Help a noob install GTX 750 Ti for CUDA acceleration

Hello, forum members.

I am trying to install a GTX 750 Ti in Linux to run the application NAMD with CUDA acceleration.
I can’t physically install the card before installing the driver because the screen scrolls with “[sequentially increasing number]…nouveau…unhandled intr …[memory address]” errors.

I have experienced the series of events described below both in Fedora 20 and the latest stable version of Debian.

I have tried to install the nvidia driver by downloading the .run file for the 334.21 version of the driver by downloading it, logging in to my user account, opening a terminal, executing ‘su’ to log in as root, then executing “init 3” to kill the X server, then logging in as root again when the terminal apears. From there, I copy the .run file to /tmp and execute:

sh N[…].run -k $(uname -r)

to install the driver. This proceeds OK and I power off the computer, physically install the card, and power the computer back on. When I boot the computer, I get a small multi-tone blue rectangle in the lower left corner of the screen. I have also tried using the integrated graphics after installing the graphics card by selecting ‘iGPU’ as the primary display option in the BIOS. That didn’t work either, but it was late last night, so I don’t recall what exactly happened at that point.

Any advice that anybody can give for getting this card to work in linux for CUDA acceleration in NAMD would be helpful and greatly appreciated!

remove the nouveau driver first. you’re going to have to do it at some point anyway. If you google “remove nouveau” and the name of whatever OS you are using, you’ll find about a million articles with instructions.

Thank you! I was putting “blacklist nouveau” in the wrong file. I guess I should have realized that [facepalm]

I’m in the same boat Ambidestrian. What did you do?