Expansion Board for G2 Engine.

[blue hell]

Maybe some examples would illustrate things best. Please ask questions when I'm not clear, I don't really know what it is that you need to hear to make you understand better.

First this. All resource percentages in the editor are relative to one DSP chip. So when you use 100% that's equivalent to one fully loaded DSP. I know there are two percentages given, one for DSP cycles and another for memory, but that doesn't really matter. Here we just assume the highest of those two to be the DSP resource usage.

Lets assume you have one patch only and it's in some slot as the only active slot. This patch happens to use 50% of a DSP for its voice area and another 50% of a DSP for its FX area. Now when you have requested one voice for this patch it will fill one entire DSP. And when 3 voices are requested it will fill two DSPs completely, and not three. This is because the FX area does not get duplicated when more voices are requested. So for three voices you need three voice areas and one on FX area. All these areas were 50%, so that adds up to 200%, or two full DSPs.

So for this particualr patch we can see that we could have 7 voices, as 7 times a voice area plus one FX area would be eight areas, and each area being 50% we get a total of 400% or four DSPs. Likewise when we would have eight DSPs we could get 15 voices for the patch, so it more than doubles the possible polyphony (from 7 to 15) for this particular example.

The trick for easy calculation is that you work with patch-areas. For an FX area you always have one patch area but for a voice area you get as many patch-areas as there are voices allocated for the patch. The patch areas have size that are percentages of a full DSP, and the special thing is that the patch-areas corresponding to the voice-areas voices can be allocated to different DSPs.

Now lets look at another example. Suppose you have four different patches and each patch is pretty big, it uses (near) 100% for the voice area and anotehr 100% for the FX area. So each of these patches needs two DSPs at least (when the voice count is set to one), and you can see that on an 8 DSP system you could actually load thes four patches into the four slots, but that on a four DSP system this would be impossible - the first two loaded would have consumed all DSP space already. The G2 system will set the allocated voice count for the 2nd two patches to zero, meaning : out of DSPs.

Now in this last example the four patches could have been made to cooperate closely. You can do this by using the G2's audio routing capabilities, but als by sending MIDI messages from slot to slot. So in a limited sense this could be considered to be one huge patch.

[Acidfever]

Aaaaahh, i think i understand now.

If it's just about loading 4 patches it would be just as good to have 2 G2's then. Only when you want to be able to get enough voices from that 1 supercomplex patch you would really need the expansion.

I have recently traded one of my G2 engines for a Poly Evolver Rack. So far a nice machine, but i cannot guerentee i won't trade it back for a G2 engine or even an G2 keyboard. The g2 is just that nice!