[gmx-users] GPU not found

Mirco Wahab mirco.wahab at chemie.tu-freiberg.de
Tue May 13 12:14:21 CEST 2014


On 13.05.2014 12:00, Albert wrote:
> Hello:
>
> thx for reply.
> it is very strange, I cannot run this command as normal user:
> Failed to initialize NVML: Insufficient Permissions
> here is the output I use sudo to run the command you mentioned:

Please look into the solutions posted here:
http://stackoverflow.com/questions/13054262/cuda-runtime-api-error-38-no-cuda-capable-device-is-detected

Otherwise, I usually exec the following script on headless
nodes to get all the nvidia-devices user-accessible:

------- 8< --- [cut here: cudainit.sh ] ----------------

#!/bin/sh
if [ "$?" -eq 0 ]; then
   # Count the number of NVIDIA controllers found.
   N3D=`/sbin/lspci | grep -i NVIDIA | grep "3D controller" | wc -l`
   NVGA=`/sbin/lspci | grep -i NVIDIA | grep "VGA compatible controller"
   wc -l`

   N=`expr $N3D + $NVGA - 1`

   for i in `seq 0 $N`; do
     mknod -m 666 /dev/nvidia$i c 195 $i;
   done

   mknod -m 666 /dev/nvidiactl c 195 255
else
   exit 1
fi

------- 8< --- [cut here: cudainit.sh ] ----------------

(run as root). This is not always necessary, seems to
depend on CUDA vs. driver version vs operating system
taste.

Regards

M.



More information about the gromacs.org_gmx-users mailing list