Screen become blinking after run ./deviceQuery need to run deviceQuery to enable CUDA, and then scre

[minhtuan]

I have installed C2050 on a number of HP Z800 machines. Ubuntu servers are installed on them, except one has GNOME enabled to use graphics. On this machine, a separate GPU GTX380 is dedicated for graphics display. On the other machines, monitor is plugged into one of the C2050 for console display.

When the machines start, I need to run ./deviceQuery as root on every machine. Otherwise, my program using GPU cannot detect any GPU available on the machines. I’m not sure why, but this is fine, except that after running ./deviceQuery, it messed up with the C2050 display causing all monitor become unavailable, i.e. it shows some blinking rectangles (attached file). It avoids me to work with the machine directly. Any one has a similar issue or a solution to resolve this.

The driver is 280.13.

Thanks,
Tuan

[tera]

You need to create device entries in /dev with suitable access rights (which happens automatically if you run deviceQuery as root, but fails for normal users because of insufficient permissions). The release notes have a script to automatically do this at boot time.

[minhtuan]

You need to create device entries in /dev with suitable access rights (which happens automatically if you run deviceQuery as root, but fails for normal users because of insufficient permissions). The release notes have a script to automatically do this at boot time.

Hi Tera,

I could not found the proper information on the release note file

http://developer.download.nvidia.com/compute/DevZone/docs/html/C/doc/CUDA_Toolkit_Release_Notes.txt

Can you give more detail.

Thanks,

Tuan

[tera]

It used to be there before CUDA 4.1.

• In order to run CUDA applications, the CUDA module must be

  loaded and the entries in /dev created. This may be achieved

  by initializing X Windows, or by creating a script to load the

  kernel module and create the entries.

An example script (to be run at boot time):

#!/bin/bash

/sbin/modprobe nvidia

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

[minhtuan]

It used to be there before CUDA 4.1.

Thanks a lot Tera. I realized that the problem is resolved in CUDA 4.1.

Tuan