Hi,
I am running an application on Intel Xeon E5 using OpenCL SDK XE, and when I checked the system monitor to observe thread utilization it shows that most of the item only one thread is running at 100% and rest running at very low percentage. I am just wondering if we have to declare any specific parameter to make sure the application is mapped on all threads on a single processor.
FYI: When I run second instance of algorithm in parallel I observe another thread being utilizing 100% so it meas the code is the limiting factor in mapping the kernel on multiple workitems and workgroups. I also declare large workgroup and workitem number in my application code.
Thank you,
Sai.