search

martijnpieters / mcgpu

Automatically exported from code.google.com/p/mcgpu
1 stars 0 forks source link

NVCC problem #27

Open GoogleCodeExporter opened 8 years ago

GoogleCodeExporter commented 8 years ago 
Hello.

Currently, I am using MC-GPU v1.3 version.

In linux command, I made a running the script "make.sh"
, but had a massage the below. 

'nvcc' has been the problem...

please, solve the problem..

Thank you 

root@CentOS MC-GPU_v1.3_RELEASE_2012-12-12]# "./make_MC-GPU_v1.3.sh"

 -- Compiling MC-GPU v1.3 with CUDA 5.0 for both compute capability 2.0 and 3.0 (64 bits), with MPI:
    To run a simulation in parallel with openMPI execute:
      $ time mpirun --tag-output -v -x LD_LIBRARY_PATH -hostfile hostfile_gpunodes -n 22 /GPU_cluster/MC-GPU_v1.3.x /GPU_cluster/MC-GPU_v1.3.in | tee MC-GPU_v1.3.out

- /usr/local/cuda-6.5/bin/nvcc
nvcc MC-GPU_v1.3.cu -o MC-GPU_v1.3.x -m64 -O3 -use_fast_math -DUSING_CUDA 
-DUSING_MPI -I. -I/usr/local/cuda/include 
-I/usr/local/cuda-6.5/samples/common/inc 
-I/usr/local/cuda-6.5/samples/shared/inc/ -I/usr/include/openmpi -L/usr/lib/ 
-lmpi -lz --ptxas-options=-v -gencode=arch=compute_20,code=sm_20 
-gencode=arch=compute_30,code=sm_30
./make_MC-GPU_v1.3.sh: line 29: nvcc: command not found
.

Original issue reported on code.google.com by diracn...@gmail.com on 4 Feb 2015 at 9:18

GoogleCodeExporter commented 8 years ago 
The NVIDIA CUDA libraries and the CUDA compiler nvcc have to be correctly 
installed and accessible before trying to compile MC-GPU. Make sure you can 
compile and execute the sample codes distributed with the CUDA package before 
using MC-GPU.

Original comment by andre...@gmail.com on 9 Feb 2015 at 9:31