When you uncook a COMP with the X button, what changes under the hood besides the obvious? I’ve tried looking this information up in the documentation but wasn’t successful in finding it. I have a big .toe with 95k OPs in it, so I feel like even if the difference between an un-pulled and an un-cooked COMP is small, it may not be unimportant, so I’m trying to better understand.
My understanding that nothing is unloaded from RAM, but is anything unloaded from VRAM?
Are additional CPU/GPU cycles saved, even if nothing is pulling the COMP to begin with? Or does an ‘un-pulled’ COMP truly use zero CPU/GPU?
Can OPs storage be altered inside an uncooked COMP? (eg. appendRow on a DAT)
Are there any other differences, no matter how seemingly small?
I created a Base, dragged a few palette toxs into it, and turned off the viewer. In theory, it’s not being pulled by anything outside of itself, but it is still cooking something inside of itself (looks like videoPlayer palette).
If I copy paste it, and press the X to un-cook, then it seems truly inert.
I have many times heard and read the advice that you should never have to manually turn off cooking. But what I’m seeing in my example is that even if I’m not pulling/referencing a COMP, I could save a bit of CPU by manually un-cooking it. Am I off in my thinking here?
There are a bunch of cases where things inside cook, even if you are not connecting the outputs to anything, nor displaying the viewer, nor fetching a node inside with a select (CHOP/TOP/SOP/COMP…).
For example, Audio Device Out CHOPs will cause cooking to generate its audio streams, as will other OPs that output to devices like laser etc - some of these OPs have Active parameters to control whether it cooks always or not. Folder DAT, OP Find DAT, Timer CHOPs… just some examples.
To analyze what’s going on inside, I think “probe” is good - it doesn’t interfere much if at all.
A wiki page may be useful for this. Some of it’s here:
