ('define MOVEACTORY_H_).Īfter configuring Cmake (I ran the GUI version) to use CUDA, I got an error and blamed it on a bad install, so I restarted without CUDA option, but still ran in the same error, namely: When running make, I got a warning regarding MoveFactory.h.

#Recompile with cmake driver

I have installed CMake and GNU Make, as well as the latest CUDA driver (7.5.20).