I recently upgraded to an AMD 3950x with 16 cores/32 threads from an Intel i5-9600k with 6 cores.
In my Bayesian Optimization for my Neural Networks, my CPU usage was at 100% across all 6 cores. Now on my new processor, it appears I'm using fewer cores/just as many at only 10% CPU usage (along with several programs and browsers open).
I was hoping for a more significant speedup since it was so taxing on my last processor. Is this just as fast as it can go and the process is instead being bottlenecked by my GPU since that is the execution environment it is set to?
Any way for me to speed this up/allocate more usage such as starting a local parallel pool?
Attached a picture showing my utilization.
Best Answer