I know there is something called graphics “frame rate” in GPU device and the kernel call should never take longer than this “frame rate”. otherwise results will get ugly.
I am trying to figure out if there is a bug in my kernel or my kernel is taking longer than “frame rate”. because my results get ugly only when a while loop is included inside my kernel.
Is there a way to find “frame rate” for each GPU device? I have a GTX 260 and Tesla C1060.
what could be a work around if my kernel call is longer than “frame rate”? can i make another kernel call from inside a kernel?
I will appreciate any discussion leading to resolve this issue. I am running code on Ubuntu 10.04 64x.