C14 only using 4 cores?

I am the author of the report !

The person who posted the graph, posted a link to the report to place it in context, and as a reference !

You can lead a horse to water….. !

1 Like

I read that page and didn’t see the baseline numbers. Where are they? Pleae don’t tell me they’re on your Facebook page.

How much time do you have , LOL ?!

After Steinberg were blindsided by the MMCSS mess around 2017/2018 ( Cubase 9.5 era) , they made some changes to the thread management at Cubase 10 that has had a knock on effect, and is getting more and more convoluted the more they attempt to maintain bespoke thread management routines, instead of handing the scheduling to the OS and appropriate thread managers like Intels Thread Director.

V

1 Like

What baseline numbers ?

I have already explained the baseline is a heavy mix session, which has no consistent metric because each DAW has a different method of reporting resource overhead.

I can detail number of tracks , different plugins instances, groups, FX Sends, etc , which will prove what exactly ?

Its the same session and processing load across all the respective DAW’s.

The only variable is how the DAW’s utilize the remaining overhead available over and above the baseline.

Thats the result listed in the chart.

This thread was initially started by the O.P because of poor thread management, and was prematurely overloading ASIOGuard leaving huge resource reserves in TM, which is exactly what I am displaying in the mix environment test I am reporting on.

1 Like

It’s simply this. The graph posted above shows the number of additional plugins that could be run under Cubase, Studio One and Reaper. The questions is, additional to what?

The number of plugins for the DAWbench DSP tests seems to be documented here (in the readme.txt file from the DAWbench DSP zip download on dawbench.com):

DAWbench DSP 2023

Reference Test DAW : Reaper

Session Details and Instructions :

DAWbench DSP

5 x Stereo Tracks of Music Content, 80 x Stereo Tracks of Sine Waves - Parallel Load, 640 Insert Plugins - 8 per Sine Wave track.

DAWbench DSP BUS

5 x Stereo Tracks of Music Content, 80 x Stereo Tracks of Sine Waves - Parallel Load, 640 Insert Plugins - 8 per Sine Wave track, 10 x BUS Channels - Serial Load, 8 Sine Tracks assigned per BUS Channel, 320 Insert Plugins - 32 per BUS

Native Test sessions available for Reaper Only.

But I think the MIX tests being described are something new that isn’t one of these two tests. So I have no idea exactly what is in them either. I did read that this new test is meant to represent more of a real world mixing setup to address concerns about the earlier benchmarks not being real-world situations, and saw a screenshot showing the session with some of the plugins visible, but not too much more than that.

EDIT: screenshot here:

I believe the bar graph is the quantity of “FrankCS” plugins on the right that could be enabled (blue) before it hits the max on the ASIO meter. Although I see 14 of the FrankCS plugins in blue which doesn’t exactly correspond with the quantities on the bar graphs, but maybe one of them is part of the baseline? I have no idea, I’m not sure I’m understanding this. Or maybe 14 was the intial finding but was revised down after doing more testing.

From seeing this screenshot though, I suspect the issue with providing a baseline quantity is that the existing plugins in the baseline are a bunch of EQ’s and compressors and things like that, while the overhead is just comprised of copies of the FrankCS. So quantity of FrankCS doesn’t necessarily directly translate to quantity of other plugins of different designs and functions, apples and oranges, since they may not load the system in the same way. I have no idea if TDR Nova loads the system more than a FrankCS or less.

1 Like

I think that if the point is simply to show that there is a difference between the DAWs then cropping the graph (or focusing only on the additionally added plugins above the common baseline) is fine. We can see the difference. And I think that that’s the point of the article, unless I’m missing something.

Given what we do with DAWs the difference seems significant to me. 70 to 80+ plugins is nothing to sneeze at.

Again though, maybe I’m missing something.

1 Like

As detailed in the report and already noted again here multiple times, over an above the baseline resources being used by the mix session !

Picture worth a thousand words

*Yep, thats the test, but running on a 14900K in that screenshot and @512 from memory, and it was my inhouse testing. Results in the report are from Pete @ Scan. Could well be on the day and on that particular configuration /BIOS/W11 build, etc result scaled down a bit on the 285K

I wouldn’t read too much into the slight degradation in the result. The results for this test in Cubase do not scale as expected per core/architecture, etc, as ASIOGuard collapses before it can scale and utilize the available resources.

