cg vs. CUDA vs. Two GPUs


I have two GPUs in my machine: 1 quatro fx 4600 and 1 fx5200.

I have an application which executes using CG a part of the code and another part using CUDA.

I want to run both parts of code using my quatro fx 4600.

I don’t have problems with this configuration and CUDA but the CG part is not working:

  1. If the fx5200 is my primary monitor, I’m getting an error from the CG part
  2. If the quatro fx 4600 is my primiray monitor, the data output from the CG part is only 0s.

Do you know what it is going?
Do you know about any solution?


I don’t know anything about Cg or Quatro’s but maybe if you change your monitor to on or another, that GPU becomes the primary and in your code you select the primary adapter somewhere so thats why it won’t work. If you know what I mean. its a bit hard to explain my thoughts :P

It’s not clear exactly what your problem is here, but you could use the GPU affinity extension to control which GPU the OpenGL context gets created on:…pu_affinity.txt

This is only supported on Quadro, btw.