I always spend a couple of months checking out the latest known working system specs when I build a new DAW. This has meant I’ve not had issues. I took a chance when i built my latest 7950x AMD machine as it was all new, but having had a 3950x then a 5950x machine over the preceding year or so I was confident it would be OK. … It was …the 7950x is a stunning DAW. Powerful, Silent and stable as a rock with windows 11 and RME audio hardware. It has been runnign Reaper for the last 18 months as C12 wasn’t as stable as I needed. C13 seems better for me already.
if anyone is interested my old 5950x machine is for sale at the moment.
P.M for details.
To follow up, I begrudgingly updated to Windows 11. Tried it last year and the process failed: I got a black screen. Tried it again, didn’t really change anything, and it somehow worked. NVIDIA driver issue, perhaps. Don’t really know.
Anyway, performance is way, WAY better so far: much more stable VST performance meter, load seems significantly more evenly distributed across all 32 cores (I’m running an i9-13900K). With Windows 10, it was parking a lot of cores and only ganging up on the performance cores, overloading them.
So I guess the process scheduler has improved markedly in Windows 11. I’m cautiously optimistic here: this is really seeming a lot better so far. I can’t get my taskbar where I want it (left side), but there are worse things, I suppose. Just want to get some music done without these random spikes.
Hi! One month ago I found all that great world of plugins and Cubase. I was so excited and greatful. It was really all I needed. I started composing something big. But without my friend’s help I think I could be mad and seriously depressed right now.
How is this possible? I used a lot of money for Cubase and plugins and I can’t do anything because of the peak problems. I’m glad my speakers are alive. It’s been a war here because of peak problems.
Steinberg’s quality? Is there one?
I lost one month of my busy life and I need technicians to go further.
Is this really ok? How Cubase compensates this?
And see, this goes back to my thread-opener: “Steinberg. Fix this.”
I’m so, so very tired of this falling on me to troubleshoot my system when it’s so clear that Cubase’s engineering is falling down here. If they need to entirely re-write the audio engine for modern computers, then they need to totally re-write the audio engine for modern computers. This is insane.
Their competitors at Ableton clearly have way, way more of a handle on this, as do Presonus from the anecdotes I’m hearing (but which I don’t own).
Respectfully, I really don’t want anymore advice on things to try. I want Steinberg to make real, substantive change in their performance engineering and testing.
Cubase 13 is a nightmare to use, top to bottom. Do anything of any complexity and the random spikes show up: your VST meter is chugging along at around 30-40% for 30 seconds or so and then spike! It’s absurd.
There should be no way any VST plugin should have the authority to make the host program so unusable for any reason. I am tired of the excuses and I’m tired of the endless troubleshooting.
This is my last shout into the void of the internet. Software companies are making us all insane, more so every year, because they’re not doing their effing jobs by releasing stable, robust products. Gotta get those new features out the door as quickly as possible, or all the money dries up!
No, all the money dries up when you continue to abuse your customers’ trust and a horrible reputation starts to follow you among professionals. That’s what’s up.
You’ve seen my old thread about the spikes yes ? I’m also a gobshite on the forum yes ?
I have never been able to run a Cubase with anything less that 512 while tracking before a complete reinstall and CPU upgrade yeh ? Always constant spikes or some sort of bottlenecking , be it graphics or what nots BUT , biting the bullet and having a pro windows install with a newly flashed BIOS has wielding great results . This is an old machine still thou 9900 BUT for the first time EVER i can now work STABLE with C13.0.20 and the latest mediabay download of 128 buffer and 80 tracks and loads of FX , yes the ASIO meter jumps about like crazy in real time but if i increase the buffer to 1024 samples , It’s stable AF NOW , this is a big jump forward .
I noticed somewhere that you have updated to Win11 ,? was this an update or a fresh install ? As there seems to be a lot of conflicting reports of things being trashed or broken taking the upgrade via win 10 path … Can you see what i’m getting at ?
I would really love it if Steinberg via Yamaha perhaps would maybe band together with other developers in the DAW space to put joint pressure on Microsoft and NVIDIA at the bare minimum to address this stuff. It’s really obnoxious that we can’t use stock power plans as-is and feel any relief from these latency spikes.
In 2024, I don’t think there’s any possible excuse for trampling all over the real time audio stream by any other driver for any reason whatsoever. Audio is as basic as basic gets and its consistency should be considered sacred already.
“Thall shalt not interrupt the audio stream for any reason whatsoever” should be their mantra, and if they won’t enforce it willingly, they should be made to pay a real financial penalty when they screw it up. Enough is enough. As consumers, we’re fed up!
