Are all hosts share more or less the same CPU performance when running plugins?

Audio Plugin Hosts and other audio software applications discussion
Post Reply New Topic
RELATED
PRODUCTS

Post

Out of curiosity, I ran a few stress tests with CPU intensive plugins on Reaper and Tracktion and its really hard to say if there is any difference in managing CPU intenstive tasks...any ideas?

Post

Boy this can turn out to be a good thread to watch. There are a lot of things other than the Host's or VST's that contribute to poor or better CPU optimization.

Audio cards, interfaces OS's and countless other factors can change between User's that an accurate assessment will prove different.

I will wait and keep an eye on this thread as I think there may be some interesting comparisons but it also could be useful for some of us.

Post

Depends on the plugin.
And on whether the host properly supports multi-core or not.
Scale above 2 cores and you'll find reaper does better than tracktion 3, for that reason alone.
Mostly it comes down to how the host divies the plugins out to different cores, nowadays - other differences are very small, performance-wise.

Post

To my observations:

Nearly all hosts do +-10% the same job about plugin performance except one:

Reaper!!!

Their "Anticipative" processing, which afaik is a kind of prerendering works really really well...

In most tests I did for myself, I could run about 30-40% more content in Reaper than in other hosts, which in real numbers were 15-20 instances MORE ...

That's really a lot...
Nevertheless, these are test, which will never reflect a real life situation, where often smart disabling techniques like in FL Studio and MuLab work very efficient too...

And: All host behave more or less equal at running CPU hogs like DIVA, Diversion, LUSH and others... if a single patch eats 60% CPU in one host, you'll have the same situations in the others too

Differences are mostly only noticeable by running "normal" plugins

Post

Live is more hungry, probably 20% less plug performance than the others.
dedication to flying

Post

In theory, no, in reality, not unless there's something wrong with one of them.

All of the hardcore processing that takes CPU time is done within the plugin's code, regardless of which host it's in. As long as the host does a good job of scheduling different plugins on different cores(which is not a given), then there's little possibility for difference. Then there's the possibility of oddball stuff like the host passing unaligned buffers to the plugins that can also hurt performance, but I think most developers figure out that pretty quickly, due to the sheer size of the performance hit.

Some hosts also do smarter disabling of plugins when they're not doing anything, like Reaper, for example... That means that if DAW-XYZ isn't disabling plugins at idle, and is idling at 20% CPU and Reaper at 1%, then it will always have that 19% disadvantage during playback, because most people will have more idle plugins than active during any point in playback.

But then again, there's even more factors as to when dropouts start to occur at a given latency, so even if DAW-A uses less CPU at idle, it could start having dropouts at 50% CPU usage due to internal inefficiency and delays, and DAW-B idles at 20%, but can make it to 80% before dropouts occur because the audio/MIDI engine is more efficient.

Post

jeffh wrote:in reality, not unless there's something wrong with one of them.
Well, there's nothing wrong with Reaper, so I guess there's something wrong with most others? ;)

Post

EvilDragon wrote:
jeffh wrote:in reality, not unless there's something wrong with one of them.
Well, there's nothing wrong with Reaper, so I guess there's something wrong with most others? ;)
Eh... I don't think you've qualified that statement with hard numbers or probably even any worthwhile amount of personal experience, but you have made it quite clear which DAW is your favorite :D

EDIT: but yes, I've used Reaper, and it does a decent job of scheduling and sleeping the plugins, but it's hardly in a league of it's own..

Post

Well, I've used FL Studio, Sonar, and now Reaper, and its CPU utilization is clearly the best of the three from my experience, since the same damn plugin is using less CPU in it for whatever reason (betterer multithreading in Reaper? I dunno, but it works), and I get to stuff more instances of it before crackles occur. It's magic. :)

Post

ASIO driver that your audio interface uses makes a difference, too. Some are more efficient than others. Like RME or Lynx in comparison to TC Electronic or M-Audio ASIO driver. I've got TC Konnekt, RME 9652, M-audio 2496, and Creative SB Hyperultramusicsomething [right..], so I'm aware of the difference between these drivers. RME wins hands down in efficiency. Don't even start with ASIO4All or Creative ASIO... M-Audio wins in that case, especially at low latencies like 64-128 samples. At 256 samples up the difference isn't so big, and if you're satisfied with 512 samples of latency, you shouldn't care about anything I wrote. Probably. Perhaps. :D

Even SB Hyperultramusicblah works OK at 512 samples, or ASIO4all. But CPU struggles a bit more, too.

