I'm porting some cuda code to ATI stream + OpenCL. I have a machine with a ATI card installed, but some tests need to be run in a linux machine with a nvidia card. Right now i'm running this code with the CL_DEVICE_CPU option but i wonder if there's another way to do this.
Is there any possibility to get this card working with the same code? (the compilation could be host-dependant). Maybe with some driver-magic?
I want to run this code in a linux 64bits machine.