SX / Nuendo Performance Loss - A Myth ?

[Sascha Franck]

SX2, significant copy protection/encryption embedded
SX3, much much more copy protection/encryption embedded
Both the same playback performance.

I allready said so in another thread: We will never know how much this is indeed affecting performance.

Yes, however as has been pointed out many times in this thread it can't automatically be inferred that any performance difference is due to the copy protection overhead as you know SX3 has many more features.

I agree on the copy protection part...
But I don't agree on the "many more features" thing. I do know that full PDC is expected to add a certain CPU hit. But that shouldn't add more than 5-10% at best.
Then, which of the other features should add anything more, unless I actually use them?
And fwiw, Logic 7.1 (assuming it's running well... which it clearly isn't on all machines) seems to be just as efficient - and it's got a close to identical feature set. Due to the absence of any Mac at my place I can't do any scientific tests.

The issues of application performance and copy protection overhead are now being mixed, which is simply going to add more confusion to the myth for other users.

Yeah well... you're probably right on that one.

Furthermore, anyone testing application performance must ensure ASIO buffer settings, number of ASIO I/O channels enabled etc are all constant between tests.

I'm not new to any comparison tests, so my settings obviously were 100% identic.

"Constrain delay compensation" doesn't switch off the PDC engine from what I can tell from the user manual, it just ignores the delay for recording purposes, I don't think full PDC can be fully switched off in SX3. [/quote]

Yeah, that's what I allready suspected.
Would even be understandable that with "constrain delay compensation" CPU hit is slightly higher, as tracks had to be treated differently.

[ericj23]

OK i found sascha results interesting - I wondered if cubase's poor showing was due to its low latency performance - I also remembered from my logic days that you could stack the fx up to the cpu's limit - something sx isnt so hot at

so i altered my original test so that the cpu useage was up at cubase's max - i increased the audio tracks from 6 to 12 - each group of 3 contained a reaktor 5 flatblaster and a kjaerhaus auto filter (so 4 of each)

kontakt 2 with 3 instruments - one of which was a large mono guitar with had a reaktor 5 analogical filter (or whatever its called) followed by the demo of camel audios distortion thingie

another 3 voice instrument had the camelspace dub thing preset on it

finally there was a fm7 playing 6 voices chord and a mono pro 53 using 8 voices in unison

cubase would crackle occasionally - but oddly enough on my set up it played better at 10ms latency than 20 - cpu 80 odd ram about 500

set this up in tracktion 2 demo - it just wouldnt play back ! - cpu and ram useage were much lower but whatever i tried got me nowhere -just crackles

So i give up - i don't know why logic performs better but i suspect that it is just better at high loads - any chance of cn=onfirming my suspicions by running your comparison where the cpu is not being run into the ground - just to see whther there is still a significant difference

or we could just call it a day and go and do soemthing more useful

[egbert]

I am not sure why people expect PDC to consume CPU at all. It is simply a different (variable) offset for each of the audio streams. The audio streams are already buffered and PDC simply involves an extra adjustable buffer for each channel according to the total added latency of all the latency inducing plugins in it's path to an audio output. If there are no latency inducing plugins, the extra buffer is zero samples.

I could see why PDC might consume a bit of extra RAM - more audio might need to be buffered there - but as far as CPU cylces - nada.

Eg

[carmen]

theres definitely a performance loss when using SXNuendo. try Ardour on Linux, or Sawstudio, if you want software not boggled down by bloated code, a crappy underlying OS, and copy protection which just makes things worse..

[egbert]

theres definitely a performance loss when using SXNuendo. try Ardour on Linux, or Sawstudio, if you want software not boggled down by bloated code, a crappy underlying OS, and copy protection which just makes things worse..

Sounds like a really scientific comparison.

[Josmoker]

theres definitely a performance loss when using SXNuendo. try Ardour on Linux, or Sawstudio, if you want software not boggled down by bloated code, a crappy underlying OS, and copy protection which just makes things worse..

why not go all the way, pick up a guitar or something and bypass all those friggin computers!

[martian]

i think sashas test is completely invalid. compare cubase vst to logic 5 and maybe more valid.

does logic 5.5 even use 32bit float signal path? and you cant completely switch pdc off in sx, its part of the whole structure of the program.

do you really expect an ferrari to have the same fuel consption as a lupo (slightly exagurated for the point )?

[stag]

There are some features on SX that are a very important way of manage the CPU power.
Background rendering and offline processing, aside from freeze which i don´t like because of the slowing impact on workflow .
I have 2 hard drives and SX can stream 70 mano audio tracks at 32 bits, without i noticing any disk activity, i know of people capable of stream 250 stereo audio tracks.
I can´t say anything about others performance, i have used logic 5.5 on PC a long ago and didn´t grab my attention enough to evaluate it but AFAIK, background rendering and offline processing, features that virtually let the app run virtually forever are not present at all.
Then there is workflow, i like more SX than anything else, it doesn´t prove it´s better, it only proves that the programers at Steinberg do think as i , or program with people like i in mind, The transposition of hardware into software maybe an old paradigm, but still works when people don´t want to read lots of manuals instead of recording a simple pop tune: some synths, drums, bass, guitar and ocasional orchertral arrangements.
Well if someone ask me if it´s a good options to trade sequencers or platform, because of what is said and written anwhere, i´m not in favour of trading, unless someone is very unsatisfied with what he has at the present time.
I´ll never forget how hard it were my begginer days... learn to make music on a PC, by myself, was rather difficult, i don´t want to take that inconvinience anymore, i´d rather use my time on my lame projects or other things.

[Sascha Franck]

does logic 5.5 even use 32bit float signal path? and you cant completely switch pdc off in sx, its part of the whole structure of the program.

