[Neurodebian-users] fsl-5.0-gpu

[Alex Waite]

Hey Jonathan,

I don't have any experience with fsl-5.0-gpu (one of the other guys on 
here will have to chime in to answer that question); however, I do have 
a few comments about your nouveau "blocking."

> for FSL) card. For that purpose I installed debian 7 wheezy, the
> neurodebian repository, and the nvidia driver and the cuda toolkit. This
> time I did it all through apt-get instead of installing with the
> NVIDIA*.run files (maybe I shouldn't have?)

Installing through via the package manager is almost always a vastly 
superior choice. You did the right thing. :-)

 > which is always a nightmare
> since you have to block "nouveau" and it's never easy.

Are you blocking the nouveau package from being installed via apt-get, 
or is this a process specific to CUDA? If you're talking about the 
package manager, there's no need to block nouveau.

https://wiki.debian.org/NvidiaGraphicsDrivers#Debian_7_.22Wheezy.22

This link shows how easy it is now. Nouveau /can/ be installed at the 
same time as nvidia's proprietary blob, but nvidia's GLX libs will 
"win." Then, if you create the /etc/X11/xorg.conf.d/20-nvidia.conf as 
they suggest, X11 will use the nvidia driver rather than nouveau. No 
need for blocking, pinning, etc.

> I have first a preliminary question that has nothing to do with
> neurodebian but rather with cuda. Why should we block nouveau and
> install the nvidia grafical driver since at the end we want to avoid
> using the nvidia card as a graphical adapter but rather as a computing
> gpu?

Are there directions you're following from somewhere that tell you to do 
this?

If, indeed, you cannot use a GPU for both computing and for display, 
then your driver, nvidia or nouveau won't matter. They're two different 
drivers for the same GPU. Your laptop may simply not have enough GPUs 
(e.g. 2+) for the task at hand.

Many modern Intel CPUs come with a GPU on-die. If you have a laptop with 
that type of CPU, you /might/ be able to use the Intel GPU for display 
and then dedicate the nvidia GPU to computational tasks. Do you know 
what your CPU is? cat /proc/cpuinfo | grep 'model name' should tell you.

Sorry I don't know more about the fsl-5.0-gpu package, but hopefully 
some of what I provided will help.

---Alex