Lets create a Cubase 14 community test project as a control for troubleshooting

Due to the amount of performance-related posts, and the concern over Cubase performance vs individual machine performance (i.e, is it your computer or is it a Cubase bug), I think that it would be beneficial to create a standardized project that consists of royalty-free audio, and only uses the plugins that are included with Cubase so that everybody can open the project equally.
The project would be loaded with inserts and VSTis in such a way that it should load and run smoothly on a computer that meets the specified requirements. This way, if any of the users cannot run the project smoothly, we can start isolating not only user issues (that we can address individually), but also helps us pinpoint certain performance issues on the Cubase side.
The second thing it does is give us a base level control for the experiment. It’s exactly the same for everybody and once it is accepted that it can run smoothly on the standard audio computer, we can declare it a control for all performance experiments going forward…until Cubase 15 arrives.

Questions:

How many audio tracks should be included?
How long should the project be?
How many VSTis should be included, and which ones?
Which inserts should go on the tracks?

I’ll be happy to put together a test project if we can start to agree on what the starting parameters should look like. Or, if the group wants to create a project together, that would be even better. Somebody starts it, another person amends it, until it’s considered a good test.

Then, whenever somebody says Cubase is having performance issues, we tell them to run the test project before anything else. If they can’t run it successfully, you get the idea.

4 Likes

One thing to keep in mind is that it can’t just be a regular old project - it has to be considered stressful enough to be worthy of a test, but also not too stressful that most people can’t run it.

Hi,
this is a very good idea because I am not sure if this one is still relevant

Maybe someone else who is more knowledgeable in this area can chime in?

What in your opinion has changed so much to make that standard test used for a time now no longer relevent?
There will never be a test that reflects everybodys way of working but as a standaed benchmark to compare one machine to another it works well imo.

For interest I get 170 Tracks Ur44c and 9/ms RTL at 44.1khz 14700k processor

Hippo

You have just answered your own question. As I said, I am not up to date if there was another set of benchmark projects out there which take different general requirements into account. Therefore, I think that the OP’s idea is worth a discussion.

So to start a discussion what would you think should be the new new way of standard testing always with the general rule of course it should only use the stuff in the cubase “box” other wise its not open to all.

Hippo

I’m not sure, if saying “we should do something” has much chances of kicking off a meaningful project.

The more common avenue to success I’ve observed over the years involves someone investing their own time to implement a first version of an idea. Or at the very least present a reasonably well fleshed out set of details that reflects a deeper understanding of the subject area. And detailed rationale for a new approach, tool or product, and explaining how it differs from what already exists.

And that better approach would likely already include well reasoned answers to the questions in the original post, like “how many tracks” which VSTi’s and FX. And why/how the world would benefit from an alternative to Tom&Dom’s work.

1 Like

@Steve_Day I just did a quick test using the reference test project @Reco29 linked to on my laptop I carry around. Your post peeked my curiosity so I wanted to do something really quick and see the results. See the results below:
I was able to get 165 tracks. After that I started having dropouts.

Laptop Specs/Settings

  • Dell Alienware m18r2
  • Apogee symphony audio interface / Buffer 512 samples
  • ASIO guard normal
  • Steinberg Audio Power Scheme set to on
  • 64GB memory
  • i9-14900HX 2.20GHz

But my fans in the laptop kept coming on periodically.

1 Like

For what is worth:
132 tracks/44.1kHz/256 sample buffer/ASIO guard normal/Metric Halo ULN-8
Hackintosh MacOs 14.7.2
13700k/64GB

2 Likes

AFAIK that test is still relevant.

There are so many possible configurations for a PC that one might want to consider whether a Mac is the solution. The extra cost might be worth it!

I’m keeping my fingers crossed, because I very rarely have any problems with my PC and Cubase usually runs very smooth.

When I see this types of tests, they mostly are covered by asio guards. But you also need to think about what usecase is it going to test? A big mixing session where everything is recorded is very different from a interactive composing session. Interactive is that is in realtime mode.
Im sure if there is any good way to this, but a hardware midi loopback is a working way. Eg, a midi track that is sent to a hardware output that goes back to the computer that runs a vsti.
Sidechain is also problematic so it should go in to the tests. Letting the realtime channel control a recorded part running a vst compressor with sidechain! And some FX-bus with a reverb that is feed from our realtime vsti mixed with other channels.

The point with the loopback is that is should be repeatable test.

I think it´s a great idea!
I suggest a 100 track project, in my experience the lenght is not so significant so it could be 4-5 min long to cover 95% of real projects.
It should include plugins as Vocalchain which is a heavy one.
I´m not sure what to use to stress RAM, usually it´s stressed byt Kontakt libraries or similar…

I don’t think it is a good idea to add non included plugins. Some halion sonic maybe?

Not so interesting for me as my typical productions “only” have about 20-25 tracks. I cannot fathom working on a song with 100 tracks !

And in terms of performance, for many of us it will DEFINITELY be about how those third party VST FX and instruments behave.

But if this test idea works for some of you, great. For me I’m more interested in Steinberg doing its utmost to improve stability with all kinds of plugins… As it seems some are JUCE, others C++, different graphical frameworks, etc,… all of which can cause various issues.,

True,
Build a cubase project with enough midi and audio with max processing with cubase only plugins with increasing channel counts and max plugins would be similar to the DAWbench stress test app of old.
A demo session that would stress the stated cubase minimum system requirement without it falling over would be a good start.
The biggest issue outside of Cubase is audio interface driver interaction with Cubase, which can greatly affect system stability which tools back in the day like ASIO4ALL which gave you asio controls for your built in audio card to attempt to improve system stability when using a DAW( which back then I used with PT, Cubase, SoundForge, Ableton and Vegas Pro when not connected to my audio interface rack…)

The one thing to remember is that we can figure out what impact a component has on performance by exchanging that component. This means that the problem for us is that we have different system with more than one different parameter. If I compare my system to someone else’s the different performance could be because of the CPU, memory, mobo, interface, and so on.

So I think taking in this information and then drawing conclusions from it might be difficult.

That’s the benefit of tests like DAWbench when they are run by Pete Kaine and Vin Curigliano, because since they sell systems they actually test systems where they swap one component or change one setting at a time and can therefore learn a lot more about that one parameter or component.

I would be more interested in how the data would be used in practice.

1 Like

I’ve basically given up on CB . After upgrading computers twice , with the latest one having 64 GB of RAM , I7 processors , it still has persistent dropouts , Audio Performance Meter going red with even one track . Multiple remote sessions with Steinberg and trying everything I’m comfortable with as far as with settings in the computer . I or no one else has sorted it out . So , like I’ve said , I’ve given up at least until i can get someone who know’s enough to figure out why this happens . It’s sad because i really like CB , but it would be like try to drive and your car shuts off randomly and often . Makes no sense to me
Martin Tate

The entire purpose of doing this test is for one reason only: To eliminate Cubase as the culprit. It’s not to compare systems, or see what different system variables there are. The single purpose is to make people run the project who post on here, saying that they are having performance issues. Because chances are that they won’t be able to play the test project either. And when that happens, because the test project has already been vetted as working fine, we now can isolate the issue as being a user problem and not a Cubase performance problem.

@Steve_Day in that case the Dom Sigales/Scan test will be perfectly fine for that purpose.

M

1 Like