Ok please forgive my ignorance here but I have seen references to both disabling hyper threading and making sure all my CPU cores are being used.
Do I just find hyperthreading in bios and click disable? What does this do?
And how can I ensure I am “using all my cores?” I am not over clocking.
ASUS ROG Strix F gaming mobo
13900KF
4090, 64 GB DDR5 6400 mhz
Windows 11
Oculus 2/Virtual Desktop
Thanks!
G
Hyperthreading (as Intel calls it; AMD calls it Symmetric Multi-Threading - SMT) puts two front-end computing contexts onto each core; each of the two “threads” decodes and runs different instructions, but they both share the same arithmetic and memory units. Sometimes both threads can run things at the same time, other times they have to wait on each other. The end result of hyperthreading in a heavily-multithreaded workload is usually a modest increase in throughput (total amount of data computed) but more variability in latency – because two “threads” in one “core” don’t equal the performance of two full “cores”.
In MSFS, there’s a mix of workloads that are limited to a single thread (especially “the main thread”) and other work that’s split over many threads. Beware that here, “thread” means something different from the previous paragraph – it now means a thread of software execution, which may be placed on any “logical core” (either an entire core, if HT is off, or one of the two hardware “threads”, if HT is on)/
This combination may cause the main thread to lose performance, because it can be placed against another working thread and have to share some of its time. Your mileage may vary though.
As far as checking if cores are well used; there’s not much you can do about it but you can look at the per-core breakdown in any CPU monitor tool such as Task Manager. However these are hard to read clearly. What you’ll tend to see is that 1 or 2 logical processors are pretty consistently high (probably running the main thread) while others will be lower or higher depending on what’s going on in the simulation. In big airports at busy photogrammetry cities you’ll see higher overall usage, for instance, while flying high in the middle of nowhere you’ll usually see less overall usage and most concentrated on 1 or 2 cores.