I have a new computer that came with an ATI Radeon HD 3400 graphics card. I wanted to do some CUDA computing so I purchased a Quadro NVS 420 card. The quadro card did not come with the proper video cables for my display, but I considered that of little consequence since I just wanted the card for GPU computing. I installed it side-by-side with the ATI card. The ATI card is still driving my display, but the NVIDIA card is also present. I installed the drivers, the CUDA v2.3 toolkit and the NVIDIA SDK. I have a C++ program that I’ve written that uses CUDA and works fine on several other cuda-capable systems. This program fails to run on the new system - the one with the ATI/NVIDIA combo. The program simply calls to CUFFT.DLL to perform some FFTs.
I’m wondering if there are special considerations that I need to take into account when running with two different boards installed in this manner.
Thanks for any help.
GB