Multi core settings and ASIO gaurd

If you’re going to compare the performance across DAWs you need to really load them up the same, have you done that? Maybe test with 50-100 tracks all with like-for-like third party plugins on each track, and then see how low the buffers can go with each.

Studio One is very efficient, but i’ve found bigger projects Cubase chomps through better. Your mileage may differ however, as i know the newer versions of Studio One have leaped in regards to ASIO Engine/CPU usage.

But just make sure it’s a fairer real world comparison that you’re doing, i’ve seen claims that Studio One can 30-40% more efficient with ASIO load… I just don’t know under what conditions those claims are made. I’ve also seen people claim that Cubase is getting hungrier since ver 7 - so maybe it’s a case of one DAW falling down and the other rising up?

The big question is which one works best for you within those lower latency restrictions.

I am not comparing, I am simply trying to get shot of the clicks and pops, As I said earlier I have AFF exported the exact same project to Studio One and had none of the issues, UAD, and Antelope drivers are working absolutely perfectly well with both SO and PT native but not with Cubase. We are not even talking a massive project 25 tracks at 96kHz maybe 30 plugins running which on a system like this should not even blink the meters and doesn’t on both of the DAWS mentioned. I want Cubase to work properly again that’s all. it’s the best out of all of them, but I have to say the Studio one guys have got that audio engine working like a dream, external hardware inserts, etc all work a treat. The codings all to deep for me but Cubase is the most expensive out of all 3 platforms. The hotfixes do imply there is drouble in the house

Yes Studio One has come on leaps and bounds, and the audio world knows it. It’s tempting for a lot of Cubase users, due to the history of the dev team that founded Studio One too. (Won’t even mention the dongle lol)

However, The point i’m making is that for smaller projects Studio One may be far more efficient, but with larger project sizes Cubase (In my experience) starts to show it’s strengths and retains the latency/buffer settings well that you started with, whereas with Studio One you have to start pushing it higher.

Whether there is something built into the Cubase audio engine that has a fixed/set overhead from the start, i don’t know, but it feels like there is to me. But, this doesn’t help you in anyway of course

At what point do the pops and crackles stop? @ 256 buffer / 512 / more?
And have you tried with no tracks set to monitor/record?

I run an old Firewire interface and 128 buffer fine for me, but i’m not trying to run on 96khz.

This thread has some interesting info:-

I think that the following may be worth looking into:-

  1. “I have seen that, too. It improved a lot by switching the audio file format back to 32-bit float from the new default 64-bit float, which seems like some severe overkill.”

  2. “I had spikes and dropouts all over the place (Windows 10 and Cubase Pro 10.0.50), but setting the Power Option to High Performance has so far - fingers crossed - worked flawlessly for me.”

Hi thanks for your input. yep I have tried with an empty project, same issue. If I am lucky it’s gone at 256 but a heavy track count leads to 512. I have also flipped the Apollo back to FW800 to see if the issue continued and I have also run the Antelope on USB. Same issue.I have kept the audio engine on 32bit(64 bit? unnecessary lol)
Yes I am sure SO starts to choke as it goes higher, and yes agreed the coding team I believe are old Steinberg and Emagic coders. They are certainly pulling something together and yes! Dongle!I use it normally just on the laptop for remote work
I am a Cubase user its what I want to work for me, trying to submit a ticket is a pointless task as I just bounce back to the main support page so this forum seems to be my last hope Obi-Wan

I was using studio one this weekend and noticed that the actual monitoring round-trip latency figure is in the processing tab when you setup your audio interface. So as someone mentioned earlier, just check that you’re comparing like for like.

