What's new

Cubase 10 Performance MacOS Mojave vs Windows 10

Is this different from Channel Batch Export?
Yeah. I just like to hit the track freeze "snowflake" icon when I get to a stopping place in my sessions but Cubase makes you do this on track at a time, whereas Pro Tools and others let you freeze as many as you want in one click. Plenty of other great export and rendering options in Cubase though!
 
Yeah. I just like to hit the track freeze "snowflake" icon when I get to a stopping place in my sessions but Cubase makes you do this on track at a time, whereas Pro Tools and others let you freeze as many as you want in one click. Plenty of other great export and rendering options in Cubase though!
Ah, gotcha. Strange to not have that feature available...but I guess all DAWs have that handful of "wishlist" features!
 
Its bugging me so I am going to try to replicate this test without VEP on Cubase. I did it on LogicPro and dropped CPU usage from 35% down to 25% without VEP involved. I'm curious what cubase will do without VEP connection. Give me a few minutes to work that out....

FWIW, I spent some time and recreated the non-VEP test in Cubase10. 100 Tracks. All ViPro instruments with MirPro and Miracle.

Result? Not good. It won't play all the way through without dropping out audio entirely and crapping out, CPU ~3x of what LogicPro was using for the same tracks.

So what I can say is for me, LogicPro is out-performing Cubase10 by perhaps factor of 3x with 100 tracks of VSL instruments.

while I was building up the project track by track, in the earlier phases with only a few tracks, it was performing well and I was hopeful. As I got past 10 or 20 tracks I started to notice the CPU has straining increasingly but still entirely usable. By 50 tracks it was already using considerably more CPU then LogicPro for the same tracks, but still usable. By 75 tracks I had to close MirPro window and make sure Retina display mode not enabled to get playback. By all 100 tracks, it won't play all the way through no matter what I do.

I think with smaller projects cubase works fine and certainly seems smooth and has many useful features. As I was setting things up I kept thinking, hmm, that is cool, I wish LogicPro did that. But anyway, performance is a factor on mac. It ought to be able to play 100 tracks without sweating. LogicPro certainly can.

I am not a Cubase expert and its possible I am missing some opportunities to optimize the project in some way. Any suggestions welcome.
 
I had a chance to put together a Cubase session of 272 audio and sampler tracks and some plugins to test on the Mac and Windows partitions of my Hackintosh. With the same buffer and ASIO guard settings the MacOS version outperformed the Windows version with the CPU at about 88%, whereas on Windows the CPU was maxed out and the session didn't play back properly. Here's a link to the session file: https://www.dropbox.com/s/1pltwwdq2snclc5/CPU Test.zip?dl=0

A second test with 100 or so Steinberg reverbs also outperformed on MacOS compared to Windows. I'm a little surprised but there you have it!

My CPU is an Intel 9900K and I have 64gb of RAM.

EDIT: The reverb test (320 instances of RoomWorks being fed audio) had the CPU at 63% on MacOS and 98% on Windows.
 
Last edited:
Sierra. I'm shocked by his results frankly. What plugins was he using? I was using all VSL instruments and MIRPro, so its always possible VSL is not good on mac.

Anyway, I digress, at least for me here, Cubase is a dog compared to LogicPro. I couldn't even play 100 tracks much less 272 as he is doing on his hackintosh. I've heard numerous reports of mac version running less efficiently, this is the first time I've heard the opposite. (shrug), it is what it is... I guess roll the dice and buy it if you want to find out if it will work on your mac.
 
Last edited:
I am on Mojave. My tests are pretty unscientific, I just added an audio loop and made a few tracks with the Cubase sampler, noodled some MIDI, and added some stock plugins from Steinberg and duplicated until my CPU meter started to max out then compared the results on the other OS. Windows is optimized for performance, the only difference is that it boots from a standard SSD rather than NVME but when I did disk read/write tests the speeds were about the same.

I just did another test with 160 instances of ZebraHZ (VST3) noodling MIDI and duplicating as before. About 55% CPU on MacOS, and 75% on Windows. MacOS wins 3/3 on my build.

It's worth noting that the 9900k CPU completely smokes the 2012-13 Mac Pro CPUs especially for single core performance.

EDIT: 256 buffer
 
I'd love to see some tests from someone with an iMac Pro comparing Logic and Cubase and Cubase on MacOS and Windows!
 
Nonetheless, on my old crappy system I get 3x the performance out of LogicPro compared to Cubase.
Wasn’t trying to knock your system by the way! But I was really surprised about Cubase performance improvements going from a 12 core cheese grater to my home built 6 core i7 8700k rig last year (since upgraded to 9900k). Thanks for taking the time to run those tests and sharing the results.
 
I had a chance to put together a Cubase session of 272 audio and sampler tracks and some plugins to test on the Mac and Windows partitions of my Hackintosh. With the same buffer and ASIO guard settings the MacOS version outperformed the Windows version with the CPU at about 88%, whereas on Windows the CPU was maxed out and the session didn't play back properly. Here's a link to the session file: https://www.dropbox.com/s/1pltwwdq2snclc5/CPU Test.zip?dl=0

A second test with 100 or so Steinberg reverbs also outperformed on MacOS compared to Windows. I'm a little surprised but there you have it!

My CPU is an Intel 9900K and I have 64gb of RAM.

