Automatically Run Device Query

Accelerated Computing CUDA CUDA Programming and Performance
1

In order to use CUDA 3.1 after a machine reboot the deviceQueryDrv command (located in: /usr/local/NVIDIA_GPU_Computing_SDK/C/bin/linux/release) has to be ran. Is there a way to run this command automatically everytime the server reboots?

Thanks!

2

In order to use CUDA 3.1 after a machine reboot the deviceQueryDrv command (located in: /usr/local/NVIDIA_GPU_Computing_SDK/C/bin/linux/release) has to be ran. Is there a way to run this command automatically everytime the server reboots?

Thanks!

3

You should be able to run it from /etc/rc.local I think.

It sounds like I have the same problem as you. I’m trying to install several Dell M610x blades in a cluster that have Tesla cards. I’m using the 3.1 SDK and the driver listed on the download page for the 3.1 sdk (256.40). When I boot up and try to run the examples in the SDK as a regular user, they fail. If I run deviceQuery as root, it succeeds, and then subsequent tries to run the examples programs as regular users succeed.

Do we really need to run deviceQuery to get things working?

4

You should be able to run it from /etc/rc.local I think.

It sounds like I have the same problem as you. I’m trying to install several Dell M610x blades in a cluster that have Tesla cards. I’m using the 3.1 SDK and the driver listed on the download page for the 3.1 sdk (256.40). When I boot up and try to run the examples in the SDK as a regular user, they fail. If I run deviceQuery as root, it succeeds, and then subsequent tries to run the examples programs as regular users succeed.

Do we really need to run deviceQuery to get things working?

5

You could use the script that is in the 3.1 release notes that will load the kernel module and initialize the /dev/nvidia* entries for multiuser access.

6

You could use the script that is in the 3.1 release notes that will load the kernel module and initialize the /dev/nvidia* entries for multiuser access.

7

Ok thanks, I hadn’t noticed that script, it seems to be working for me.

Does this still need to be done with version 3.2 if we decide to upgrade? I don’t see it in the release notes for that version.

8

Ok thanks, I hadn’t noticed that script, it seems to be working for me.

Does this still need to be done with version 3.2 if we decide to upgrade? I don’t see it in the release notes for that version.