Hi - Adam, Make sure you check this hot fix for ASIO at Steinberg Forums
Cubase 10.5.12 fixes a performance issue caused by ASIO overloads

To be honest it’s so obvious to the ear. SO settings buffer set to 32 monitoring through and working like a dream
Cubase I have to crank the buffers to 256 or more if I want to monitor through and i have a noticeable latency, its small at around 9ms but its there, a performer can feel it and really! if I can get spotless out of SO and not with CB on the same rig there is an issue. If I drop to 128 in CB just monitoring I am hearing a cycle thumb click thing all the time around 3 second interval and then random glitches when I start to record with pops and clicks. up the buffer to 256 and it’s hardly there. I think people are not understanding the issue here. I am not comparing the platforms. I am saying there is still an issue with the audio engine that is not fixed with either hotfix. Various people have suggested its the UAD drivers or the Antelope Drivers. Which have either thunderbolt USB or FW800 options? all of which I have tried and all which work perfectly with Studio One and Pro Tools but not with Cubase

You say you can hear ‘by ear’ but that’s the pops and crackles, i presume? - the question i’m asking is if S1 actually running at the same latency as Cubase which is crucial here. i.e. is there additional overhead in the S1 buffer settings that don’t exist in Cubase? There was an additional fixed software buffer (64 samples?) in Studio One, for example. I don’t know if that still exists - but it’s these hidden values that you sometimes have to be wary of when referencing performance across DAWs.

You need to set them both to be the same round trip setting to compare this, not setting the same buffers. Just because you set a number, doesn’t mean that’s what’s being used in the processing of audio. i.e. it’s entirely possible that despite the buffer settings you’re entering that Cubase is running at a real life lower latency as it’s more transparent in it’s reporting.

The S1 equivalent of ASIO guard, also, may be set far higher of a latency of 50-100ms - and would be taking a lot of strain from the CPU which you may not be allowing Cubase to do (I don’t know how you have it setup). But as you’ve posted above the real test is when recording, as that’s where your buffer and subsequent latency comes into effect.

Also i’m not doubting your issue, as i kinda expect Cubase to sit in S1’s shadow based on their recent improvements in regards to CPU - i just want to make sure you’re referencing apples to apples, because i’d never even noticed the values in the processing tab on S1 before, until this weekend and the actual values where much higher than i originally thought - which is why i posted back.

Also, i know you’re not comparing DAWs, but simply referencing them - what else can you do, you need a baseline to compare Cubase too as youre getting pops and crackles, so please don’t think you’re being misunderstood and this is a comparison debate. I appreciate that the comparison has arrived due to the issues you had, not the other way round.

Hi Adam - I am asking the same question that “Norbury Brook” on feb24th, asked?

  1. "IS your Antelope connected to the Apollo?
  2. so your Apollo is the only thing connected ASIO wise"?
  3. Or when you run the UAD drivers and the Antelope Drivers do you have to use ASIO4ALL?
  4. Do you have to monitor through the Antelope?
  • those are important questions to get an understanding of your setup.

I would not be to happy if I had your CPU and had to run a 512 buffer either! My systems performance is not any better on the Cubase 10.5.12 either, I saw no improvement at all. But my system is not as powerful as yours and I don’t have your latency issue mostly I think because I use the UAD method of “direct monitoring” and I DO NOT monitor though Cubase which provides ZERO latency! Is it possible for you to do that?

Can you download Reaper and use that as your baseline. There’s nothing hidden in Reaper and it’s a good benchmark as it’s all good clean code.

the ASIO4 all thing was to use your on board sound card to take either interface and drivers out of the equation.


M

REAPER will walk over Cubase, it will be even worse, there’s practically no overhead in comparison to Cubase! :slight_smile:

128/96 = 1.3 ms latency
256/96 = 2.7 ms latency
128/44.1 = 2.9 ms latency.

But this is only for the soundcard buffers. There’s also other system latencies.
Remember that Midi also introduce latencies, especially when using a lot of midi-Messages, VSTi’s and fx’s.

Is there still a 14 core limit with Windows 10. My understanding was that the 14 core limit was for 32 threads minus 4 reserved or 28 thread or a total of 28 thread. However, my understanding was that was for a 32 bit OS, and for a 64 bit OS, that the limit 256 threads.

So if you are running Windows 10, 64 bit is there still a core limit of 14? I this a limit of the Multimedia Class Scheduler Service (MMCSS)?

If this is the case, can you get around this by using a Microsoft Server OS like Server 2012?

I see lots more threads in 16 cores - not even above and idle
AudioPerfMeter.jpg