Unclear calculation of PDC

Hi,

I have some questions regarding the PDC report.

Preparation:
Please install TB Buscompressor v3 VST for macos (intel only), use an arm mac machine. Also MRatio VST3 by Melda Production. Renoise 3.43

Open the following song file:
bug-pdc.xrns (9.7 KB)

Load it into Renoise, disable sandboxing (though bus compressor will run in cpu bridge).

  1. Open PDC compensation report, looks like this here:
  • Why is the track “sn” split into input and output delay, because of the sidechain send?
  • Why the track “rev kick” only having 767 samples?
  • Why are all lead tracks getting the same input latency and not the group “front” instead? Could this lead into issues while using multiple nested groups, in parallel? Isn’t input latency also delaying meta signal input (e.g. env follower)?
  1. Now mouse-over an info icon of a tb bus compressor instance:
  • Why is it reporting less latency (512 samples) than the PDC dialogue (767 samples)? Which value is the correct one, is the PDC report maybe faulty? Same happens for example with Uhbik-A plugin. Or does the bridge add additional latency? If so, shouldn’t this be reported in any case?
  1. now remove the track “rev tom”. Again see PDC report:
  • Why is the track “sn” not anymore split into input/output latency? Because sidechain latency is the same as the other tracks (having input latency)?

BTW. I think it would be cool, if the report window was not exclusive/blocking popup, and instead, it would refresh the report each 2 seconds or so, and you could live edit the song in parallel…

Ok, now tried with non bridged arm native plugins, and then it reports the exact same latency everywhere.

I guess the bridge always will add 256 samples extra buffer… This seems to be quite a lot, especially if you use bridged plugins in serial. Will easily add up then. Can’t this be reduced at least for arm macs?

BTW. it might be wrongly calculated then, for example:
Uhbik-A uses 16 samples + 256 samples = 272 samples. Yet PDC report tells me 271 samples… Rounding error or intentional?