Cubase 13: random CPU spikes still here, is maddening

I wish you all the best :+1:t3:

1 Like

Thank you. Sincerely appreciate that.

@DrWashington again, I’m trying to help here. I think the issue is YOU are having issues and MOST overyone else isnt.

I’ve just spent the past 2 days doing sessions with a film composer for a new film and he was running an old 5950x machine with RME interface and windows 11 24h2 and everything was a smooth as silk.

Large template projects with lots of Kontact and various other samples, various VSTi soft synths etc etc etc , Video being played back on separate screen. Tracking live intsruments along to picture.
C14. worked perfectly… as it does in mine and all my other Cubase professsional colleagues Studios.

So what I’m saying here is for MOST of us… and I’m talking about working professionals as well not bedroom beat makers, Cubase 14 is working well enough we’re all getting our work done and delivered to clients with few issues.

My post was in response to me thinkling about what you said and you mentioned a VERY SPECIFIC scenario that was causing a problem…and as I’ve said your issues must be very specific as most of us don’t have the issue.

Again, I hope you can find a workflow with whatever DAW/OS platform that serves your creativity.

M

1 Like

@Norbury_Brook - look man, you’re not helping (even though you claim you’re trying to).

I’ve already explained why you’re wrong multiple times above and am not in the mood to reiterate. I’m so happy for you that things are working out, but fact of the matter is your world is only anecdotal to me and I know for a fact other Cubase users are facing similar issues. I suspect it may be due to the Intel 13th & 14th generation core architectures, but I don’t know for sure. It could be a number of factors. That’s why I’ve been proposing the solution I have that I also don’t feel like reiterating for the 4th or 5th time.

So, I exhort you: go enjoy your system. Be productive.

Saying that only @DrWashington has issues and nobody else is a lie. Many people here, including me, had a chance to start similar topic. I can also understand @DrWashington because I’m in the same position: 18 years working on Cubase without issues, investing money, loving the ecosystem and so on, I can even say that I became Steinberg’s fanboy some time ago. That’s why I expect the professional grade product and don’t want to switch my DAW to other. I don’t want to duplicate my other posts so I just add that Steinberg does not meet the statement they did in the user’s manual regarding multi processing feature. I mean this one:

multiprocessing

This is how CPU load looks:


These clearly don’t go together.

1 Like

And yet

I’m not sure what duplicating a bunch of Retrologue tracks is supposed to prove: these issues are happening with third-party plugins. There’s a larger issue here and that is that Steinberg of course controls the VST3 plugin format and the parameters of the same.

It’s pretty clear that certain developers are doing things they shouldn’t be doing but the architecture is allowing these things to destabilize the entire DAW (crashing), introduce CPU spiking (performance issues to the point of unusability if you’re one who needs that sense of immersion to achieve flow state, as I do), etc.

My overarching point is that at the end of the day, the buck stops with Steinberg and they’re going to need to get better at stabilizing their environment or suffer the economic consequences. I have been putting up with years of this and it’s almost destroyed my desire to make music in general at this point. It’s very mentally and emotionally taxing to keep encountering so many bugs that never get addressed: I just want my DAW to feel solid, stable, performant, and dependable. Then I can relax. Then, flow state can be achieved and I can feel productive again.

First order of business would be for Steinberg to develop diagnostic tools that can be distributed to their customers, even built in to the DAW, and have telemetry options that allow Steinberg to see exactly what’s going on remotely so they can actually pinpoint and address the problem.

I can’t believe I have to be the one to actually lay this out, but this is where we find ourselves, because the beta testing they’ve been doing is clearly more like alpha testing and WE end up being the true beta testers–except things don’t end up actually getting fixed. Still. After years of complaints on these forums, so many bugs that they have to have internally acknowledged and know about still have not been given priority.

Who are these PMs who are determining what is actually priority, I have to wonder, and why are they so out-of-touch?

1 Like

[quote="I believe you
You don`t understand
If you cant reproduce the test, poven on many machines , you set up is starting from a lower base line than most and that has to to be put right first before you can prove any futher statements regarding the software.

1 Like

There is always a risk if you absolutely trust synthetic test rather than real life scenario. The test which results you are posting has different background than mine. So, there is a result of another synthetic test that Cubase fails. This is more towards my needs than yours as we can see. I didn’t use any VST to make this test even more lab-like.

