You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I am trying to improve upon the workload partitioning problem solutions for het. platforms. THe SHOC benchmarks run on GPUs only. How do I compare the CPU-only performance with this? How do i partition the workload between CPU and GPU?
The text was updated successfully, but these errors were encountered:
You can run on the CPU by compiling the OpenCL version and using an implementation that exposes the CPU as an OpenCL device. You can measure CPU and GPU simultaneously by using MPI and OpenCL. There is not a way to dynamically balance the workload, though.
I am trying to improve upon the workload partitioning problem solutions for het. platforms. THe SHOC benchmarks run on GPUs only. How do I compare the CPU-only performance with this? How do i partition the workload between CPU and GPU?
The text was updated successfully, but these errors were encountered: