Do I need to download NVIDIA driver for onboard GPU if I am to use CUDA?

Accelerated Computing CUDA CUDA Programming and Performance
1

I just finished installing RHEL 5.3 on a new build that uses the Foxconn Destroyer motherboard.
So far I have only the onboard GPU, which is NVIDIA GeForce 8200 and is supposed to be able to
produce max resolution of 1920x1200, but RHEL 5.3 assigns it to a VESA driver that results in a much
lower resolution.

Since I intend to use this system with one or more Tesla C1060 cards, such that the monitor will be driven by the
onboard GPU, do I need to install a separate NVIDIA driver to be able to get the max resolution of 1920x1200, or
will installing CUDA 2.2 allow me to get the max resolution of 1920x1200?

Thank you!

2

AFAIK as someone new to cuda:
Installing the “normal” drivers will be enough. However, it is my understanding that all recent drivers are cuda-enabled (so that end-users will be able to benefit from cuda too) so there’s not really a difference.
You’ll need the cuda sdk however to develop against the teslas, but that doesn’t have anything to do with you resolution.

3

install CUDA 2.1 along with 180.22, and you’ll be good.

4

A new driver, 180.29, has just been released.

Does it make sense to install it instead of the 180.22?