Logics internal resolution is at 32bit float.
So where's the difference?
It doesn't handle 32bit float files, that's right. But I don't use them anyways.

do you really expect an ferrari to have the same fuel consption as a lupo (slightly exagurated for the point )?

That's a nice Ferrari you got there... losing almost any race.
Seriously, there's SO many areas where Logic still outperforms SX big time - the Ferrari statement doesn't hold much water, if any at all.

[Sascha Franck]

There are some features on SX that are a very important way of manage the CPU power.
Background rendering and offline processing, aside from freeze which i don´t like because of the slowing impact on workflow .

But that doesn't help. I prefer my stuff being realtime-enabled as long as possible.

[Scotrick]

I do think we can't compare Logic 5 with Cubase SX 3... a better comparison might be Logic 5 with Cubase SX 1...

it's a shame, but comparing computer's performance and software performance and DAW performance is pretty hopeless... there's just too many damn variables...

and the most important variable, to my way of thinking?

The User Interface, and whether the program makes some sort of 'sense,' in its layout...

all this performance talk is fine, but finally, would it be better to have a program that can play one more Reverb or would it be better to have a host you're comfortable with and like the way it works, user interface wise?

Over all productivity is what it's about... and there are so many variables involved, personal preference and computer set-up, that trying to compare is very very difficult, as has been demonstrated by this thread...

I too felt a little twinge of sadness when I read the hacker's report, here on this forum and at Cubase.net... but not because I was worried about the performance of the program, but rather because all of the hacking and cracking and stealing and dongles and copy protection have left us in quite a muddle... while Steinberg 'wastes,' time programming the protection into their software, H2O 'wastes,' time cracking it and around and around we go... and then we all come here and talk about it...

[Sascha Franck]

I do think we can't compare Logic 5 with Cubase SX 3... a better comparison might be Logic 5 with Cubase SX 1...

Why?
Just because SX 3 excels in some areas?
Logic 5 excels in others. Where's the point?

[Alan]

you can't simply compare to SX2 or SX1 since all Cubase versions always contained encrypted dongle-calls in the code, even the Atari versions back in the 90s!
So this kind of "shit" has a long tradition at Steinberg...

cheers,
Chris

I did notice a lot of complaining on the Nuendo forums concerning performance when 2.0 arrived with the new USB dongle,I also remember moderators suggesting system upgrades as well.Cubase/Nuendo definitely took a performance hit as soon as the USB dongle was introduced and I remeber it clearly as one of the reasons I decided not to upgrade Nuendo past 1.6.Dongle calls may have been present since the Atari days but the introduction of the USB dongle was clearly the time users noticed a performance difference.My guess is there is more going on here and I think there is a lot to be said in Steinbergs lack of response,but then again anything they say will hurt them either way.I'm just glad I Followed my instincts and continued on with Cakewalk/Sonar rather than upgrading.

[stag]

There are some features on SX that are a very important way of manage the CPU power.
Background rendering and offline processing, aside from freeze which i don´t like because of the slowing impact on workflow .

But that doesn't help. I prefer my stuff being realtime-enabled as long as possible.

Well, i too but on SX before crashing it recommends politely to save your work. you know next time, is time for rendering, it´s a workaround , probably done too early but since offline offers unlimited undos, is the best solution i know to keep up.

[talkbanana2000]

AUTO-ADMIN: Non-MP3, WAV, OGG, SoundCloud, YouTube, Vimeo, Twitter and Facebook links in this post have been protected automatically. Once the member reaches 5 posts the links will function as normal.

Sorry. Sonar is not the answer to all. I finally had to give up on Sonar 3 cause it was spazzy on Vstis. Sanity returned when I went back to Sx 2.35.

Well, if you look here (http://forum.nuendo.com/phpbb2/viewtopic.php?t=5127), you'll see that they're pretty upfront about having tried the crack and compared it to the protected version. The bullet point that sticks out for me is: "There is no reliably measurable difference in CPU and disk resource use between the legitimate software products and the illegal versions". You can choose to believe Steinberg or not.

Hello Houston,

Thanks for posting the link to the Steinberg statement. I have been very concerned by the issue of the usage of CPU by the protection system. This prompted me to cancel my order for upgrades to Cubase SX 3 and all other Steinberg software. I have also purchased Sonar 4 Producer Edition. I am glad that Steinberg has finally replied to the questions.

I have done some independent testing with Sonar 4. At this time it seems that the overhead of the protection system is signficantly less than the overhead of the VST wrapper in Sonar.

My test was with the following project on both platforms:

- 5 instances of VirSyn TERA 2 with 35 active presets
- 35 MIDI tracks

On this project, Sonar 4 uses between 20% and 30% more CPU than Cubase SX 2. This is at idle, at peak, and on average. The CPU was monitored using Windows XP Performance and Counter logs feature for both tests.

With monitoring turned off, and observing only the built-in CPU monitors in the respective programs, Sonar 4 is much closer in performance to Cubase SX 2. But Sonar 4 still looks to be using about 5% to 10% more CPU than Cubase SX.

At least on playback of projects with VST instruments only, and no host based effects, the CPU impact of the protection system is within acceptable limits on SX2.

There are two differences between the two platforms. These are:

The 'first note glitch'; and
long close times.

The 'first note glitch' is the first note on a part produces a CPU spike on the first playback of a project. This happens only in Cubase SX 2.

Sonar 4 closes much more quickly than Cubase SX 2.

This is my experience with one project and one synth. I do not consider this to be conclusive. Further comparisons between the two platforms are needed. This includes synths with both DXi and VST versions; quality of included effects; and audio project comparisons.

Sincerely,

George