For me, Reaper uses less CPU than EnergyXT 1, yes. It works great. I have nothing else so I can't compare with anything else, but I think I remember EnergyXT 1 being somewhere in the middle of the "efficiency chart". :D Either way I'm a happy camper with Reaper. I wonder if new Traction works better than the old one with this new audio engine. Traction has a lovely, very streamlined, user interface. I've always been rather fond of it, but hesitated to even try it due to being rather inefficient and buggy.

Cheers!
Last edited by DuX on Sat Feb 09, 2013 12:03 am, edited 1 time in total.
It is no measure of health to be well adjusted to a profoundly sick society. - Jiddu Krishnamurti

Post

I know from first hand experience (on a mac at least) that different daws running the same plugins will tax your CPU differently. I did this test with Logic VS cubase running the exact same mix with the exact same plugins, and the mix which would run at like 95% CPU on Logic 9 (basically maxing out) would not even come CLOSE to running successfully on cubase 6. Even after removing all my amplitubes (in which I was running like 6 of them) Cubase would still not run it the mix successfully.

Now this was a "real world" test and I was running a VERY large project using a ton of different plugins, virtual instruments, and audio tracks on a full mix. Normal projects would run fine on Cubase, or you would find work arounds (freezing, bouncing stems, etc). Its also important to note that Logic runs AU plugins while Cubase runs VST, so its conceivable that some of these 3rd party plugins are better optimized for AU vs VST and its not 100% the hosts fault. It would be interesting for someone to do a test in Reaper running all AU's in one project, then the exact same mix as just VST's in a different project, and see if the CPU hit is the same.

That being said, it seems to be a "well known fact" that logic is very CPU efficient by design. I dont know how or why, but do a search for "logic efficiency" in google and you will find tons of people who have run similar tests finding Logic to be a very CPU efficient DAW.
My progressive rock band - free demos here!! (and if you do listen please let me know what you think!) http://www.aeonsatori.com/news/free-downloads

Post

In the old times, Cubase was more efficient than Logic [5.5] on Windows. Yeah, it also depends on the OS you're running. Reaper probably doesn't work so efficiently under OS-X, since the OS-X version development lags a bit behind the Windows version. It would be interesting to know how efficiently it works under OS-X. ;) I know for a fact that Cubase on OS-X doesn't work as efficiently as on Windows, so Cubase lovers shouldn't go OS-X, since the difference isn't so small. Rather like 20-30% on average, and that's a lot. However, many people don't give a f*** about efficiency unless they peg the CPU in every project... :D I do, however.
Last edited by DuX on Sat Feb 09, 2013 12:15 am, edited 1 time in total.
It is no measure of health to be well adjusted to a profoundly sick society. - Jiddu Krishnamurti

Post

lolz

I like this subject, but this will end badly.

I can take the same exact computer with the same exact plugs and get strange differences between them.

Reap is best, by far, not even close. Did I mention how much better it is?

Sonar is second, but uses the first core and then starts fanning stuff out on the other 7....eventually. Kinda strange

Live actually supports a ton of plugs, but it's weird too. I can load one instance of a plug and it will say 56% cpu. Then I can add oh.......GOBS more and it will sometimes lower the CPU use :nutter: RARELY does it spike. I've got a couple of projects with so many cpu intensive amp sims (like 8 tracks) and all kinds of comps/eq's all over the place and it says......you guessed it......about 50% :shrug:

Studio one was a nightmare, and it's still kinda scary, but I'm not getting the cpu spikes like I was prior to 2.5 (kontakt was particularly bad) It's now similar to live......you look like you are running out quickly and you can keep adding tracks and it's fine. Very strange.

Yep, I still have FL and it's still the worst by far :lol:

Anyways, this is all on the same computer with the same specs so, it is what it is.

Post

EvilDragon wrote:Well, I've used FL Studio, Sonar, and now Reaper, and its CPU utilization is clearly the best of the three from my experience, since the same damn plugin is using less CPU in it for whatever reason (betterer multithreading in Reaper? I dunno, but it works), and I get to stuff more instances of it before crackles occur. It's magic. :)
Well, Reaper is better than FL Studio and Sonar for sure (at least the versions I tried 7-10(?) years ago), but Cubase dominates the market for a reason... Despite Steinberg's ridiculous treatment of users...

OTOH, it's not like Reaper users are sitting pretty on Atom netbooks while Cubase, Logic, etc... users are sitting around waiting for faster 16 core CPUs to come out so that they can finally make some music. At some point years back, it became fanboy elitism to try to run projects with 300 tracks at 8 samples of latency, while real musicians are out there making great music with far fewer tracks and at far higher latency ;)

Don't make me start telling stories about running Cubase SX1 on a single core *gasp* 2.0GHZ *gasp* CPU :P

Post

Logic is efficient because it hardly uses any CPU for a plugin that's loaded unless that plugin is processing audio.
You are currently reading my signature.

Post Reply

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