Banging uzi
As you can see I am writing a cosine into a multislider
The yellow numbox defines pi radians ,and it's also banging the uzi for a (near) realtime update , in the correct order ( banging is happening last )
Issue is that I still need to manually bang the button again to get the correct cycles .
Notice how 2pi radians represents a faulty shape

And hear I manually bang it again , now the shape is correct

This is worrying since I want to control the amplitudes of mc.filters this way
In pure data none of those issues appear , I can bang an uzi at near audio rate and write into an arrray without a glitch
I can hardly see your code at all in those screenshots. But my guess is you are doing something wrong with zl.group.

you can not count a numberbox to output all numbers between x and y when you drag it with the mouse, you only get what the mouse update frequency offers.
there are allways some limits when using mouse controlled GUI objects.
But not that bad as in your screenshots.

You can speculate about what the problem is. But I rebuilt your code with my solution using uzi's "done" bang to output zl.group and that sure makes it seem like your method for zl.group usage ended up with about 32 "overflow" entries at some point. Which means you forever have 32 overflow entries which pushes your list out of "phase", because you've never purged zl.group's contents by bang or zlclear.
I also added another solution for giggles that caches the 128 item list.

Here's something funky ( right click pitcure and open in new tab )
A two operator phase modulation structure with feedback on the modulator , all done in the scheduler and written into the multislider .
The uzi needs to be rebanged to show the update in the multislider .
It does it but not verry fluently , pure data otoh has absolutly no problem with stuff like this , the uzi can handle anything you throw at it while tweaking a numbox at the same time

( right click pitcure and open in new tab )
i would prefer to be able to click on 'Copy patch' and past the result as a new patch in Max.
Could you 'Copy compressed' your patch and paste the result here so we don't have to rebuild it from gigantic screenshot in order to test things on it? Here's how to do it. And please make it an habit for when you post here.
that is not really optimal patching.
you have no idea what triggers what and when.
One should put all factors or parameters into a list and then
make calculations in optimal order.
typical mistake is make one construct and the add and add on top of it,
not taking really care about the flow.
Sorry guys , I can't use copy compressed since the demo trial period has expired
I am aware of that , it was just to showcase operator feedback in the schedular , and the multislider shows it correctly
A 2 operator ratio 2;1 square wave , with operator feedbackon the modulator