Reaper significantly outperforms Cubase in DAWBench...why

In my DAW using the latest DAWBench (2017 edition) and the x64 version of the RXC compressor, I can hit 266 instances active before audio break-up. Same exact setup, no changes, same audio driver, same version of the RXC plugin, etc. in Cubase 9.0.20 tops out at 232 instances. Have tried it with ASIO Guard on/off and it’s even worse with ASIO Guard disabled. Yes all the usual performance tweaks in Cubase are on…multi-processing, no difference audio priority set to high, etc. It’s just simply getting outperformed by a $60 DAW.

I love Cubase, been using it for years and years, and I’m not looking to switch. Its current performance is sufficient for what I need it for.

But this is a 12.5% difference, a significant amount. I’d at least like to hear some theories for why this is occurring?

-Ben

This is definitely something Steinberg should look into if they agree that this DAWBench test is legitimate.

I don’t see how you could argue it isn’t…it’s a very practical, real-world way of testing DAW performance, apples-to-apples. Maybe the explanation is just that Reaper handles multi-core processing more effectively, but at least there’d be an explanation for it.

I didn’t use DAWBench, but I have the same experience with Reaper and Pro Tools (and Studio One when I recently tried it)…Cubase falls noticeably behind in CPU efficiency on my computer.

Sorry… I don’t buy that DAWBench is a real world, apples to apples, or practical way of testing your DAW performance… I don’t use Cubase to just see how many compressors I can open, and not get audio drop-outs.

For instance, if you try to use outboard effects using the Reaper plug-in ReaInsert, you’ll soon figure out it doesn’t work, then you’ll realize a real world situation where Cubase out performs Reaper!

I’m not saying Cubase is better than Reaper, as a mater of fact, I happen to like Reaper a lot, but I also like my outboard gear! Until they fix ReaInsert… Reaper is dead in the water for me.

I guess my point is that a 12.5% difference according to DAWBench doesn’t tell the whole story.

I’m of the opinion these days, that it isn’t wise to put all of my eggs into one basket…so to speak. There is no reason to limit myself to one DAW that limits my options during my creative process.

“Real world” to me are tools that give me what I want, when I need It… sometimes that’s Cubase, sometimes that’s Reaper, and at the moment… I’m starting to feel like giving Studio One 3 a shot.

Al

Reaper is great and a lot of people love it.

But DAWbench is totally not a real measure of ‘performance’!

The real measure? take two brothers- twins. But them in the studio and tell them to make a song.
Whoever gets more hits on Soundcloud wins :slight_smile:

My meaning? workflow optimization… controller support… scoring… all these things are more important that CPU efficiency IMO. and if Cubase has to spend clock cycles to support those things… that is OK with me. Does that make sense?

NOT that I am hating on Reaper. People should use whatever tools work for them, and Reaper is a first-class DAW.

Well there are a lot of things we have learnt about Cubase over the years which me know will impact on performance -

There are a significantly large amount of calls to the Syncrosoft dongle as part of the copy protection system.
It’s support for multi-core processors is not great - you only have to look at a cpu meter to see this.
It does not support any GPU acceleration.

On top of this you only have to read the posts towards the top of the forum to see that Steinberg have some serious issues to address regarding performance in C9. Their focus in the last few years has been on new features to drive sales, and I’m afraid stability, the fixing of long standing issues and performance optimisation has taken a back seat.

Hopefully we’ll see them addressing these problems soon.

MoPro and cameron206, great points and I don’t disagree. I didn’t say DAWBench was the only performance measurement that matters, but it is absolutely a relevant measurement. I often work with large mixing projects that might have 150+ tracks and potentially hundreds of insert effects, and DAWBench provides useful insight into just how much of that you can do. So while none of us use Cubase just to see how many insert effects it can run, DAWBench absolutely tells you something useful about the limits of the projects your DAW can handle, in a way which is meaningful to the work I do (versus, say, just raw system benchmarking measurements like PassMark, CPU-Z, etc.)

