Hi there, I am learning about MPS’s thread percentage setting. It is really an interesting and useful property! And I have a question about it: is it possible to dynamically change the active thread percentage of a running MPS process?
I really believe dynamic thread limitation can further improve deployment flexibility. So I wonder is there a way, such like an exposure through procfs or something, to change it.
Have you figured it out?
I have the same question for MPS active thread percentage.
Hello, I have the same problem. Is there a solution? If there is a solution, please share it with me. I would be extremely grateful.
You cannot edit the percentage assigned to a process today. We are actively looking into it.
If you control the application source code, you could handle this by creating multiple contexts and switching which context you submit to. This is much easier to accomplish with execution contexts now ( CUDA Runtime API :: CUDA Toolkit Documentation ).