Hi, I am a novice CUDA developer trying to improve CUDA program’s performance using MPS.
I found there are two ways to improve performance of MPS - one is CPU pinning using taskset -c ~ , and the other one is Volta MPS Execution Resource Provisioning - from this reference : https://docs.nvidia.com/deploy/mps/index.html
CPU pinning method worked very well. Now I am working on resource provisioning method.
I wonder is there any other way to improve CUDA MPS performance except methods that I mentioned.
Tae Young Yeon