Concurrent Kernels

Hello :)

I’m currently implementing pthreads to my existing CUDA code. I was wondering cause I’m a little confused. Ever since 3.0 came out we can run concurrent kernels or it’s just for the fermi arch?

I’m running on a GeForce GT 220. I wish to run 6 threads and each of them containing a kernel call. The code runs to a certain point and then crashes.

It is only for Fermi. You must launch the kernels with different streams, not with different threads… :D

So my approach with pthreads is never going to work?

What do you mean streams?

I think yes…

You can read about it in CUDA C Programming Guide Version 3.1.1

Page 37-38 ( and