Liquid Filter brings my PC to its knees and it begs for help, I pat it and say "there there" but it doesn't improve things... lol.
Hello,
I have been away from Cardinal for awhile but I have come back and tried to load EtherPad from the demo / example projects as a test and Liquid Filter seems to quickly bring my PC to its knees and the performance meter is through the roof for that particular module. I have tried both VST2/3 versions (bizzarely I have no main variant of the plugin in VST2 for some odd reason). It then forces resets.
Anyone else having this problem / know how to solve it?
Thanks.
EDIT: By the way, this is with me using Reaper.

does it always happen with it by itself or other modules, or is it specific to that patch?
I have just been testing further with a simpler patch of my own and it seems to be working fine. Why is that patch so heavy though?
something in that patch triggers undefined behaviour, likely resulting in NaNs which are typically super heavy for filter FXs. I kinda already suspect what is going on, will fix it for 22.08
Ah! Nice one. That would make sense. No worries. :D.
Oh btw, when I wire aftertouch to the frequency CV in it also hammers the performance. EDIT: Or anything even. Such as an LFO
Thnx for testing and reporting!
bizzarely I have no main variant of the plugin in VST2 for some odd reason
Btw, this is because VST2 does not support CV ports (only LV2 and VST3 have a Main variant)
Ah I see. No problem. I will keep testing things. I don't know if it is somewhat related to the same issue but... yes. CVs are hammering it.

It does seem to be the CVs, especially the Gain input. On my computer, connecting anything to the Gain input increases the CPU usage by about 10x, Resonance is 2x, and frequency and FM barely change it. The easiest way to "fix" the example patch is to use the envelope to control a VCA before the filter, instead of connecting it to the Gain input.
Have you tested this on main VCV Rack? If this is an issue upstream it should of course be fixed there.
Have you tested this on main VCV Rack? If this is an issue upstream it should of course be fixed there.
This does not happen in VCV Rack 2.1.2 (Windows 10).
So can you please give a simple example patch that we can use for testing? I don't see any spikes on liquid filter on linux.
So can you please give a simple example patch that we can use for testing? I don't see any spikes on liquid filter on linux.
GitHub can't upload .vcv files, so I just changed the file extension to .txt. Change it back to .vcv before opening it. The issue may be specific to Windows. I don't know.
You can use the export file function, it is just a decompressed json.
I'll try this out.
Works fine (VST3 here):

Works fine Maybe it is a Windows-only bug.
Could be due to missing flush-denormals-to-zero CPU flags.
I dislike setting those on the plugin side, it messes up with global state which is never a good thing in plugins. But we can enable it on standalone, and use it as a test.. Will do a build with this soon
latest release (and even before) now disables denormals when processing plugins. closing ticket as we confirm that was the issue