Twist, LPG and FM

[baconpaul]

From the massive kvr thread user https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=203660 reports:

Found a bug in the new TWIST dsp module ,
When lpg is enabled for osc 1 ( also the only osc enabled in the mixer ) and osc fm routing is set to 2--1 , fm depth to zero , this somehow influences osc 1 , bug goes away lpg is disabled .
Steps to reproduce
-for osc 1 choose TWIST , engine waveshaper ..increase fold to add some harmonic content
-Enable Low pass gate ( right click enable )
-Set low pass gate response to 40 % , decay to 100 %
-Set all mixer channels to zero except for osc 1
-Now alternatae between NO FM ,2-1 in the oscillator fm routing , make sure the FM amount slider is set to zero
Notice how switching between FM modes actually changes the sound of osc 1 while there is no FM going on , when one of thefm tabs is enabled , it actually shortens the lpg decay ??
Disabling the lpg mode gets rid of it

[baconpaul]

OK here's what I think is happening here

In regular twist we run 4 samples 'at a process time' and then load them into the resampler and so on.

In FM mode we can't do that - we have to run it one sample at a time

I think we have that wrong somewhere.

[baconpaul]

So if I change the sub-block size in non-FM to '1' it is identical to the FM case. But I'm not sure which is the right one. Tricky. I'll have to dig into the euro rack LPG code to find this.