Avenger 2: performance issues and tests thread

[noiseboyuk]

Some people are experiencing some performance issues with Avenger 2, while others have no issues. This thread is an attempt to help track down any common factors by providing a standardised test as far as possible across all OSs and DAWs. The test is based on a video Manuel posted here:



To reproduce this test:

1. Start a blank project in your DAW, using whatever your normal settings are.
2. Add a blank track, unrouted.
3. Add an instance of Avenger on its own track.
4. Open the patch Acid, Factory 2 Library in the ARP section.
5. Close the GUI.
6. Record 4 bars of C2.
7. Reduce fader to -40db (just to stop the audio levels going nuts).
8. Duplicate track.
9. Select your blank track and play. (This is so there are no live tracks).
10. If it plays fine, duplicate all your avenger tracks and repeat step 9.
11. When you get a reported CPU over, or if there are audio glitches or instability, reduce the number of tracks until you have a final comfortable maximum.

When you post results here, post the following info:

OS version
DAW version
RAM
Soundcard
Buffer size
Avenger version
Any other performance factors (such as level of ASIOguard in Cubase)

This is as good as a standardised test as I can think of, but it won't catch GUI related problems. NOTE - there is a patch in 2.0.2 to fix the GUI CPU issues some had with 2.0.0.

Here's my own results

MACBOOK AIR M2
Cubase 13, 24GB RAM, Core Audio buffer 256, Asioguard Medium, Avenger 2.0.2
48 tracks
After this point I got CPU overs but it would still play after a little grumbling.

WIN 7820X (overclocked to 3.6ghz)
Cubase 13, 64GB RAM, RME Babyface buffer 256, Asioguard Low, Avenger 2.0.2
72 tracks
More tracks made Cubase 13 unstable, I'd get CPU overs but it was often playing okay at 80 tracks.

[Lbdunequest]

How are you able to open so many instances? Im getting out of memory with less instances. 32gb ddr5, fl studio 21.2.1

[noiseboyuk]

How are you able to open so many instances? Im getting out of memory with less instances. 32gb ddr5, fl studio 21.2.1

Well that's sorta the purpose of the thread! Why are people having such different experiences?

I have 24gb of RAM in the Macbook, 64gb of RAM in the 7820X. I'll add RAM to the OP in terms of things to list.

EDIT - on the PC, 64 instances takes 35gb of RAM.

[ozinga]

133 Tracks. After that stops with System overload message.

Macbook Pro M1 Max
OS version Mac OS Sonoma 14.1

DAW version Logic Pro 10.8.1
Processing Threads: Automatic
Buffer Range: Medium
Summing: 64bit

RAM 64
Soundcard UAD Apollo x6
Buffer size 128
Avenger version 2.0.3


>>>Update<<<<

Studio One 6.5.1
Avenger VST3 Version
115 Tracks with Dropout Protection set to Medium
67 Tracks with Dropout Protection set to Off

[robert_p]

Here's my own results

MACBOOK AIR M2
Cubase 13, 24GB RAM, Core Audio buffer 256, Asioguard Medium, Avenger 2.0.2
48 tracks
After this point I got CPU overs but it would still play after a little grumbling.

WIN 7820X (overclocked to 3.6ghz)
Cubase 13, 64GB RAM, RME Babyface buffer 256, Asioguard Low, Avenger 2.0.2
72 tracks
More tracks made Cubase 13 unstable, I'd get CPU overs but it was often playing okay at 80 tracks.

These are pretty good results. Could you also share some details about RAM consumption?

[robert_p]

EDIT - on the PC, 64 instances takes 35gb of RAM.

Thanks for the info!

[noiseboyuk]

...and just to add in the Mac, Cubase was reportedly using 30GB of RAM on 48 tracks. Which is quite astonishing, given there's only 24gb of RAM installed!

It's obviously using disk swapping. Amazing that it plays happily.

[Hanz Meyzer]

Nice idea, but the test should require asio guard to be disabled, since it changes the latency. Also you should provide a set a settings as requirements, e.g. “asio guard off, 48khz, 256 samples buffer” or similar. Login also seems to have a similar mechanism like asio guard, which then needs to be deactivated, too.

