The Original issue was fixed, I'm closing this thread to stop it from being necroed.
If you have any issues with Battlefield 2042 please make a new topic about your problems and what kind of steps you have tried to solve them.
/Atic
The game is very heavy for CPU. Low GPU usage is caused by CPU limitation - 100% utilization makes no room to prepare more frames for GPU to process. However the default CPU cores and threads utilization is not optimal. If you see that GPU utilization is below 100% and CPU does not reach 100% then you may in fact try the following:
@LeMairyHuffDon’t check it through the game’s console. This won’t ever change. Monitor your CPU threads utilization and overall CPU %. In case of 6c/12t CPU, the default would be 6 worker threads (not 8). Try increasing Thread.MaxProcessorCount starting from 7 to 12 via User.cfg before starting the game. Also set Thread.ProcessorCount to the fixed 6.
i played around with these settings right on the first day of the early access, and there is something weird happening. setting the processor count to *far* below actual available physical/logical cores of course reduces performance heavily. however, setting it to somewhat below the actual available cores still may use *all* threads at 100%, which shouldnt happen.
the game seems to use 8 threads, and if one has eg. 8c/16t, it mostly uses the 1st thread of each physical core > 8 threads, while the second remains unused; occasional usage of both threads of a core is most likely due to windows power settings (overutilization > switch to another thread; or even constant duty cycling), however using high/gaming/ultimate performance power plans, or a custom one as i do, actually should prevent this from happening > another indication of an issue if it still happens, except it is due to heat throttling; stuttering may also occur when a task hops from one thread to another, this is why setting the game's worker threads above the actual available threads leads to stuttering or even crashes (task reserved for a non-existing core/thread needs to be queued back to an existing one, and the task currently performed on this one needs to be halted, then the same happens with the next task > microstutter).
