[gmx-users] no GPU detected

Albert mailmd2011 at gmail.com
Wed Jul 8 08:17:50 CEST 2015


I've got a GPU machine with 4 GPU cards. I compiled gromacs-5.0.5 with 
GPU support successfully, and I try to run some jobs.

However, the job always failed with following messages:

  Command line:
   mdrun_mpi -s md2.tpr -g -v md.log -c md.gro -x md.xtc -e md.edr 
-gpu_id 01 -ntomp 12
Back Off! I just backed up md.log to ./#md.log.6#
Number of hardware threads detected (48) does not match the number 
reported by OpenMP (24).
Consider setting the launch configuration manually!
NOTE: Error occurred during GPU detection:
       no CUDA-capable device is detected
       Can not use GPU acceleration, will fall back to CPU kernels.
Reading file md2.tpr, VERSION 5.0.5 (single precision)
Changing nstlist from 10 to 20, rlist from 1 to 1.028
Using 2 MPI processes
Using 12 OpenMP threads per MPI process
No GPUs detected on host cudaC

I am just wondering, what's problem? I use nvidia-smi and can find the 
GPU without any problem. I've installed CUDA7 in the linux X64 system.

thank you very much.


More information about the gromacs.org_gmx-users mailing list