Gpu and multiple processes

Hello,
I have a Tesla C2050, I know host multiple threads can use this device Simultaneously.
I ask, Can multiple processes (host) use this device simultaneously?
Thanks

I have a gtx 470, which is same chip as the Tesla C2050.

I can run several instance of cuda sdk example (for example, 3 nbody). And I can run several instance of my own application.

I have a gtx 470, which is same chip as the Tesla C2050.

I can run several instance of cuda sdk example (for example, 3 nbody). And I can run several instance of my own application.

Yes, but the performance is worse with 20 (for example) jobs simultaneously. I fear the kernel of the jobs are not run simultaneously (like kernel of the thread), but sequentially. There is also a way to run different kernel of different jobs simultaneously?

Yes, but the performance is worse with 20 (for example) jobs simultaneously. I fear the kernel of the jobs are not run simultaneously (like kernel of the thread), but sequentially. There is also a way to run different kernel of different jobs simultaneously?

According to the programming guide, no. Only kernels belonging to the same context can run concurrently.

According to the programming guide, no. Only kernels belonging to the same context can run concurrently.