Closed pawanjoe closed 8 years ago
Newer versions of CUDA don't support pre-Fermi compute capabilities.
You should remove this flag from CUDA_CPPFLAGS in ./config/common.mk
Based on the GPU you are testing, you should change the 12 or 20 to the compute capability for your GPU. You can look that up here:
https://developer.nvidia.com/cuda-gpus
Example with Titan X (remove all -gencode except):
-gencode=arch=compute_52,code=sm_52
Thanks! That worked.
It gives an error: " nvcc fatal : Unsupported gpu architecture 'compute_12' make[4]: * [BusSpeedDownload.o] Error 1 make[4]: Leaving directory `/home/pawan/Desktop/summer_school/Intern/shoc-master/src/cuda/level0/epmpi' make[3]: * [all-recursive] Error 1 make[3]: Leaving directory
/home/pawan/Desktop/summer_school/Intern/shoc-master/src/cuda/level0' make[2]: **\* [all-recursive] Error 1 make[2]: Leaving directory
/home/pawan/Desktop/summer_school/Intern/shoc-master/src/cuda' make[1]: [all-recursive] Error 1 make[1]: Leaving directory`/home/pawan/Desktop/summer_school/Intern/shoc-master/src' make: \ [all-recursive] Error 1 "I have set the right PATH, and configure worked.