One other item I wanted to add about total voices, using that video where he uses Massive as an example. If you'll notice, the CPU load from all three tracks at the beginning is not that much greater than when he auditions a single Massive instance.
There are a number of reasons for this, but with Dune2 you can essentially do what he's doing with a single instance of Dune2 (treat the voices like independent layers to tweak and detune separately, on different envelopes even and fatten them up). And you can pretty quickly watch your CPU meter go up if you create a fat dreamtrance synth sound like he's doing, but you'll also see that if you add more instances of Dune2, the CPU meter does not rise much more, presumably because of the overhead of the first instance and/or the benefits of multithreading (just a guess).
But then, take that same fat sound that's showing 25-30% CPU usage, and create something equally as fat on the Virus-- you'll get as many notes as you have fingers to hold down keys on Dune2, but on the Virus you'll hear note-stealing happening, more noticeable if you have a long release.
Something else I've noticed is that CPU meters seem to be based on initial clock frequency of the CPU. For example on my PC if I leave the CPU at its default clock but let it go into turbo (i.e. upping the clock speed of individual cores as demand requires), the CPU meter in both Cubase and FLS will appear that it's struggling more. Or, I can go in and set it to a fixed clock speed at its max stable frequency, and the CPU will appear to be far less taxed, but the musical result is the same.
Point being that I don't trust CPU meters inside the DAW too much, I try to use them more as a relatively measurement tool than something that really represents an upper processing limit.
|