Cubase Pro 8 Performance Comparison Mac OS X 10.10.5 and Windows 7 Video

Audio Plugin Hosts and other audio software applications discussion
RELATED
PRODUCTS

Post

Well, I was curious so I did a direct comparison using the same project running on the same computer. The only thing different is one time I ran the project in Windows 7 and the other time in OS X Yosemite 10.10.5. I did a screen recording of the results on both platforms and slapped together a video. The results are surprising to me as to how close the performance is.

I'm now wondering how close the performance gap is between Logic and Cubase on OS X. Here is the link to the video below:

https://youtu.be/MjU3ISBeTFo

Post

Machine specs would be nice please.

Post

Intel Core i7 2600k at 4.5Ghz
16GB RAM
Z68 based motherboard
RME AiO PCIe Audio Interface
Both OSs are loaded onto separate Toshiba Pro Q series solid state drives
Samples and project files are on separate 7200 HDDs

I was able to use NTFS under OS X using Paragon so the projects were actually being read from the same drive.

Post

Thank you..... I know the parity is something that Steinberg has worked on very hard to achieve, including introducing energy modes for the computer and asio guard (especially 2).
thanks for doing this.
rsp
sound sculptist

Post

I went back and disabled ASIO Guard and the project behaved pretty much the same on both operating systems. 32 samples with this stuff going is a no go. 64 sample buffer was tough and operating around an average of 75% ASIO performance but would peak red every so often on both operating systems. 128 sample buffer was the same result on both operating systems at about 60% without any overloads.

2 things I have realized here is that ASIO Guard works and the performance gap between the different operating systems isn't large now. The old tests from DAW bench are from Cubase 5 I believe and OS X 10.6 and Windows XP. Those tests made OS X look like a slug.

I will say that regardless of OS Cubase seems to work really well now.

Post

Very interesting.
thanks
rsp
sound sculptist

Post

I posted this as a comment on the video page, but figured I'd post it here too. :)

I did the same test on a Boot-Camped Mac Pro 4,1 a couple of years ago (so I would have been on Mavericks and Cubase 7.5 or maybe 7) and had pretty much the same result. As you did, I kept my video objective - just pointed out facts and let people draw their own conclusions. For some reason it really seemed to offend Windows folks. Some of them posted harsh comments and gave me Thumbs Down. (And the irony to that is I was doing the test for myself to see if I'd be better of - or at least get better performance - by running Cubase in a Windows environment.)

Indeed ASIO Guard changed substantially in version 8. The CPU usage seems higher overall but the performance is completely stable. Your video also illustrates that, with the new ASIO Guard, the latency setting has very little affect on the ASIO Performance Meter.
Brock
2013 Mac Pro 12 core, 32GB RAM, 1TB Flash | UA Thunderbolt Apollo QUAD | Cubase/Nuendo (latest) | Axe-Fx II

Post

is this through bootcamp? I don't believe it to be an ideal environment.

If it is on a proper optimised dedicated windows machine, very very interesting.

I plan to do the test myself when i get C8 so will see if there is anything to add.

will be using an old 2.3ghz ivy bridge windows machine that runs very very well,

vs a 2.8ghz current haswell mac.

of course the latency chosen has little bearing on the asio meter because it's using the 23 ms asio guard buffer which didn't change through any of the latency settings. Just like logic whether 32 or 1024 samples, the project plays back the same.

when it WILL make a difference is when an instr track or audio track with many fx is selected and thus goes "live". That's when the 32 samples will struggle vs say 128.

Post

It is a Hackintosh set up. Effectively performance is the same from Windows 7, Windows 10, and OS X 10.10.5 for me. Even with ASIO Guard off. I'd say on a modern computer the performance differences would be subtle since the hardware has become so powerful.

Although I won't use Windows 10. There appears to be some kind of bug that causes spikes when clicking the mouse on the canvas and also when selected multiple midi notes. It has something to do with USB input or the display drivers. This bug doesn't seem to effect me in Windows 7 or OS X so I'm thinking it is driver related with Windows 10.

Hey though, Windows will always have at least one leg up over Mac and Linux, and that is Games!!! Although more and more are becoming available for Mac and Linux.

Post

Very interesting. if it is a hackintosh, that means indeed the windows part is running the best it can (i presume you have high performance mode activated in windows power settings, right? And asio on "backround services"?"

The reason i ask is as I have been doing performance testing like this, to a very intense degree, for much of my daw life. And i am a mac user and will always be one primarily, so please don't think I am posting as a windows guy, as i only bought the windows laptop for a dirt cheap sum for travel so i wouldn't take my expensive macbook. it just so happens it performs really well for audio, it was juts a fluke that I got a good off the shelf one with low dpc latency.

Anyway, i have always found the opposite - the mac is always obliterated by windows in audio plugin counts.

However, i am thinking the fact that it's a hackintosh is at play here..

have you disabled speed step in bios? If so, there's your answer there - the mac WILL perform much better than a native machine because the native ones are always juggling the mhz like crazy.

This is their downfall.. the mac efi vs the windows bios. If i were to go and buy the parts identical to my mac to make a hackintosh, i can guarantee i would double the performance doing so - but i don't bother as i want turnkey and i am happy with the performance as it is anyway.

Your results will only be accurate for the majority of mac users when you test on an actual mac with similar specs to the windows box.

In fact, my imac is a 2011 model, and is a 2600K inside, with ssd, and 16gb ram. I do have all the plugins in your vid but i doubt you;d want to send me your project. so why don't we make a standard test between us, you test windows, let's say, pick a plugin, spread across 16 audio tracks, and do instance count on mac vs pc.

You do the pc, i do the mac.

Post

I haven't really messed with the Windows settings too much. I know there are a bunch of "tweaks" you can do. The problem though, sometimes these "tweaks" mess other things up. My DPC latency is in the sub 50 micro seconds range which is decent. I think it goes lower if I turn of HPET in the bios. I don't have a problem with core parking or anything like that either. I think people search out these "tweaks" when they are having issues. My machine works so I don't really seek out them as the performance is awesome already. I just checked and it is set to favor programs and not background services. But again, I don't have issues. If I was getting pops and cracks etc... I would go hunt down the issue and change stuff.

Yea, I disable Speedstep and all the power settings and let it run at it's over clocked speed. I did notice that though when Speedstep is on within Mac OS X that the processor speed is just constantly fluctuating. Within Windows it clocks down when it is idle and doesn't need the power, and then ramps up when it needs it. It isn't constantly fluctuating all over the place. I even created special settings within OS X to try to control this using a SSDT but it is sporadic still. Although when I launched Cubase in OS X and started working it would stay around 4.5ghz even with the energy settings on. But, I'd rather just disable them.

Hackintosh is a odd route at first but once you understand what is actually going on, it isn't too hard to figure out. You can even copy somebody else's build and just follow their install method. It's pretty easy. All I had to do was purchase Snow Leopard back in the day as Apple still hasn't charged for an update since. Well and you need access to a Mac to make a bootable installer too.

I could do that test. Honestly, A.O.M Invisible Limiter running with 16x oversampling is a heavy duty plug in and may only need a minimal amount of instances. The only thing is that my i7 even though the same model as yours runs faster. I think the normal turbo boost is 3.8ghz on it, so that is as fast as yours will go. So I have an extra 700Mhz working in my favor. Which I guess that is the benefit of building your own machine.

Anyway, what buffer setting do you want to use? ASIO guard on or off?

Post

I am curious to know what is the intent of testing on entirely different machines?
What in the world would that prove?
I think the original test on the same machine is the only type of test that gives any useful info. Done both by monkeymanx on a hackintosh and Brock Jon on a mac....

It use to be that PC's on Cubendo give better performance. I don't' think that is any longer true, as both monkeymanx and Brock Jon have testified/shown.

Btw Cubendo on PC overrides (unless you tell it not to do) the energy savings mode on your PC and forces it to prioritize cpu power.

rsp
sound sculptist

Post

zvenx wrote:I am curious to know what is the intent of testing on entirely different machines?
What in the world would that prove?
I think the original test on the same machine is the only type of test that gives any useful info. Done both by monkeymanx on a hackintosh and Brock Jon on a mac....

It use to be that PC's on Cubendo give better performance. I don't' think that is any longer true, as both monkeymanx and Brock Jon have testified/shown.

Btw Cubendo on PC overrides (unless you tell it not to do) the energy savings mode on your PC and forces it to prioritize cpu power.

rsp
because we are talking real world that way - He has a 2600K. I have a 2600K. In a real mac. With the original mac EFI. The h'tosh audio community is estimated at 1% of mac owners. I am talking REAL WORLD - this is real world for only a tiny tiny, did i say tiny?, percentage of OSX users. This test means nothing sorry to say - we are talking about a hacked machine running an apple eula violating installed OSX. The test is only valid when the results are done on a real mac vs similar specced PC. There is just NO WAY that a 2600K in a real mac will be on par with a 2600K on a dedicated windows box.

I did the 7.5 tests myself, extensively, and windows thrashed the mac, but haven't done the 8 - it will be closer, but not 1:1.

I DO know that cubase 8 on a mac does not perform better than logic on the same mac, but on par-ish. Have done those tests over the demo period.

And I can tell you this - a similar specced PC with Studio one at 128 samples, or cubase 6 (which has no asio guard of any kind) - blows a similarly specced real mac using logic and cubase 8 away. With asio guard on! I have done extensive testing on this myself..many many days worth of tests. We are talking in the region of 30% more plugins instances.

This is a cubase 8, windows 7 vs osx performance comparison on the same hacked machine - not a cubase 8 win vs mac comparison as it's simply not valid and I am surprised you don't see that. This is why I had to ask what was used as the results initially stunned me. Sure it proves there is no inherent problem with osx itself but it doesn't prove there is no inherent problem with macs and audio performance in general vs pc. I believe if it was an asio guard OFF scenario however, the osx, even on a hackintosh, would have been thrashed.

I might install bootcamp myself, on a machine with native mac efi, and do the tests - and i guarantee you windows will win in that case. The problem is the mac efi for audio.

Post

monkeymanx wrote:I haven't really messed with the Windows settings too much. I know there are a bunch of "tweaks" you can do. The problem though, sometimes these "tweaks" mess other things up. My DPC latency is in the sub 50 micro seconds range which is decent. I think it goes lower if I turn of HPET in the bios. I don't have a problem with core parking or anything like that either. I think people search out these "tweaks" when they are having issues. My machine works so I don't really seek out them as the performance is awesome already. I just checked and it is set to favor programs and not background services. But again, I don't have issues. If I was getting pops and cracks etc... I would go hunt down the issue and change stuff.

Yea, I disable Speedstep and all the power settings and let it run at it's over clocked speed. I did notice that though when Speedstep is on within Mac OS X that the processor speed is just constantly fluctuating. Within Windows it clocks down when it is idle and doesn't need the power, and then ramps up when it needs it. It isn't constantly fluctuating all over the place. I even created special settings within OS X to try to control this using a SSDT but it is sporadic still. Although when I launched Cubase in OS X and started working it would stay around 4.5ghz even with the energy settings on. But, I'd rather just disable them.

Hackintosh is a odd route at first but once you understand what is actually going on, it isn't too hard to figure out. You can even copy somebody else's build and just follow their install method. It's pretty easy. All I had to do was purchase Snow Leopard back in the day as Apple still hasn't charged for an update since. Well and you need access to a Mac to make a bootable installer too.

I could do that test. Honestly, A.O.M Invisible Limiter running with 16x oversampling is a heavy duty plug in and may only need a minimal amount of instances. The only thing is that my i7 even though the same model as yours runs faster. I think the normal turbo boost is 3.8ghz on it, so that is as fast as yours will go. So I have an extra 700Mhz working in my favor. Which I guess that is the benefit of building your own machine.

Anyway, what buffer setting do you want to use? ASIO guard on or off?

the logical thing IMO would be asio guard on *and* off, same test, just simply enabling and disabling AG in settings, getting counts both ways.

However unless you are willing to let your 2600K run at it's normal clock for the sake of this test, there is no point. ;)

I can tell you this - in raw plug in count, my $800 2.3ghz ivy bridge 2012 windows laptop at 128 buffer, outperforms a $2800 current retina macbook pro 2.8ghz haswell!

Raw count before total overload at HIGH latency is higher for the retina, but it cracks and pops earlier than the dated windows box! So this little cheap windows lappy has more *usable* performance than a current retina macbook top of the line!

Post

One of the guys over at Gearslutz made a good point that the ASIO meters are not really that reliable. I looked at the current DAW bench tests but you need to run some plug ins in demo modes and some of them require an iLok. I was thinking of running them but that is just a hassle. Would be different if it was all with free plug ins. The guy over at Gearslutz did tell me the performance difference is due to core audio. It most likely has more overhead than ASIO. And even though Cubase on Mac says ASIO we all know it is actually core audio. Think about it, if you try to use the normal WDM mode in windows the damn thing can't even play stuff back at low latency. So I think it is more about audio protocols than anything.

Just for information though I did go back and turn ASIO Guard off and the results were still pretty similar. Again though, maybe a project with a bunch of heavy duty VI's would be different.

About the EULA... I put the Apple sticker that came with Snow Leopard on my case, hence making it "Apple branded" to comply with the EULA... I'm joking but I'm not... I really did put the sticker on there. Come to think of it, it is odd that Apple would put an unlocked processor in their machine. Unless you can over clock a normal Mac.

Anyway, What to do. Mono audio files processed through AOM Invisible Limiter with 16x oversampling on? See when it overloads? That sound good?

Post Reply

Return to “Hosts & Applications (Sequencers, DAWs, Audio Editors, etc.)”