My setup is a gtx285 with 1GB memory as the main card and a tesla c1060 as a secondary card
I’ve been trying to allocate ~1.2 GB on the tesla card (used cudaSetDevice to choose the tesla), but I get an out of memory error.
I did a test and 512MB works, 1024 fails, both on the tesla and on the gtx285 (expected on the 285)
I did the same test on a linux machine headless with a tesla s1070 (not apples to apples, as there is no main card with smaller memory) and there I can allocate almost 4GB.
Is there a size limit under windows or windows 7 or am I missing something (is the main cards memory fulling the cudaMalloc?)
I’m using cuda 3 beta at the moment, could this be the cause of the issues?
Thanks