I love Retrologue, Padshop and Halion but I use them to give my music some additional dimension. I would like to record guitars in real time with the best possible latency supported by my hardware.

There is always a risk if you absolutely trust synthetic test rather than real life scenario

I would never rely on it for any thing else than proving the harware and system set up
I dont think you get the real point , if the benchchmark , proven to run flawlessly on cubase and all other other well set up setup machines, will not return good results on yours there is a non cubase problem local to you problem.
When it runs well then you can start with the other issues

1 Like

I couldn’t disagree more with this way of thinking. I’ve already hammered on the strategy I would use if I were Steinberg to ensure broad compatibility so I’ll not reiterate.

The best I can tell so far is that the issues I’m facing have something to do with the way Cubase is handling graphics that neither Ableton nor Bitwig do: I’m simply not encountering the same issues on them.

Its been so long what is this specific graphics problem?

If I knew exactly what it was, I’d address it, or at least name it. All I can tell is that certain plugins are much more likely to cause CPU spikes than others and it sure seems to be related to manipulating things on them graphically. The exact same plugins cause no issues in Bitwig or Ableton.

Edit: but then, here I am, just created a totally new project with nothing running at all and I’m seeing the real time CPU meter jump almost up to 1/4. With nothing there. It’s just playing back nothing, recording nothing, and jumping.

Dear Mr Washington
Never seen or heard anything like that on a proper set up .
I still await your step 1 2 3 or YOUR demo project or down loadable or video examples.
So far only words and rants, nothing practical.
Take on the challenge and do the ā€œstupidā€ standard test and pass it.
Publish your result
Then and only then after your set up and machine is proved to the universal low minimum then please provide your step 123, demo project or down loadable or video example of your specific problem so that the rest of us can have a go at your problem.
On a personal level I hope you never do.
You brighten up the end my day with your rants

Sad I know but Ive already made my money in a large part since using cubit from 1980s and enjoying a well-earned retirement.

Until tomorrow

Hippo

1 Like

Dear Mr Hippo,

I’ve already put a bunch of practical stuff up years ago, crash reports mostly.

I’ll probably get to your test at some point, but I’ve already gone down a deep rabbit hole here the last few days and am actually making some progress. I found an old UAD forum post from 2008 that mentions disabling Cubase’s CPU affinity for core 0 but leaving all other cores ticked in the Task Manager. This has made a very significant difference, indeed. Seems something is competing for core 0.

I just installed Windows 11 24H2 last night and uninstalled a bunch of older software and drivers I don’t need. That didn’t seem to make any difference so far, but this has.

Ever so glad you’ve been enjoying both my rants and your well-earned retirement!

But, I’ll say it again: proper built-in diagnostics with telemetry and possibly some well-considered ML could likely make short work of tons of various issues that hit certain setups but not others. I think it’s pretty apparent that ordinary beta testing isn’t cutting it, that there’s a fairly large contingent of people who hit things like this that can be especially hard to diagnose.

Sir, please set the lowest possible buffer size on your UR44C (isn’t it 32 samples ?), then add one audio channel and set monitoring on it. It’s the scenario from my last post. Your results probably would be better than mine, because of the minimum buffer size possible on your hardware. I have UR22C and the smallest buffer possible to set is 32 samples. Going further is unusable in Cubase.

Not sure what you want but this is what i came up with but dont see the point

So, using Process Lasso to disable Core 0 for Cubase seems to be working a treat so far: whatever this badness was, it was attacking Core 0 and that was causing dropouts. It’s still not perfect, but it’s at least usable again.

Hope this helps someone out there and really hope Steinberg looks into this and begins to take it more seriously: something’s wonky, nothing should be allowed higher priority than Cubase’s audio stream on Core 0 or any other while Cubase is running. Whatever is doing this, it should be forced at the system level to wait its turn, and Cubase should be able to install whatever it needs to to enforce that, IMO.

2 Likes

I don’t see ASIO performance meter on your screenshot so I can’t see the point either.

Try this
150 vst tracks playing at the same time + 5 audio playing and 1 activly recording + 32 buffer + aiso meter
Now you told me what you actually wanted to see i provide it.

Processing: For forum 2.jpg…