* Quick Edit : Just checked the current DAWbench MIX EXT session , each track/group has an additional TUBA insert, the busses have additional ChowTape Plugins, so the preload on that session in the screenshot, which was an early BETA, was a lot lower, so the 14900K would not achieve the result of 14 it did in that initial test. Also Cubase 13 v Cubase 14, all variables.

^
The initial premise for the test was just to mimic a RW mix session, which we did to the identical layout and resource overhead to Marcus’s session, we then beefed it up ( EXT ) , and then I added the facility to scale it even further with the incremental loading of the additional FrankCS plugins in the mix busses, as shown.

Each stage gave us more and more information regards the thread management behavior across the respective DAW’s.

2 Likes

The initial point of the article was actually to show the lack of scaling from 32-64 cores across all the DAW’s, but because we decided to include some additional DAW’s instead of just focusing on Reaper, the comparative performance aspect gained more attention.

Yep, and if you followed along at the other forum threads, Marcus managed to utilise the additional resources successfully in Cubase, by using Audio Gridder, which is another rabbit warren when a freeware app can thread manage and reliably load balance the remaining resources left on the table.

3 Likes

Didn’t I just cover this and clarify in further detail in the other thread, why are you attempting to misrepresent this ?!

Thank you for pitching in. I was just thinking exactly the same. This is very odd behaviour, indeed. And so not in the interest of those of us who use Cubase on Windows.

Unbelievable…

Best,

Magnus

1 Like

I do not believe what I had originally said was directly addressed by the clarification in the other thread (which was addressed more generally). I still cannot see that when I read that reply. It has, however, been clarified in this thread. I’m not trying to misrepresenting things as I said. There were no ill intentions on my part. His clarification in the other thread did make it clear that he felt the issue was serious and dramatic, but at least for me there were many things that remained unclear after that, and remained so since what I specifically said was not directly addressed from what I could see. I didn’t ask for further clarification because I don’t like to bother people who are busy people, and asking a ton of follow up questions to try to get details could have seemed like an interrogation, so I just asked one that I felt was most critical and didn’t specifically ask about the details of the session. It has been clarified in this thread.

I am a Cubase user on Windows too, for heaven’s sake. Any general issues with this impact me as well…

Thanks for digging in and getting some clarity.
It’s now clear that the delta can only be measured in terms of additional plugins run and not as a relative percentage gain.
And yes, we’re all Cubase users here - no conspiracy theories are required.

It’s not worth it, unless you’re having issues with your setup. I had recurrent performance and stability issues with Cubase on Windows; switching to Mac solved most of my problems. UAD and Apple Silicon took care of the rest.

These days, the Thunderbolt ports’ throughput can handle just about any I/O needs you may have for recording, producing, mixing, and mastering music with Cubase.

I hear you, but that’s a different story. I see Cubase as music production tool–not a live-show controller. I’d only use dedicated appliances for shows–not anything running on general-purpose computers with software from Apple or Microsoft.

I have “some” old projects that never got finished, and would definitely not work on my current setup. If I decide to work on any of them, I’ll have to start over with the raw audio and MIDI tracks, and lots of missing plugin warnings. For most of them I have mixdowns, so I can just start over completely if I think it’s worth it.

There are plenty of storage systems that work fine with MacOS. I usually just track to an external OWC with NVMe since it’s fast and portable.

It’s generally not a good idea to run lots of “other applications” on a DAW–it’s not a media server. It should function as an appliance.

One doesn’t really need screens to record music, but if the situation calls for a large number of video outputs–a dedicated separate machine, or a production switcher would make more sense.

That’s true, but “C14 only using 4 cores?” says a lot. On Mac, I turn on my machine and just make music. I don’t have to spend time thinking about driver & OS updates, or researching the latest tweaks.

1 Like

On my new Windows system, I haven’t really had to worry about driver and OS updates either, or doing tweaks, I’m just using a plain vanilla install, not stripping out anything from Microsoft. I am getting much better performance in Cubase and Dorico than I was on my old 3900X, which I am relieved about because the composition projects that were overloading my old system and making them impossible to work on are being handled easily on this system. No pops and clicks - everything seems solid.

However, I might be avoiding many of the issues that others are experiencing due to the fact that I’m hosting my entire template (all instruments and plugins) in Vienna Ensemble Pro on the same system and only have a couple plugins loaded directly in Cubase. I do all my mixing in the Vienna Ensemble Pro mixer (which has feature parity with a DAW mixer) and send Cubase a single stereo return back which then goes to the output. This was not done for performance reasons, but instead for portability with Dorico so that I could move my template back and forth between the two programs by having it in a separate program. I’ve been using this setup with Vienna Ensemble Pro for years. It is possible that this setup has shielded me from noticing performance issues.