Good day Tj99 ! I had this problem with the spikey sound in Cubase 13.0.10 but after I updated to Cubase 13.0.20 the problem was solved, at least in my studio pc. Before the update, I had this weird problem which I will show you right away. The problem was that my Universal Audio Apollo X would not overload my Asio Guard but it was overloading my CPU with the message Processing Overload ! Click to Reset Display
Microsoft needs to improve windows audio further, and ASIO needs to go away. Right now Windows knows how to prioritize Pro Audio threads. The issue is Windows Audio is not good for people who need beyond Stereo I/O and a lot of Pro Audio software doesn’t support it.
We need to move towards a native driver with class compliant devices, like on macOS. ASIO drivers themselves are an additional potential problem point on Windows. It’s a roulette.
Many older DAW code bases may have issues moving to Windows Audio, but others who are better set up for that can bridge the gap for many users.
I have an RME Fireface UFX and it’s been rock-solid everywhere else but Cubase, but I agree: it’s time for Microsoft, Intel, NVIDIA, etc. to get really serious about professional audio on PCs. Apple just makes too many unnecessary, breaking changes for no discernible reason other than “buy the stuff you already own again, mmkay?”, so there needs to be a viable, competitive alternative.
I’ve gotten my latency way, way down by employing so many various tweaks I can’t even remember them all, and for the first time in a long time, I had a long session yesterday without any spiking. My “Ultimate Performance” power scheme has been tweaked with PowerSettingsExplorer to not promote/demote; I’ve stopped core parking (forget how at this point); and on and on and on. So now my PC is burning lots of energy unnecessarily, but at least I’m getting some work done. Yay.
Oh, cool: the spiking is back. What drives me crazy is the randomness of this: things are looking just great, around 30-35% VST meter on average on this project. I just get these large spikes every so often seemingly out of nowhere, then like every fourth or fifth one clips it. So far, audio hasn’t seemed to glitch much today which is a big relief, but having to reset the meter is all sorts of annoying after a while.
I’ve said my piece on this thread already as have so, so many others. Thanks everyone for chiming in: I’m hopeful that Armand and others are taking this seriously now and that the performance team will start to hold themselves to a higher standard.
The fact is, I just can’t recall Ableton behaving like this. VST3 and Cubase just need to get smarter about distributing load and demanding from the OS the resources it needs when it needs it, or even “planning ahead” in microsecond or millisecond terms so that it can get processed what it needs to before there’s a spike.
I know this may be hard, but engineers do hard things: that’s why they’re paid to be engineers.
… aaaaaand, it just crashed right when I was very much in the zone and feeling incredibly inspired by my (very seriously underrated) DSI Mopho’s out-of-control filter + feedback + other stuff. Just got a Studio Electronics synth I’ve been playing with all weekend that I’m already quite in love with, too.
All I did was hit play. It sputtered, overloaded, and ultimately crashed. Was working fine not ten minutes ago.
Was there any crash dump/log file created.? Sorry, not read the whole thread, if you’ve gone in to this before… And you’re sure its Cubase that’s bringing the system down - and not something else causing Cubase to ‘crash’.?
The ‘spiking’ I understand is a completely separate matter - I hope the devs can improve things swiftly, for you and many others.
There’s lots of variables it could be , you’re not mentioning Buffer size , you mention you are using RME but you don’t mention bandwidth ? Are you using Analogue only in the driver setup or AN/Dig/Adat ?
I use a lot of extenal gear through my RMe with the Adat disabled , WMD on 1 , buffer size 128 /256 with great results , round trip of 4ms and absolutely NO crashes , if you have the RME buffer up to high then you might have stuttering of Cubase due to trying to cope with the latency .
I’m using RME too. Buffers at 128 so I can track midi and mix. It will work well at 64 but struggles on a large project at that so 128 allows low latency still. One wdm device also. The more wdm devices enabled the more processing used. I am using adat with no noticeable difference in performance. These problems do seem to be computer based.
I’ve normally been using 512 samples; I tend to get the best results with it and don’t start to notice the key strike to sound delay really till 1024. Sure, I’d like to use lower if possible, though.
I don’t have an option to disable the ADAT devices, or I would. I use my AES output for loopback recording, primarily, and tend to use it as my primary Windows sound device.
I disabled MMCSS for ASIO just now since according to an RME thread, Steinberg should be handling this in their own way now. We’ll see.
Thanks for the tips, but original point stands: tired of the constant tweaking. We’re well into the age of machine learning now in 2024, so I’d hope Steinberg partners with Microsoft and others to ensure automated solutions to these such issues or at the very least effective automated troubleshooting going forward.
