Why does this Rapture instrument take so much CPU power?

As far as I can tell, Adam, the only difference in the progression of 6 Elements is an increasing amount of Detune in each one. (Very methodical approach; I like that. ). The other Elements get buried by that last Detune of 42 in E6. For my money, you could lose the first 5 and retain most of the patch's character. That should drop the CPU cost significantly.

If you still want some variation from note to note, set a destination in the MIDI Matrix to Detune 6 at a middle Depth, like 21. Adjust the GUI Detune to the same number. Then select a source of velocity, or one of the random generators. For what it's worth, I just ran this patch through an optimized 2800+ and got the same CPU cost. Dropping the first 5 Elements [switched off] saved 50%.

I 'thought' a supersaw was many sawtooth waveforms detuned from one another. What am I missing here?

It is, Adam. But you can reach a saturation point. Detune ranges from 0-100 cents, and the oscillator voices (3v-9v) are spread equally within that range. They're also panned to different locations, and it almost sounds like a phase shift as well. So when you've got 9 voices at Detune levels up to almost half a semitone, things get pretty thick pretty fast. The other Elements each have 9 more voices apiece spread at those panning positions, but they're detuned from their base oscillator and the other Elements. In this case, I think that the Detune of 42 wins out. That's really quite a bit.

You can get just the panning effects with a Detune of 0, but try smaller amounts of Detune, like 1,3, or 5. Even the default setting of 10 is a little 'phat'. You can get some complex thickness by varying these small Detune amounts over Elements with different multi-oscillator voice numbers. Then add a very small Depth of Pitch LFO at a very slow Freq. (if you like). You'll get very full-bodied sound without the oscillators stepping all over each other, and clarity will improve.

Yea, there is a wikky somewheres where the spread is shown. Basically each extra voice (3-5-7-9) adds a stereo voice or ocscillator (as I would call it) that is spread out across the soundfield. The detune stretches the detune across the cent range, as I understand it. So having a 2nd similar element detuned to a smaller amount probably won't show up that well, unless you fade the two elements in and out or use the LFO to vary the pitch differnetial, which should add some movement. I think that is what Mr. Rock was saying.

Thanks, B Rock and AT!!!

So, when I initially said that a SuperSaw is many sawtooth waveforms detuned from another, I could really do that with one element set to 9V being that it is evenly spread out. So, basically, what I had was 54 (6 elements * 9 voices each) sawtooth waveforms detuned from one another!!!

No wonder why it took so much CPU power!! I was doing a mega supersaw!!! Geez!!

Thanks again!! Right when I think I had something all figured, I learn something else. This is great!

Just to clarify the what's, wherefore's and why's, check out René's reply in [this somewhat related thread]

I'd like to give a slight insight on a techie detail about the multisaw.

Both Rapture and Dimension Pro can play the above multisaw.sfz file. So is that the same thing as having the MULTI switch turned on in Rapture?

Not really. Rapture MULTI mode will create 3, 5, 7 or 9 oscillators with the loaded samples, which then will be routed to a -single- dsp chain. This means, the two filters will process the mix of all oscillators as a whole.

This effect results in a higher control when applying distortion and lo-fi effect, and reduces enormously the CPU load.
In other words, if you apply to the sfz file above a filter, 3 EGs and 2 LFOs, and then play a single note, then you'll get six filters, 18 EGs and 12 LFOs pumping at once. Using Rapture MULTI mode, you get one filter, 3 EGs and 2 LFOs. This allows you to stack another element with another MULTI osc, and still having your room with a decent temperature.

The sfz has some advantages though: you can specify the six voices to play with a random value affecting something like filter cutoff on each, so they behave individually from start to end like in the old synthies unison modes. Expensive, but cool results could come out of it.


-René