Nuendo 12.0.7 CPU Spikes when Monitoring VSTi

Hello all. I’m no stranger to Nuendo, I’ve been with the program (and with Steinberg) since N3. I’ve dealt with my share of CPU spike issues, but this one is a head scratcher. I know a lot of people have struggled with these in C12 / N12, but I thought I’d share my situation.

By and large, my sessions run great. The only time when I have severe CPU spike issues is with virtual instruments. Specifically, when monitoring or recording them. I can play back tons of them. I can run 100+ track sessions with very minimal spiking (at least, nothing out of the ordinary). Typically I run at 256 buffer. But for the life of me, I’ve tried everything to figure out the CPU spikes when monitoring virtual instruments. I put ONE instance of Omnisphere in a totally blank session and I was getting up to 70% on the CPU meter when monitoring. If I recorded and then played back, however, it was barely 10-15%. And as we know, the Windows Task Manager is always going to show a bit less than what’s in Cubase/Nuendo. So when N12 would show 10-15%, Windows Task Manager showed about 6-9%. Nothing too wild there.

And for the record, I have tried Omnisphere, IK Multimedia B3X, Serum, Lokomotiv, a TON of different VSTis, and I get similar results in spikiness. But again, only when monitoring/recording. It’s pretty darn stable when playing back, to be honest.

One curious thing - a lot of folks claimed “it was so much better in N11/C11” but one strange thing I noticed was that the Performance Meter is VERY different in N11 than it is in N12. The one in N11 seems to only have an “average” and a “maximum,” and the meter we see on our mixer or whatever tends to be more indicative of the average. However, in N12, we have 4 meters - Real Time, Peak, Asio Guard, Disk Cache (if I’m recalling correctly). And the meter we see on the mixer is…well…sort of whichever one is doing the most at that time. If ASIOGuard is working hard, our meter will show that. If real time / peak is working hard, our meter will show that. So, despite the claims of things working better in N11, I don’t actually think the performance difference is THAT different - I think we may be interpreting the meter readings differently.

For example, In N11, my realtime peak when playing Omnisphere goes crazy. But, my mixer meter only shows my average. So it APPEARS to be more stable. In N12, the mixer meter shows whichever one appears to be doing the most work, and so the realtime peak goes crazy and so does my meter. Now yes, N12 does seem to use a little bit more CPU usage overall, but not by much, at least on my machine. I can open full mixes in N11 and N12 and have similar CPU readings.

I guess what I’m saying is - I don’t know if it’s an issue with 3rd party VSTis, or Cubase/Nuendo, some of both, but I really hope it gets resolved quickly, and if anyone has any input to my specific situation as to why my CPU ONLY goes crazy during monitoring/recording of virtual instruments, I’m all ears if you have a potential solution.

Intel i9, 10 core, 32GB RAM, Win10Pro, nVME drives, AMD Graphics Card. etc. Nothing out of the ordinary in terms of temperatures, overclocking, the typical tuning/settings we all do to every audio PC. Same stuff we’ve done for 10+ years. Lynx Aurora N interface, most recent firmware and drivers, everything is as up to date as it can be, including Windows, Midi Keyboard, iLok, etc etc


Once you enable the Monitor or the Record Enable on the track, the track switches to the “real-time” mode. It means, it doesn’t use the ASIO Guard anymore and all is processed in real-time. Therefore it needs more CPU usage.