Why S870 does not work in init3 mode?

[queeten]

Hello!
I have got my S870 work normally in X window, ie init 5 mode. But if I start the system to init 3 mode, it does not work. As I try to launch any cuda program, it responded that libcudart.so.2 cannot be found. Anyone can help me?
Thanks!

[tmurray]

that just sounds like a path problem. you’re going to have to set up the device entries as well once you resolve that–look at the 2.1 beta release notes for the script to do that (they’re not in the 2.0 notes for whatever reason, but they are in the docs for 2.1).

[queeten]

that just sounds like a path problem. you’re going to have to set up the device entries as well once you resolve that–look at the 2.1 beta release notes for the script to do that (they’re not in the 2.0 notes for whatever reason, but they are in the docs for 2.1).

Thanks for a quick response. The cuda version is 2.0, and my system is rhel4.x. I notice that in init3 mode, the color of tesla’s light is orange, not yellow, as in the normal condition. I think maybe S870 is utterly not recognized by the system.

And, another problem is that, every time I start my system, it takes a long time, some 2 minutes for X window to start. and I check the manual and attach Option “IgnoreDisplayDevices” “DFP, TV” in the xconfig file. But the problem still exists. So what’'s wrong?

Thanks!

[MisterAnderson42]

Thanks for a quick response. The cuda version is 2.0, and my system is rhel4.x. I notice that in init3 mode, the color of tesla’s light is orange, not yellow, as in the normal condition. I think maybe S870 is utterly not recognized by the system.

Because the driver hasn’t initialized it yet. Do check the release notes as tmurray suggested. The script to create the dev nodes is most certainly what you need to apply.

[queeten]

Because the driver hasn’t initialized it yet. Do check the release notes as tmurray suggested. The script to create the dev nodes is most certainly what you need to apply.

Thank you for your answer. I check 2.0 release notes and notices the following warning:

"While X does not need to be running in order to use CUDA,

X must have been initialized at least once after booting

in order to properly load the NVIDIA kernel module. The

NVIDIA kernel module remains loaded even after X shuts down,

allowing CUDA to continue to function."

But I have launched SDK in X, which means that, according to release notes, kernel has been loaded. So, what’s wrong in this case?

[MisterAnderson42]

You again failed to read tmurray’s instructions. He said that it was not in the 2.0 release notes for some reason! If you were to look in the 1.1 or 2.1 release notes you would find this

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

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

[netllama]

Its unclear to me exactly what problem you’re having or how to replicate it. Initially you stated that you were getting errors trying to locate libcudart.so.2. The most likely causes for that problem are:

• you didn’t install the CUDA toolkit
• you didn’t set LD_LIBRARY_PATH and/or add the toolkit library path to /etc/ld.so.conf (per the documentation)

Then you stated that the “S870 is utterly not recognized by the system”. I don’t know what that means. If the S870 GPUs don’t appear in lspci output, then that is either a hardware problem or a Linux kernel bug or an SBIOS bug (or some combination of the 3). Its definitely not a CUDA bug.

Its unlikely that anyone is going to be able to offer further assistance until you provide detailed descriptions of the problem(s) you are reporting, and how to reproduce them.