I'm running a PC with Windows 11 and a I9 14900K processor.
I'm running a multi-threaded application and use 30 out of the 32 possible threads.
The software creates 30 individual batch jobs
During the run, the p-cores run at 5+Ghz and the e-cores run at 4+Ghz.
After 14 hours the p-core finished processing and their activity dropped to zero and the e-cores continued to run at 100%.
After a few minutes of down time the system appeared to redistribute the remaining 15% of the workload throughout all the core once again. However, the CPU's did not resume their peak performance speed. Instead, all the CPU's dropped down into the 2Ghz or less range. As each batch job completed, the problem got slower. This resulted in 5 additional hours of run time for just a fraction of the overall dataset.
I was expecting that the computer would continue to process each batch job at top performance instead of spreading it out all over the place, slowing down the processing speed and taking more than twice the amount of time to finish.
The first snapshot is the performance of the computer when all the batch jobs are running. All the CPU's are maxed out. The second is what I captured shortly after redistribution, I only caught the usage and not the speed. The usage is at 97% for all CPU's at the beginning and ultimately drops down and below 50% as the remaining data is processed.
How do I configure the machine to run a multi-threaded job at high processing speed from beginning to end?
Thanks, Paul
Link Copied
