Compatible mode simply ensures that two instances don’t process at the same time. It’s mainly required for some old/poorly implemented plugins plugins that choke if two instance run at the same time. It’s not really related to the execution model I describe above and simply locks out the other instance until the first instance is finished processing.
The only difference between aggressive and compatible mode is if two instances of the same kind of plugin happen to require processing at exactly the same time. Two volume plugins at the end of two different chains are probably going to execute at different times anyway (unless both chains take exactly the same time to execute) so it’s not that big a deal. It’s more of a problem for heavy/slow plugins running in parallel.
Having said all that… you can override the compatible/aggressive mode for individual plugins in Cantabile’s compatibility.ini file - which is in the same location as the Cantabile.exe file.
; Valid values:
;
; 1 = Always serialize processing even in Multiprocessor Aggressive Mode.
; 2 = Never serialize processing even in Multiprocessor Compatibility Mode.
So say you know a plugin is cool to run in aggressive mode, grab it’s unique id from Plugins.json in your settings folder and add an entry to Compatibility.ini like this:
PluginXXXXXXXX=2
where XXXXXXX is the plugin id.
Or, set Cantabile to aggressive mode and if you suspect a problem with a specific plugin add an entry with value 1 to force it’s execution to be serialized.
So what would you more technical users suggest for something like Diva? My unit has plenty of power… i7 and 32GB of DD4 RAM. I still seem to get some crackles on the “great” setting if I push the voices.
I would take a look at the (new) profiler. Try the double buffer option, or increase the latency. Next, do you really need Diva’s “best” sound for live playing? Maybe consider a lower (less expensive in CPU cycles)? And, this assumes you’ve already followed the best practices outlined in the forum.
What are your latency settings in audio driver? In my case I‘m good off with 256 samples. Additionally: the point cracles appear is strongly dependent on the soundcard and its drivers! In my case an RME babyface is doing a good job.
Do you use multiprocessor mode inside DIVA? This really helps as it is the only way to make use of multiple cores on one instance of DIVA.
As already mentioned: Quality mode „great“ is really high on cpu. For me „fast“ is absolutely sufficient for a live situation.
if you use more than one instance of DIVA at a time agressive mode in soundcard settings can help. But be aware: you should be sure this doesn‘t introduce instabilities with other plugs: so first investigate.
Finally: I add a binding to every song that quckly restarts engine after sing load. In some cases this can reduce cpu load by 40% if multiple plugins are used in one song. Since I‘m easily able to layer 4 instances of DIVA as layer.
