kernel time out

Accelerated Computing CUDA CUDA Programming and Performance
1

hello,

I am getting a kernel call timed out error while using a Geforce 8800 but no error while using a tesla

I’ve checked that i’m not exceeding the register limit and the kernel execution is rather short. What could be the possible cause??

I have a large array of data in global memory. I have a for loop surrounding the kernel, iteratively calling it and operating on different parts of the data. Is it something to do with accessing too much global memory or something?
what could be the possible cause?

2

forgot to mention…

if i reduce the number of threads, the kernel launches for a larger number of interations before getting a time out. I’m iteratively calling the kernel

3

Is the 8800 also being used for graphics? The OS will typically cancel running CUDA jobs if they exceed some OS specific threshold…

4

nope its not being used for graphics. The threshold is 5s right?

im running linux