I’ve just got myself a PCIe 1x GeForce NVS285 to use as a main display, at the same time I stuck into my PC 2 x GT240 PCIe x16 (one 512MB, second 1024). The operating system is Windows XP x64.
I experience a bit frustrating behaviour when running CUDA kernels. If I run it on first CUDA GPU GT240, all is fine. Though, when I run my kernel on second CUDA GPU GT240, it freezes the whole system for the time of computations …
Any idea why CUDA behaves this way? I thought these freezes happen only if I attempted to use my display card.