The reason it’s apples-to-apples in this regard is it shows me just how much the software itself is/isn’t contributing to the overall limit on a given system. Whatever Cubase is doing internally is making it less capable in this sense, which has tangible, real-world repercussions.

It’s definitely not all that matters…I wouldn’t trade Cubase for Reaper for many reasons, not the least of which is I’ve been using Cubase since it was just a MIDI sequencer and so I’m very fast and efficient with it. That doesn’t mean that the above doesn’t matter…you’re talking about a difference (12.5%) that is greater than the performance bump typically seen in new CPU generations (for example the difference between an i7-6800K and the new i7-7800K that replaces it is only 10%). That is not an insignificant performance difference.

Well - Reaper can outperform Cubase for a number of reasons. Reaper is much less complex, with far less utilities than Cubase. And a simpler setup (and less processes to run) consumes less CPU power. I am sure that a simple program like Ardour can outperform Reaper, simply because it’s less complex. So - In my opinion the performance has nothing to do with quality. It’s a PART of the quality of an program, but it’s far from the whole picture.

For example - Cubase 9 Pro has the Control room (what will cost extra processing power by the way), and that is a feature that is extreme valuable for me. As long as Reaper (or another DAW) does not has something familiar (in the same easy way as Cubase), for me Cubase is on top of my “to use” list. There are more things that are more convenient in Cubase, where you have to use work-arounds in other DAW’s, but I will leave it with this example.

Sure - Cubase is not my only DAW. I also use Ableton Live (and even -don’t shoot me- FL Studio), simply because it has a different work flow that is more convenient in different circumstances. Cubase, however, remains my primary DAW that I use most.

It also just might be less efficient in the way it’s programmed, multi-core processing deficiencies being one known example. I’m not so prepared to give Steinberg a pass just because Reaper is simpler. We pay a lot for these sequencing packages and I expect continual improvement, not just more features.

I don’t buy that - Steinberg have implemented things like the plug-in sentinel, the last bug fix was to improve the gui performance on macs etc.

Reaper is a much lighter program so it will out perform Cubase. Thats not to say Steinberg shouldn’t look at DAWs like reaper and try to improve performance in terms of the multicore usage etc.

I can’t say I agree at all. I put some pro/con as I tested from 2010 when I first discovered Reaper.

I think ReaperTeam has a better designed product when it comes to resources and how it operates. Justin Frankel is maybe a better system designer than most on the planet.

Cubase has true nested VCA faders, Reaper not. But instead has true nested track folders with integrated bus, so VCA need is less since you can make self sufficient branches with effect returns that ends in 8-10 stems on top level or whatever you aim for. So single level VCAs goes a long way.

Control Room offer more things than Solo In Front dimming to make essential parts best level for new recordings, but while not activated or used, hardly explain the load on system. Reaper feels featherlight, almost instant startup and operation overall. And a mouse modifier made me adjust SoloInFront dimming level just turning mouse wheel to decide background level compared to recording parts.

One thing bothers me a bit, and that is how Reaper handle pdc - and it adds as many multiples of set asio buffers to handle total plugin delays. If you add a plugin with 1 sample latency - Reaper adds a full asio buffer of 32-64 samples( for most of us). But maybe it is smart resourcewise - you don’t need to preprocess anything like Asio Guard - you just add the outgoing delay and possibly less tricks needed. Reaper does increase roundtrip latency quicker for monitoring through - but maybe gains overall performance. It finally align recorded clips on the sample, no difference there, but realtime monitoring is a hazzle if you need to use that later in mixing process. Common for us composing/arranging while recording.

Maybe the amount of history involved for the product plays a roll. But does not explain why StudioOne feels pretty much like Cubase when it comes to performance. Reaper shines above all in this field, it seems.

ProTools for Windows is a cpu hog, compared to anything almost. It would be Digital Performer that needs optomizing as bad as PT. I think this is to do with that main development is made on Mac, and just transferred to Windows. So developers creating new things do it on Mac first. Maybe this is the case for Cubase too, don’t know how it performs to other Mac daws.