And why not then use bitwig for the tests? It provides a detailed cpu graph, which then can be posted here.

[noiseboyuk]

Nice idea, but the test should require asio guard to be disabled, since it changes the latency. Also you should provide a set a settings as requirements, e.g. “asio guard off, 48khz, 256 samples buffer” or similar. Login also seems to have a similar mechanism like asio guard, which then needs to be deactivated, too.

And why not then use bitwig for the tests? It provides a detailed cpu graph, which then can be posted here.

What would be the point of the tests if everyone standardised to the exact same setup? The whole point is to find out real world use. That's much more meaningful, because of course everyone works differently. People aren't going to download and demo another DAW.

[Hanz Meyzer]

The point is that you then can compare which system configurations work worse or better. Your mentioned “real world use” is actually more a very individual setup. Bitwig demo is really easy to install, like 4 clicks.

[Teksonik]

Are these CPU issues mostly Mac based? I'm working with a rapidly ageing i7 8700k non-overclocked on Win 10. I haven't noticed any CPU issues and I've spent quite a lot of hours working with version 2 since its release.

Yes I can push the CPU demand on more complex patches but I don't find AV2 unreasonably demanding compared to other synths in my collection.

I don't think 133 instances of Avenger 2 is a real world test but if the point is just to system brag then do carry on.

I'll be working with AV2 a lot going forward so if I encounter any issues I will certainly report them but I'm not going to waste time seeing how many instances I can load before melting down my studio system.

The point of complex synths like Avenger is that you can make more complex patches and therefore eliminate the need to run multiple instances.

When you consider one AV2 patch could replace 2,3,4, or more instances of another synth then the CPU demand becomes even more reasonable in the grand scheme of things.

If there are any issues beyond CPU demand that need testing I'll be happy to test those.

[Hanz Meyzer]

Yeah, whatever... Mac users (and most certainly most windows users) should not buy this update simply, thankfully I didn't... And why should we do the job for Vengeance, doing proper testing? The stubbornness and ignorance of this company and fanboys are unbelievable. Better invest your money in a solidly programmed synth, which does not cpu spike (mostly all others). This is really the one of the worst company experience I ever had. SIgning off, have fun with your alpha state software...

[Teksonik]

Yeah, whatever... Mac users (and most certainly most windows users) should not buy this update

Hey I don't know what to tell you. I'm not having issues with Avenger 2 here and I'm very happy I purchased the upgrade (and use a Windows computer).

I plan on making AV2 one of my most used synths going forward.

If that makes me a "fanboi" then so be it. Here it's an awesome sounding synth that has not given me any issues.

If there were any issues here I would certainly report them to support but....it's working fine. Is it perfect? No of course not, nothing is perfect but I'm happy I did the upgrade and would do it again.

Keep in mind there have already been two updates since the original release so they're obviously listening to user input.

[DrWashington]

I'm definitely having issues with 2.0 upon opening & closing the GUI: it spikes the VST meter in both Cubase 12 & 13 to dropouts pretty reliably.

[Jac459]

Are these CPU issues mostly Mac based? I'm working with a rapidly ageing i7 8700k non-overclocked on Win 10. I haven't noticed any CPU issues and I've spent quite a lot of hours working with version 2 since its release.

Yes I can push the CPU demand on more complex patches but I don't find AV2 unreasonably demanding compared to other synths in my collection.

I don't think 133 instances of Avenger 2 is a real world test but if the point is just to system brag then do carry on.

I'll be working with AV2 a lot going forward so if I encounter any issues I will certainly report them but I'm not going to waste time seeing how many instances I can load before melting down my studio system.

The point of complex synths like Avenger is that you can make more complex patches and therefore eliminate the need to run multiple instances.

When you consider one AV2 patch could replace 2,3,4, or more instances of another synth then the CPU demand becomes even more reasonable in the grand scheme of things.

If there are any issues beyond CPU demand that need testing I'll be happy to test those.

V1 was having CPU spike issues. Not a problem in real life but benchmarks were weird.
It is fully fixed on my system with v2.0.5 (mbp M2 pro).

Other than that, I fully agree with you
I am also using more and more Avenger 2. It is my favourite synth as of today.