I hope it is understood, I am using the Google translator
I use VST live and in each song, a few parts, each with stacks with Neural DSP plug-ins insert, amp simulators, for guitar and bass.
I will try to be detailed with the problem.
I create a project with three songs.
Stacks with vst Neural DSP on all parts.
I apply preload parts.
Everything works well.
I create a project with 10 songs.
Stacks with Neural DSP on all parts.
I apply preload parts.
Starting from the fourth or fifth song, it takes a while for the sound going into the stack to come out.
I press Stop.
I manually step through the parts of the fifth song (Something similar to what the preload function would do)
I press play.
Everything works well.
So successively with the other songs on the list.
Change the order of the songs.
I put the latest songs first.
I apply preload of parts.
The ones that are now the first work well.
From the fourth or fifth song, the same problem.
Manually stepping through the parts solves it.
It’s like the songs that are from the fourth place on the list, they forgot the preload.
This happens only with the Rabea and Cory Wong Neural DSP plug-ins.
With Darkglass everything ok.
Anyone using Neural and a similar workflow?
musicullum has analyzed the problem a lot, and assumes that it is a problem with the plugins.
The CPU load does not exceed 20 percent.
Tested with Mac Big Sur iMac i7 32 GB Ram RME Fireface UC Buffer 128 or 256 (We also tested it with larger buffers and the issues were the same). Also tested with Mac Studio Ventura M1 Max 32 GB Apple Silicon Native. Same results.
Tested in all versions that came up until today
Finally found some time to test this.
I inserted the Cory Wong into every Stack I could find up until and including Song 7. Also assigned ports to all of your audio inputs and outputs. That took quite some time…
Then tried different things and scenarios. Preload Parts, run from Song 4 or 5, all good, no delay. I will let it run from the top later when I’m away…but if I got you right, I should get the problem when
load project
preload Parts
select Song 4
run
Correct?
I do realize that all those plugins eat a lot of memory. How much memory do you have? Check Task Manager to see how much memory is eaten, and close other applications that may eat memory. Shortage of memory may well affect performance. Maybe you can use some more Global Stacks.
We also found a bug with Stack Inputs which is fixed with the next update, maybe give it a try.
If you still want or need to continue this, it would be great if you could modify your test project to only use the Cory Wong - you only need to send the .vlprj file then, other files are already here.
Also as I said before, Neural DSP does not allow to run their plugs when a debugger is present which makes it impossible to try to analyse the situation at all. That leaves us with wild guessing, but first it would be good if we could even reproduce it. We may then have to contact Neural DSP, maybe they have answers.
Ran the entire show (set all Songs to Next Song and start) and yet no delay. That was on a Windows Laptop though. I see you are on Mac, can you get hold of a Windows machine to test, just to narrow it?
Thanks for trying. I tested it on my three machines, all Macs. One with i7 and 32 GB of RAM, another, Macboock AIR M1 with 8 GB of RAM and another MAC STUDIO m1MAX, with 32 GB of RAM. The CPU meters mark a maximum of 30 percent consumption, and the Ram memory marks 16 gigabytes free on systems with 32 Gb. The error appears on all systems, at approximately the same time. It doesn’t seem to be a problem of that type, since the tracks run smoothly, if I use other stacks simultaneously, with other plugins, they continue to work correctly, except for Cory Wong and Rabea, who have the problem. No click is heard, or audio jump, only that the audio takes time to come out, sometimes, from the stacks with Cory Wong.
I use Global Stacks for vocals, but for bass and guitar I use Stacks in the parts, since there are different sound changes (Clean, distortion, solo, etc).
I can try to reduce the number of parts so I’d use less stacks, but the current workflow is perfect for painstaking and detailed sound production. And the truth is that I can rehearse this way, I just have to manually go through the parts before playing the song. If we ever manage to fix that, it’ll be really cool, for great live sound production.
I have a laptop with windows, I will try to install the programs and test it when I have time,