I have been using gigperformer live with neural dsp plugins.
The setup i have at the moment is running 1 rackspace with 4 variations, in the rackspace i have 4 different neural plugins each plugin is designated to each variation, it is setup in a way that each variation bypasses the other 3 plugins so that i only have 1 neural plugin turned on at once as they are very cpu hungry.
Also i change the presets via midi pedal in each of the neural plugins but i am experiencing very slight lag when i change presets.
Now what i am trying as an alternative method but having no luck is this.
Instead of 1 rackspace with 4 variations i am trying to use 4 rackspaces each with their own neural plugin, this is beneficial as i can set up tails between each dsp and also i am hoping that there will be no lag.
The problem i am having is that all 4 plugins are turned on all the time and it is too much for my cpu, is there any way i can have the plugins bypassed until i hit a rackspace and for it to then turn off when i hit another rackspace.
Thanks for reading the post guys.
A plugin in a non active rackspace might consume a certain amount of RAM but shouldn’t use much CPU power. Same as a bypassed plugin…
That doesn’t make sense - if the plugins are in different rackspaces, then when the rackspaces are not active, the plugins should be bypassed and not taking up any CPU cycles – at least for the audio processing thread, which is the only one you care about.
Edit: what @schamass said!!!
Hello @Cagney, welcome here 
Just to confirm what has been said, start with a single rackspace and note the cpu value, then duplicate that rackspace as many times as you want and look at the cpu value in any rackspace…
Thanks for the answers guys, thevreason i thought all the plugins were on is because when i change rackspace the cpu shoots right up to about 70% which leads to the sound breaking up before it settles back down.
What is your sample rate and buffer size?
OK… before we proceed, please tell us all about the system you use:
PC/MAC?
WIN 10/11? MAC OS?
Which audio hardware?
Which audio drivers?
sample rate & buffer size?
On my system PC i7/4GHz / 64GB RAM / WIN10 / Scarlett Solo USB-interface / ASIO / 48kHz buffer size 256
i set up three rackspaces with a diffrent NDSP-plugin in each (Cory Wong/Plini/Nolly) and the average CPU load indicated in GP is about 15%-19%, which about doubles for the time of the audio trail between switching the rackspaces (say ~36% max).
Well… what i try to say, is… everything is relative!
Maybe your system is too weak (which we don’t know yet) or maybe there is room for improvement, which we don’t know also, since we don’t know anything about your system and use-environment!
So… please tell us more! 
As @schamass mentioned, GP allows the audio tails (e.g. delay/reverb) of the prior rackspace to continue for a few seconds when you switch to the new one. So technically, both rackspaces are active for the transition. You will therefore need some additional CPU available to support this. Which could either mean using more cpu friendly plugins, fewer plugins, or a higher spec machine.
Sorry it took so long to get back to you, ive been working away, looking at your spec compared to mine i dont think i have enough grunt🤣
PC i7-3520m cpu@ 2.90GH
8.00GB ram
Win 10
Presonus studio 24c usb interface
Asio 44100hz 64 samples (1.5ms)
Plugins Cory Wong, Tone king, Gojira.
One difference i have noticed in my setup compared to yours is you are running at 48000hz, mine is 44100hz, does this make a big difference?
Thank you for taking the time to read and answer my post.
Try 128 samples and 44.1k
48k requires even more cpu cycles