SVF3 filter varying from 10% to 30% processor level
Archive · 20116 old posts
From the original SynthMaker forum (2004–2014), with its members’ names and dates.
datsound ·
Synthmaker 1 - When I press middle C it's 8% and A# is 30% !!!
Also as I modulate the res of 5 Svf3 filters [bandpass] from 0 to 1, for example res .33 uses 8.9% CPU, and res .40 uses 30% CPU-it uses that much because there are five filters set to the same res.
it also depends on the signal going in to the filter, if I choose all the same oscillator shaps to all filters it's okay and if I mix different ones to each filter with a scanner it is goes irregular .
I tried to make a demo diagram to show you but I didn't manage yet!
Also when I reinstalled Windows XP, my synthesizer went down from 28% at most times to 8% at most times-it's crazy! Must be something to do with the libraries that Synthmaker users?
I will try again to make an example, but has anyone got any ideas what is up with the svf3?
trogluddite ·
Weird.
I've had no joy reproducing this on my system. I used just a mono osc + 5 * SVF3 with shared cutoff & resonance, both serial and parallel arrangements. The CPU remains pinned at about 4% (Q6700 quad core) - However, that's on SM2 which may make a difference.
My first guess would have been denormals - but the filter code appears to have denormal removal built in. Is there anything after the filters that might be affected by changes in the filters' output?
datsound ·
trogluddite wrote: Weird.
I've had no joy reproducing this on my system. I used just a mono osc + 5 * SVF3 with shared cutoff & resonance, both serial and parallel arrangements. The CPU remains pinned at about 4% (Q6700 quad core) - However, that's on SM2 which may make a difference.
My first guess would have been denormals - but the filter code appears to have denormal removal built in. Is there anything after the filters that might be affected by changes in the filters' output?
nice rig! Thanks a lot for trying to reproduce the bug. It's actually a really complicated synth and I'm not sure how to isolate just the filter and oscillator combinations. The filters are added and i tried sending straight to the mono combiner but the result is the same. It must be a SM code bug that is very particular to the structure and the svf3 module. fortunately all the other filters seem fine. It seems it must be denormal, even though the inputs of the filter are all ordinary floating-point numbers.
MyCo ·
The denormals may occure internally by multiplications with small numbers. Maybe they aren't passed to the output of the filter, even.
You can check for denormals with this tool:
A geek toy: Stream inspector
datsound ·
MyCo wrote: The denormals may occure internally by multiplications with small numbers. Maybe they aren't passed to the output of the filter, even.
You can check for denormals with this tool:
A geek toy: Stream inspector
Hello, time for a total overhaul of the synth for me, that stream tool is AWESOME!
...Problem solved! one of the inputs from the waveform scanner was not connected and i didnt see it, so the svf3 had an unconnected stream.
sorry i am just getting to grips with the troubleshooting routine in SM, have to copy and paste stream readout upstream until the denormal is found. Thanks!!!!
datsound ·
i was sure that the svf3 had a bug because the cpu trebled when i changed res! actually it had no stream in data!
Wrote something here? To have a post removed or your name hidden, write to [email protected].
SynthMaker