CUDA and multiple displays

I’m trying to start some CUDA work on my system. I’m running a dual monitor system, with an 8600GTS. The demos run fine if I disable my second monitor, but not if I have it on. I’m wondering why this is, and if switching to one of the 8800 cards would solve this problem.


We currently run an 8800 Ultra with 2 x 1920x1200 displays, we are running them in twinview mode though, so it only looks like one. You might try doing that as opposed to getting a different card.

(but then again, I’m no expert, and we’re running Linux)

I’ve got this configuration working currently. Do any of the demos run with both monitors (deviceQuery, bitonic,…). I’d expect problems with the interop demos such as fluids/simple d3d/gl, but not the others.

Is the 8600GTS the only graphics device in your system?

You’re running WinXP and have the CUDA 1.0 driver installed?

I just posted my experiences starting out with running
the SDK sample codes on my 8600GT and found that
two monitors in Horizontal Span Mode works as well:


I have a 8800 GTS with WinXP SP2 and a dualcore AMD.

Heres what i have found about connecting multiple displays and having different configurations:

The dual-view configuration -

  1. Normal GPGPU code works without a problem and any inefficiency.
  2. OpenGL interop does not work.
  3. DirectX interop works, but if you have the application running on the primary monitor, it will suck up a chunk of resources off the host CPU.

All othe configurations everything works fine. Hope this helps.


i also have a 8800gts and dual monitors in dualview, on winxp sp2 intel D and it has been working well for months

So what might I be doing wrong?

Nothing, there was an issue in 1.0 with OpenGL interop with multiple monitors or multiple GPUs. Will be fixed in 1.1.

could be the 8600, i should get one soon and then i will let u know if i see any problems