1 Like

A DAW and Sequencer has many applications other than ‘recording music’ and mixing it down to a stereo file. They are often integral to ‘putting on the show’. Running something like Cubase is only one of many needs at hand. Been on an amusement park ride lately? Hundreds of tracks all timed out on different speakers and screens throughout the ride? It all has to hit specific cues, right on time. Nuendo is the boss when it comes to stuff like that!

I’ve only listed a few reasons why the newest Macs are not appropriate for every situation. The point was about support for over 50 years of gear and software in the wild. Again, if you’re sitting at a $20,000 mixing console that has no Mac drivers or support, it’s not really the best option.

You’ve described scenarios that are 90% or more ‘in the box’, and are fine for a ‘single composer or producer’ in his personal studio, of which I stated there are competitively priced Mac setups that are great for this purpose.

Yes, there are ‘storage solutions’ for brand new Macs, but if you already have racks full of enterprise class, industry standard stuff that works with 98% of the systems in the world, then why would you want to replace it all just to run one program that is supposedly multi-platform?

How long does it take you to backup your system drive? How long does it take you to do a full system recovery? Can you do that with a 15 minute break between sessions? With the PC it’s simple. Pop drives in the docking bay in pairs. Everything needed to do the basic RAID thing ships with most intel/amd motherboards these days. With most of the new Mac models, you can’t swap out the system drive AT ALL. You cannot add memory. You can’t recycle parts when you upgrade in the future.

How do you deal with multiple producers doing different sessions that tweak and route everything in the studio (sometimes millions of parameters) differently from session to session? With a PC it’s simple. You walk over to the PC, unlock the door on the front, swap out the drives, and reboot. Each producer can start with a sysprepped house drive (has all the drivers needed ready to go), and bring his own complete os/software setups exactly the way he wants it. He can leave the studio with everything he did that day with 2 or more carbon copies in his briefcase (OS and everything intact), and he didn’t have to sit around for hours waiting to clone or backup anything manually.

People have numerous valid reasons for demanding Cubendo performs better on Intel/AMD platforms. If Steinberg doesn’t get around to meeting that demand (They were leaders on these issues in the past, and I do have confidence that they are working on it and becoming leaders yet again), somebody else will.

”Get A Mac” is not what the thread is about. It’s about getting more than %40 of the potential from a modern Intel/AMD build from Cubase.

If it comes down to a ‘choice’ of getting a $2,000 PC (recycling much of the previous build) and keep using my existing kit as opposed to a $5,000 Mac that obsoletes 95% of the studio, and forces swapping out thousands of dollars of basic I/O capability right out of the gate; then, the PC wins every time.

It’s nice to have both and tether it all together if need be. Thing is, it shouldn’t be necessary.

Yes, we understand that it’s complicated, and that anything Steinberg does to benefit the latest high end tech could ruin it for ‘the majority’ of people still running older, or lower spec systems. It has to be done with care and caution.

For some, getting a Mac makes sense and it fits the budget, but for others, there is the PC. The others just want Cubendo to improve when it comes to taking advantage of the most affordable, diverse, competitive, and abundant computing platform on the planet.

3 Likes

Fact remains that, on the Mac - everything works in C14, and there is nothing that doesn’t.

Perhaps you’d like to name this £20000 mixer that has no Mac OS implementation?

Perhaps you’d also wish to share the details of this imaginary facility that allows customers to install their own operating systems into the studio’s main computers.

1 Like

Way too many digital mixers, routers, switchers, and recorders to list. Even more that aren’t all digital, but offer various features for remote control, synchronization, and routing. It’d be much easier to list the ones that DO come with drivers for modern Macs than those that do not. For those that do support a more standardized protocol (I.E. AES/EBU) to drive/control them, you still need an intermediary interface that works on a modern Mac. So what if you already have several great ones, but they’re all PCIe? Getting that going on a modern Mac (assuming you can get some driver for it) will require the absolute highest speced and most expensive Mac model, PLUS some weird conversions, and it’ll get kind of messy and expensive. In the PC world, any off the shelf motherboard can take PCIe cards, and there will be native Windows drivers for it…easy peasy.

