I can confirm this. The Cubase ASIO/CPU/Performance Meter/Meters (whatever you want to call it) will rise dramatically at the beginning of a project. I’ve performed a few tests of my own comparing Cubase 7, 7.5, 8 and 8.5 with Reaper, and have always come to the same conclusion: Both will accept a similar number of (identical) Plugins before actually dropping out, regardless of how the performance meter may favor one or the other.
I’ve panicked a few times over the past 4 years after buying new hardware and discovering that Cubase was an under-performer judging solely from the VST Meter, but like others have also found out, it is sometimes best to ignore the meter in Cubase
I’ve used the same project again and this time I’ve used the windows meter as well so that’s shown in the grabs.
I also opened the project in Cubase 7.5 and that has added to the fun.
Also I’ve experimented with ASIO guard on and off and on my powerful machine ASIO guard in Cubase 8.5 seriously reduces performance.
So…Reaper; when opening the windows monitor you can see that actually its distributing very nicely across all cores , but the 12% on the reaper meter is very optimistic and under reads what I think it should be showing.
Cubase 7.5; fairy good core loading and nice smooth performance, ASIO guard is on, the Steinberg meter corresponds pretty well with the windows meter I would say.
Cubase 8.5. With ASIO guard on normal, performance is the worst here, both in the Cubase meter and the windows monitor. I cannot use the scissors tool without the project hitting the red and I hear distortion from the core s hitting 100% as well.
my thoughts are leading toward some sort of GFX coding that’s been implemented to enable cross platform consistency and performance improvement for the OSX users. However looking at it this may have hindered windows performance. this is just a guess here.
I also noticed when opening Cubase 7.5 , how clear everything was on my 30 inch cinema display, Cubase 7.5 just seemed to be a lot smoother in performance when using the tools etc , the whole GFX of 7.5 was clearer and performance a LOT better.
Thanks for doing these tests and posting, Norbury Brook.
… I’m telling you today the ‘silence on automated tracks’ and > ‘flickering when using slice/pencil/line tools’ > issues are fixed in 8.5.10 …
Maybe that will address what you’re describing in terms of the display (from a post by Fabio recently), and hopefully performance issues as well (distortion you reference hearing, and cores peaking).
Re: ASIO guard - ADK ships with ASIO guard “OFF”, and doesn’t recommend turning it on under normal circumstances, FWIW. In your n = 1, where performance was better with it “Off” than on, maybe there is something to that.
Finally, just a meandering thought, not worth much especially since I’ve never used Reaper: It’s my understanding that Reaper is great for what it does, but that it is missing a lot of functionality that Cubase has. Under those circumstances, I suppose it wouldn’t be that unexpected for Cubase to draw a heavier load on the CPU than Reaper?
Ideally perhaps the Cubase CPU load would be proportional to which features are being used at any time, but even I can see that degree of programming might not be expected in a DAW (as opposed to something programmed by NASA or brain surgery robotics …).
Each track has 8 instances of Waves Kramer tape, with 8 further instances on the master channel. No groups or FX. Steinberg MR816CSX interface with buffers set at 92 (It might be that things are different at higher buffer settings and if I do get some time, I will test that as well).
Cubase can cope with 11 tracks. Adding a 12th starts to cause audio dropouts.
Reaper can cope with 16 tracks, so almost 50% more.
Definitely interesting and certainly suggests there is room for better efficiency in Cubase.
ASIO guard gives a clear performance boost on my machine by the way.
Tell you one thing though… it took me a lot longer to add those plugins in Reaper without Steinberg’s nice plugin search function. And what am I going to do when I need to Variaudio a vocal track? And how am I going to scale the mixer to exactly match my Avid S3 controller (Oh wait, that doesn’t matter because Reaper doesn’t even support EuCon controllers!) etc etc. I think the previous poster made a very valid point: Cubase has more features, hence it’s not surprising that it hit’s performance to some degree.
it’s all very interesting, for me the fact Cubase 7 performs so much better is the interesting thing here.
Also windows 10 is throwing something into the mix with Cubase 8 as well.
I’ve just tried to open a project I finished last year before I move to windows 10 and I cannot get it to play now on windows 10 so Cubase 8 and windows 10 have some things to sort out!!!
If I can’t open and play a project I did with windows 8.1 in windows 10 on exactly the same machine then somethings not right…
I also have to say going from a MR816x to a RME Raydat connected to the MR816x, has improved performance a lot. Best money I have spent.
When it comes to CPU core performance, in my test they all were at the same level and pretty much maxed. Of cause CPU0, the first core has more to do in a Windows environment, as storage drivers and other system drivers only run on the first core.
Windows 10 looks like it does a better job at distributing, but I would have to run the same project under Win 7 to really make an objective guess.
Anyone of you with performance problems tried to change the CPU affiliation of Cubase by unticking CPU0 (the core that runs all the other stuff) ?
In Studio One you can do it from within its preferences, and it was the only way ,at the time ,for me to run it without audio problems. The problem was the onboard USB3 controller by VIA, after a driver update the problem disappeared.
In Cubase you have to go to the task manager and change it there. That only lasts as long as Cubase is running, next time you would have to do it again.
I’m on a machine I’ve been working daily with for a couple of years so all my points are relate to a known working machine that has performed faultlessly since it’s build.
I’ve been having a feeling of late that the performance wasn’t what I remembered so strated doing some tests.
bottom line is Cubase 7 and reaper perform a lot better for me on this powerful machine of mine .
ASIO guard in Cubase 7 works better too, in Cubase 8 ASIO guard really makes things a lot worse performance wise.
as I said, I cannot open a recent project now with Cubase 8.1 and windows 10 on the SAME MACHINE
I have exactly the same issues. Projects that are a breeze on mixbus 3 and reaper are stuttery cpu hogs on cubase. The issue has got steadly worse since v7.5.
I was very surprised when I read the DAWbench tests (not sure if you’ve seen that?) showing how much performance differs between different interfaces. RME seemed to come out pretty near/at the top. Any idea what kind of performance gain you got in percentage terms/plugin count?
I do love the ‘zero latency DSP’ monitoring features on the MR816 though so would be a shame to lose that.
Thanks for spending the time doing this and posting the results.
I’m fascinated by your results with and without ASIOguard. I did a little test using the Steinberg project “Eight Good Reasons” and most people who tried it, like me, found a very big performance gain by switching on ASIOguard. Specifically, the ‘real time peak’ ASIOmeter dropped to almost zero when ASIOguard was engaged, whereas your results show the real time peak higher than the average meter reading when you engage ASIOguard. I wonder if there is a fundamental difference in the way different Intel processors manage this?
What you said about possible graphical issues: I saw a significant difference in Cubase performance when I changed the default graphic options. However, this difference all but disappeared when I upgraded my BIOS.
I have seen DAW bench test…I worked with VIN to do these tests if you check the small print you’ll see my name there …Marcus C***** !!!
Also. this isn’t about the difference between audio interfaces, I’m well aware of that. This is about the current state of Cubase 8 on windows 10 performance, which isn’t good on my machine any more.
I wasn’t suggesting you have an interface related problem. Just interested to know Peakae’s experience with this because I have the same audio interface as the one he used to have and I have toyed with the idea of getting a PCI based RME interface specifically because of what I’ve seen on DAW Bench.
for me, Reaper never really outperformed Cubase by much over the years, an I’ve been involved since reaper V1.x and Cubase since Cubase III on the ATARI 1024 ST in the 1980’s
Cubase 7,5 runs perfectly and compares very favourably with Reaper in it’s core loading and general performance.
I’d rather go back to using Cubase 7.5 until this is sorted out as I’m a Steinberg guy through and through.
I might if I get time , drop my old windows 8.1 drive back in the DAW an update Cubase to 8.5 and see how it runs.
I think the record I cannot now open was done in Cubase 8.x but on windows 8.1 so maybe there’s an issue here not only with Cubase 8.1 but in conjunction with windows 10.
MC
p.s. I’ve edited the title as C7 is now in the mix.
Cubase 7.5 doesn’t run any better on my machine. Actually 8.5 appears to be better.
In the test I posted previously (8 x Waves Kramer Tape per stereo track, plus 8 instances on the Stereo Out, buffers at 92), track count is as follows (all on Windows 10):
Cubase 7.5 10 tracks -ASIO guard on (only 1 track with ASIO guard off!).
Cubase 8.5 11 tracks -ASIO Guard set to High (only 1 track with ASIO guard off!).
Reaper 16 tracks
Certainly sounds very different to your results. You say ASIO guard makes performance worse. For me, in this particular test, it multiplies performance by 10!
“There are some important tasks related to the consistency of the performance across systems (starting from the next release already). This should benefit in particular the users having performance problems and those who witness a lesser performance compared to the previous version(s). All users should see an improvement, though.”
So, maybe we get some news in the next updates? I for one, noticed the drop in performance in the last versions.
Thats a very interesting test. The only explanation I got is that some plu ns are32 bits and C8 on a 64 bit system react badly to 32 bits plug in (and well to 64 bit plug in) I am not kidding at all. Yesterday I installed a 64 bit plug ins and load a 4GIG patch and C8 was perfectly fine for the Asio 10% or something. Bt if I load SOME old 32 bit plug ins with a small amount of MB for a patch then the asio is almost red… I think THIS is where to look for a solution… I’d say try the same test with different plug ins and yes you should use the same plug in to compare otherwise you say L it should be the same I changed the plug NO !!! its not the same as I said C8 react better to a 64 bit plug in with a patch of 4 GIG than SOME 32 bit plug in with a 100Mb patch… so dont presume that there is no difference if you compare reverb. Use the same plug in for your test whatever they are… We will find a solution to this BTW I should take pics too of my system to prove what I say
Well it’s kind of hard to give a percentage gain, and that would be useless anyway as i’m sure we are using different plugins and have other needs.
First thing I noticed was a much less nervous ASIO meter real-time peak display, and I used to run the 816x at 256 buffers where I now run it on 128 buffers on the Raydat.
The second thing I noticed was that jumping around to other programs while Cubase was playing was before impossible without getting audio glitches. Now playback is rock solid.
Third thing i noticed, on the native driver if ASIO load got around 75-80% I got audio stutter, now it’s solid even at near 100%
One improvement is also the latency is lower using the Raydat, input latency @128 buffers is under 3ms, and the TotalMix monitoring software is far better than the native software.
The main point in the Raydat for me, besides the better drivers and better performance, is that I can for very cheap get 32 Inputs and 32 Outputs and even add another card if I want 64 I/O
What is not to like ?
Well the ADAT implementation on the MR816 is very poor, you will have to set it up using the MR software, and it will not change sample rate on it’s own, you will have to use the MR driver to do that.
The DSP (I LOVE the ROOM reverb in this unit) is pretty much useless, unless you waste a pair of outputs and set up a permanent reverb, that again can only be changed using the native MR software, and thus forcing you to use a firewire connection. (that btw works flawlessly even if you are using the sound going thru the Raydat)
Sorry if this got a bit long and boring for everybody else, but I think it could benefit others that have played with the idea of reusing old ADAT capable audio-units in their setup.
