I can actually confirm that - there is a noticeable decrease on the ASIOGuard load in the C12 performance meter when having one external effect inserted somewhere in the project.
I can't really explain why. It might be some extra buffer that is used when inserting an external effect. Might be just measured differently
Quote:
Originally Posted by Nevelous
It sure feel kind of stupid that I'm maxing out my ASIO guard when my CPU is only at 45%, but I guess I might just be running out of threads, since Cubase really depend on them, not the CPU power.
The overall CPU usage is not really relevant as a measurement. In earlier Cubase versions the performance meter was directly correlated on the most used CPU core in the system, so if you had one saturated core and 7 other idling, the meter was at 100%.
It is not like that anymore, though, as Cubase scheduling was adapted to more and more cores, and has become somewhat better at spreading the load over many cores. This comes at a price of more overhead for thread synchronization, as far as I can see.
I simply don't know how the performance meter measurement works, Steinberg don't tell it, so it is just guesswork.
I rarely have problems with the ASIOGuard load anyway, less so with my new Ryzen 7950, but the realtime path (peak) can still be a problem, especially with some headphone correction plugins loaded in the control room and e.g. a CPU intensive VSTi in record/monitor mode.