EDIT: The reverb test (320 instances of RoomWorks being fed audio) had the CPU at 63% on MacOS and 98% on Windows.
Do you have a proper Mac though or do you mean Hack when you say Mac?

Because this is in no way an honest test I am afraid. As someone who built four Hackintosh machines, it is important to understand that a Windows PC running macOS is not a Mac.
The Power Management will behave differently, the operating system will be different and most of all certain Processor architecture and extensions will not behave the same.
They may be present on a proper Mac and not present and disabled for Hackintosh compatibility purposes.

Remember a Real Apple machine has certain extensions and hardware that works with the SMC (System Management Controller) of the Apple machine. A Hackintosh 'fakes' i.e. emulates this functionality. Not to mention Tables that load during Post and hardware tests performed etc.
 
Sierra. I'm shocked by his results frankly. What plugins was he using? I was using all VSL instruments and MIRPro, so its always possible VSL is not good on mac.

Anyway, I digress, at least for me here, Cubase is a dog compared to LogicPro. I couldn't even play 100 tracks much less 272 as he is doing on his hackintosh. I've heard numerous reports of mac version running less efficiently, this is the first time I've heard the opposite. (shrug), it is what it is... I guess roll the dice and buy it if you want to find out if it will work on your mac.
Refer to my previous post... this would make significant difference compared to genuine Apple machines
 
what you are saying is all true, however in this case he's actually getting better performance from his hackintosh then from running windows on the same machine. I am more suspicious that there is something wrong with his Windows configuration.

Comparing someone's 2019 hackintosh to a 2012 MacPro is not very useful information for this discussion.

Another factor is that newer CPU's have certain low level CPU instruction set optimizations in them which Cubase or VEP could be using, which a 2012 MacPro cannot take advantage of, and on and on. Perhaps Cubase is coded somehow to take advantage of newer CPU's which are more common on windows by the way, while not being optimized for 2012 architectures, which like it or not are much more common on the Apple community at this time. Just thinking out loud now.

It would be interesting to have a 2019 ImacPro, or the next gen thing coming out this year, tested with both OS X and windows bootcamp, to see how they compare, but ultimately I would still expect Cubase to be highly optimized under windows and less so under OS X just because of Steinberg's history. But its possible they are getting closer and closer to performance parity between the two platforms when the same hardware is used.

What I can say is that here in my studio with my ancient 5,1 Mac Pro, Cubase10 does not perform well at all, LogicPro is performing 3x better. That is true with and without VEP involved in the setup. I can't imagine that a windows setup would perform worse then what I have, but maybe I would have to install bootcamp to find out for sure, which I don't plan to do.
 
what you are saying is all true, however in this case he's actually getting better performance from his hackintosh then from running windows on the same machine. I am more suspicious that there is something wrong with his Windows configuration.

Comparing someone's 2019 hackintosh to a 2012 MacPro is not very useful information for this discussion.

Another factor is that newer CPU's have certain low level CPU instruction set optimizations in them which Cubase or VEP could be using, which a 2012 MacPro cannot take advantage of, and on and on. Perhaps Cubase is coded somehow to take advantage of newer CPU's which are more common on windows by the way, while not being optimized for 2012 architectures, which like it or not are much more common on the Apple community at this time. Just thinking out loud now.

It would be interesting to have a 2019 ImacPro, or the next gen thing coming out this year, tested with both OS X and windows bootcamp, to see how they compare, but ultimately I would still expect Cubase to be highly optimized under windows and less so under OS X just because of Steinberg's history. But its possible they are getting closer and closer to performance parity between the two platforms when the same hardware is used.

What I can say is that here in my studio with my ancient 5,1 Mac Pro, Cubase10 does not perform well at all, LogicPro is performing 3x better. That is true with and without VEP involved in the setup. I can't imagine that a windows setup would perform worse then what I have, but maybe I would have to install bootcamp to find out for sure, which I don't plan to do.
Yes agreed on all points there

And I find the performance benchmark results with my 212 Mac Pro to be the same
 
Yes my "Mac" is a Hack, I understand it's not the same as an Apple machine. But given its home-brew nature I'm surprised it turns on at all let alone outperforms my the Windows boot. I followed the guides to optimize Windows when I built the PC for power management and CPU performance. It's possible I missed something but I used Windows on the machine for a long time with solid performance and no issues as far as Cubase was concerned.

Here's an interesting post where a member is getting surprisingly good real-time performance on an iMac Pro with Cubase: https://vi-control.net/community/th...e-performance-of-imac-pro.68459/#post-4182534

Maybe newer CPU architecture really is a big factor in all this.
 
windows also has DPC latency issues, you could have a problem related to that, which OS X in general does not suffer from.

just one example...

Until you know your windows config is completely optimized I do not think its a valid general assumption to make that Cubase runs better on mac compared to windows.

As I said earlier, I suspect that Cubase is taking advantage of newer CPU instruction sets that are not available on the 2012 MacPros. When you code things you can code things in a way to be perfectly efficient on the newer instruction set and horribly worse without it. LogicPro appears to be coding things in a way that our 2012 macs continue run performantly, while cubase is not taking that into consideration. That is just a guess.

Many of us continue to use cheesegraters for perfectly valid practical reasons. Apple has done strange things in the past few years. I think its foolish for Steinberg and others to leave behind the cheesgrater macs just yet. It is still a significant portion of the user base out there.
 
Top Bottom