I’ve seen numerous television and radio stations that have studios shared by multiple producers, each specializing in different aspects of production. The union ones can have the engineers slap in the drives that they need at the moment (OS and all), and the non-union ones, the producers can pretty much do it themselves depending on the policies and rules maintained by the head engineer. No, ‘just anybody’ can’t walk in and swap drives. There are locks and cameras all over the place. You go through processes and file tech writers to have it done, but it can and does happen all the time. In some settings, several times per day! One boss might want to use an 8 year old Pro Tools setup, another might be all in for the latest version of Cubase or Nuendo, and yet another might not even want a PC, and have a combination of all outboard gear that he prefers using.

In those studios they can not just set everything up and leave it the same for days. It has to be struck, stored, and reset for the next session on the schedule. There can be several projects going on at once, all at different times of the day, all with different producers/directors on the scene with different overall system demands.

I’ve personally helped write some of the implementation plans for preparing such ‘swappable’ system drives, and of course swapping the data drives in raid arrays is bog standard industry stuff. The sysprep for different system drives involved making sure all the drivers are in place, plans for dealing with security and software keys, decryption (if involved), and a rubric of some sort at hand to supply a template for the default settings of things in the racks that the producer wants, and a process to put it all back in a default ‘ready’ state when shutting down. Most of them involved different versions of Windows, but there were also some layouts for other OSes as well (different flavors of DOS/unix/linux/OS2/etc).

For what it’s worth, for mission critical work, test bed drives are a big deal as well. Crews would install all the ‘latest and greatest’ versions of everything (in small steps with hours of testing between each step), and sort out the stable stuff from the trash, then issue setups that are marked stable. If the stuff had to be online/on-air, they always had a redundant backup in the same rack, but with the last known stable setup. Nope, they didn’t just apply every update on the internet as it came down the pipeline! They did it in tiny steps, and it sat in the test bed, running in sync with the last stable version…so if one goes down, it’s an A/B switch to move to the other one.

I’ve been in many studios, theatres, labs, and more, of all shapes and sizes, where everything was in lockable and dockable raid arrays so backups are kept in real time, and media swaps are fast and easy. All the parts in such racks are easily replicable or even upgradable without needing much in the line of tools. The really good stuff you can even rebuild without even powering it down and taking it fully offline!

You mentioned converting everything to raw stems and such as a backup process? That’s time and money, and lots of it. Why go through all that if you can build a system that doesn’t require it?

The point of doing this ‘drive swapping’ on a very ‘standardized’ platform was simple. You have hundreds of pieces of gear in racks, it’s all going into industry standard routers. It was simpler to have one or two master PCs controlling it all, than to constantly have to be repatching stuff by hand. Producers didn’t have to have somebody spend hours every day at a patch panel, or tweaking a smart patch-bay to make things like they need it for the session. Just swap out the hard drives and run a test script real quick (if anything goes wrong make a note so it can be fixed manually, or hand patched if need be)…..

Yes, sometimes BOTH a PC and Mac were on the grid, and even sitting side by side on the same work-desk, but drivers don’t always exist for everything on both systems. There are also limits as to how many machines work into the patch-bay system without adding a lot more patch-bay equipment, which can sometimes cost a lot more than a really nice computer workstation.

A lot of those places did have Mac setups too (witnessed it from Mac OS 6 days to present). People enjoyed using them, but the arrays of external kit available to those labs were a mere fraction of what the Windows PC could use. That was less-so the case when Mac switched to Intel. Now that they’re back with something ‘new and different’, maybe 3% of the existing code and hardware in the wild works with it at present. Give it a few years, and Apple will break it even more by pulling Rosetta out of the bundle. By then it might not matter as much as more and more happens ‘in-the-box’ and more outboard gear is simply sent to the junk yard. We’re not there yet though…not even close.

Again, for everything ‘in the box’, the Mac platform is proving itself! They do have some models that are competitively priced if ‘in the box performance’ is your primary need. It does a lot of things really well! Still, there’s over 50 years worth of stuff it simply cannot do, period. It’s not appropriate for everybody, in every situation.

Recently we got some validation on this topic from a forum Steinberger here in a different forum thread.

I’m always interested to see peoples findings. Thanks for sharing your report. Lets hope things get updated and improved in the near future.

Our code was designed when schedulers were way less smart (and dedicated to audio processing) than today. But we can’t just switch the way we do our audio processing like we would, when we were building Cubase today from scratch. Changing such fundamental code takes time and caution to not break the rest which is working fine.

3 Likes

You’ll never win an argument with a “Mac” fanboy. They’re always right :joy::joy:

4 Likes

Since you’re trying to defend the indefensible, then to be honest - you’re just arguing with yourself.