I remember Reaper on Mac at some point had complaints that plugin scanning took forever, so maybe Cockos Team work on Windows primarily. Just a theory.

Reaper is delivered raw, but is more customizable than any daw. Themes, toolbars, mouse modifiers, scripting language, SWS extensions. But first impression is really that you hardly can use it as delivered. I tried hard to like Reaper a couple of times over the years, v5.4 just recently as well, waiting forever for Cubase new video engine to arrive.

I just prefer Cubase as it is still. I feel I got the headroom I want for my projects to grow without switching daw. Anything including notation stuff.

I find Cubase clumpsy in how you do many things, but you learn how to and get it done in the end. Now that stated that next maintenance for Cubase 9 will have new video engine, my hopes are back to stay on Cubase. My overall worry is that they are tempted to ditch Windows 7 soon, I could not get a statement on this from support when this will happend. Don’t feel like buying new computer to update Cubase, I just stop updating Cubase instead. I ran Reaper still on my 32-bit XP computer so they don’t drop OS premature.

Great points Larioso, I’ll have to dig up the document but I did some benchmarking between Cubase on an iMac and comparable Windows PC several years back, it was significantly less capable (as DAWBench measures capability) on the iMac than the PC, which I was surprised in a way to see. But this is a less useful test, not at all apples-to-apples, since there are so many other variables.

Interesting comments about VCAs, in all honesty even as a mix engineer I’ve never found VCAs useful in the DAW world, at least not in Cubase (since a Group track with no effects on it is effectively the same thing).

If you take out the dongle then you can do all sorts of operations without a problem, and only after some time it will find out it’s not inserted… so I can’t imagine it can be that many calls.

Low latency (DAWbench-DSP-C7-RXC-EXT.cpr) performance @ 64 samples / 1.5 ms on my PC:

Cubase 5.5.3 - 100%
Cubase 7.5.4 - 78%
Cubase 8.5.30 - 95%
(7.5 and 8.5 without ASIO guard of course)

So Steinberg made it with 8.5 to return to good old Cubase performance from 5 years ago. For the record: 8.5 with ASIO Guard - 161%.

Different Cubase versions perform very different. If you compare to Reaper, it’s very important to tell exact version and buffer size of both. And to not forget all the things Reaper can’t do for you compared to Cubase.

I think Steinberg did care a lot in the past two years to make Cubase fly. I’ve read that HALion 6 under Windows uses Direct2D to draw its GUI, so it’s likely that Cubase will follow. This adds another dependency (your graphics card must work correctly and be installed fine and the driver version must be ok), but it gives a perspective that performance will increase a bit and GUI will react much better. Simply try MS Edge and scroll a complex website and compare it to Firefox, then you get the idea how redraw performance differs between GDI und D2D.

If you use group channels as effect bus and return that into another bus together with tracks with sends to it - it would be self sufficient - and moving that final bus fader will sound exactly the same with the effect mix just different level. Some duplication of effects may arise, but that might be a sacrifice you would do.

I think Cubase FX channels are obsolete, at least what you got in Elements 7 at the time I looked. Only routed to hardware outs or something. But also read that Elements functionality was stripped of some things regarding this. Group channels do everything so did not feel the need to look in Pro.

I think the FX channels are useful in terms of house keeping if not functionality

Never use FX channels either…just Group channels (but use them in different ways, depending on what I need them for…some as FX busses, others as VCA sums, etc.)

A VCA and a BUS are totally different things.
Anyhow the way reaper works is that every plugin gets assigned its own thread.
put 8 Ozone plugins (or any other heavy plugin) on the same channel and you will see an even CPU load over the cores.
Do the same thing in Cubase and you will see one “core”/thread maxed out, as one channel with the 8 plugins gets assigned to one thread.

Afaik there’s no point in threading serial operations. You can’t execute them simultaneously if you need plugin 1’s information for plugin 2, and plugin 2’s information for plugin 3, etc.