G2 Patch load?

jamos · Posted: Sat Jul 03, 2004 6:17 pm Post subject: G2 Patch load?

I'm not clear on the meaning of the Patch Load display. Is this per slot, or global?

In other words.. can I have all four slots running at 100%, or must 100% be split between all four slots (25/15/10/50, etc.)?

egw · Posted: Sat Jul 03, 2004 7:39 pm Post subject:

It's per slot.

ian-s · Posted: Sat Jul 03, 2004 10:26 pm Post subject:

yes, per slot.

but one slot can use 200% (half of the total) if you fully load the voice and effects area.

gal · Posted: Sun Jul 04, 2004 2:15 am Post subject:

100% means 1 dsp is used.
if you use 200% in slot a and 200% in slot b and you don't have the expansion (then you have 4 dsp's), slot c and d cannot be used.

gal

Tim Kleinert · Posted: Sun Jul 04, 2004 3:57 am Post subject:

It's important to understand how the DSP computation loads are handled in the G2.

100% on the editor equals one DSP.

Generally, and logically, computations for a discrete entity (such as a single voice from any patch, or any FX area of a slot) cannot be split up between two DSPs.

So, for example, if either the memory or patch load of the VA (Voice Area) patch shows something above 50%, and the FX section too, this delivers only 3 voices (from an unexpanded G2), since every voice occupies an entire DSP, and the FX too. The rest of DSP power is wasted, so to speak.

Now, if you manage to reduce the FX section loads to the point that it can reside on one DSP together with a voice (voice% and FX% <=100%), you get a fourth voice. If you manage to reduce the voice loads below 50%, zwoop, you get seven. (two on each of 3 DSPs, one + FX on the last)

If you run 4 slots, all with individual voice designs and FX, it can become a truly complex affair to juggle the patch loads in order to get the optimum DSP usage. scratch

It may well be that a big patch cannot deliver any more voices and you think your'e done -but if you build a simple little patch in another slot, it may well still be able to provide surprisingly many voices, as they use up the "unused niches" on the DSPs left over by the "monster patch" in the other slot.

Best thing is to get an expanison. I've never seriously hit the ceiling since installing it.

BTW, I think Clavia should change the specs in the G2 presentation. They state 32 voice polyphony for the G2, but that's not quite true. It's 32 voices per slot, meaning that the G2 is actually capable of 128 voices !

I'm seriously thinking of suggesting a "voice overflow" module to the Clavia developers. A module, that, when the slot polyphony is overridden, just sends those MIDI note-ons (and offs) to other slots. 48-voice FM pianos anyone? Wink

jamos · Posted: Sun Jul 04, 2004 10:25 am Post subject:

Thanks for the replies. That is deep and interesting info.

I do have an expansion board on order. I haven't yet hot a real problem with polyphony, but can see how I might someday. Hopefiully the board will cure all ills.

blue hell · Posted: Mon Jul 05, 2004 4:44 pm Post subject:

Tim Kleinert · Posted: Mon Jul 05, 2004 5:17 pm Post subject:

Yes, I like the idea too.

The question is if, generally, "voice area" modules can access and manipulate the behaviour of the underlying dynamic voice allocation routine of the slot, as it would be needed in this case.

I personally doubt that, but who knows.

I suggested it to Clavia this morning, posting it on their mozilla comment panel. But I wonder if there is a more efficient way to get in touch with the developers, though. I don't want to hassle them, but sometimes I have good ideas which, as an ex assembler coder, I can put into programming terms quite easily.

Anyway -waiting for 1.2 (like all of you, I guess) Wink

blue hell · Posted: Mon Jul 05, 2004 5:28 pm Post subject:

Tim Kleinert · Posted: Mon Jul 05, 2004 5:42 pm Post subject:

I'm pretty sure it would take an overworking of the voice allocation routine, so that the overflowing voices do not override existing ones, but are processed by the manner described in the module: sending them to slot A-D. But then, following note-offs would also have to be tracked and sent to the other slot. And what about controllers? They would have to be shared too. So it's not as simple as it seems. The best solution probably would be some kind of "slot sharing" idea, where all CCs and note-offs are sent to all shared slots, but the note-ons are only sent to a shared slot when polyphony is overridden on the first, triggering the voice there. Something like that. scratch

Beta tester? I'd love to do that. I'm good at pushing things to the limit until the flaws show out. It's almost obsessive Wink

Who would I need to contact?

blue hell · Posted: Tue Jul 06, 2004 10:39 am Post subject: