Things I hate about... - Cubase 7

[Compyfox]

Have you actually "played back" your project where you maxed out with all synths? It should stutter like crazy or even shut off the stream. That is unacceptable, and also unreasonable to constantly push everything over 100% of load.

Neither do we still see what the "upper limit" is - it's a guessing game that I won't play along.


And this boils down to what you said:

You need to see how much work it can do.

If the "Guard Meter" is constantly at it's limit, even already "peaking red", I run into severe issues real soon. The CPU usage might not be high. But what does that help me if I have a suitable CPU (oversized even) if I run out of processing power within a short while and still didn't scratch 30% of my CPU. Should I just continue to "push more"? That's just stupid and totally lazy from sides of Steinberg. "This system is superior - it should work this way".

Not really!



In your case: 14-16 instances of FXpansions Strobe, and with merely 8 Instances of Prologue (which is elder than Strobe(!)). Some people don't even use "only" 4-5 Synths, but more. And on top of that, FX in real time.

You are lucky that your values are so low at 32samples!

The test project I got, with these values at 128 samples, I can add 3 more channels, then my stream shuts off. 4-5 if I use 256 samples. Do you think that is okay?



Again - if you like constantly running in the red, and you think that is correct - fine. But this is not(!!!) how it should work, neither how it should be.

To me, that is unacceptable. No reason to further debate this and "convince" me what I'm doing wrong, and that I need to accept this. Won't happen.

Period!

[mkdr]

Have you actually "played back" your projec where you maxed out with all synths? It should stutter like crazy. That unacceptable, and also unreasonable to constantly push everything over 100% of load.

Yes. None of them glitch or stutter like crazy. Though the C6 one does blink the light on occasion. I wanted to make sure i was getting the absolute maximum i can get out of it. It also crackled about once a second or two. C7 had no issues even though it's right at the limit of the Asio Meter.

And this boils down to what you said:

You need to see how much work it can do.

If the "Guard Meter" is constantly at it's limit, even already "peaking red", I run into severe issues real soon. The CPU usage might not be high. But what does that help me if I have a suitable CPU (oversized even) if I run out of processing power within a short while and still don't scratch 30% of my CPU. Should I just continue to "push more"? That's just stupid and totally lazy from sides of Steinberg.

No. That was not the point. Those examples are run to the limit so that it's still usable. And they stay that way too. I'm just saying that you need to find the maximum of your setup. Only then you can conclude which one runs better. Not from the "minimum numbers" or something in the middle.. like the 30%-ish you had on your pdf.

In your case: 14-16 instances of FXpansions Strobe, and with merely 8 Instances of Prologue. Proglogue - which is elder than Strobe(!). Some people don't even use "only" 4-5 Synth, but more. And on top of that, FX in real time.

Yes, well most of the CPU usage is from the effects.. there's quite a few after Prologue. I just needed some noise in there so that the FX won't shut down. And Prologue is light and included on both C6 and C7. Strobe had less effects.

You are lucky that your values are so low at 32samples!

That is what is so great with Asio Guard. It doesn't run everything at the insanely low latency. Just the things you are editing right now.

Again - if you like constantly running in the red, and you think that is correct - fine. But this is not(!!!) how it should work, neither how it should be.

To me, that is unacceptable. No reason to further debate this and "convince" me what I'm doing wrong, and that I need to accept this. Won't happen.

Period!

I'm not running "in the red". The test is just at the limit of not going to red.


Try out the project. You can remove the effects if it's too high on cpu usage.
Actually below is one where they are bypassed and there's only one channel. Just enable enough effects to get the Asio Meter upwards.. and duplicate tracks.

https://dl.dropboxusercontent.com/u/880 ... 0oneCh.cpr

[mkdr]

The test project I got, with these values at 128 samples, I can add 3 more channels, then my stream shuts off. 4-5 if I use 256 samples. Do you think that is okay?

It all depends on what you usually get. The idea was that you could benchmark the difference of C6 and C7. It's not a contest..

[Compyfox]

No. That was not the point. Those examples are run to the limit so that it's still usable. And they stay that way too. I'm just saying that you need to find the maximum of your setup. Only then you can conclude which one runs better. Not from the "minimum numbers" or something in the middle.. like the 30%-ish you had on your pdf.

Again - this is a test project that Steinberg provided me. They sure would have a reson to do so, no?

In real world examples, I'm currently happy to even reach 12 channels with about 3-4 inserts each. That - is not much. Compared to C6 - where I could use twice as much and then some.

I thought about hardware upgrading - but thanks to one fellow so far (still waiting for further results?!), this clearly showed me that I won't see a significant usability boost.

It all depends on what you usually get. The idea was that you could benchmark the difference of C6 and C7. It's not a contest..

Once more:
What do you think I did the last couple of months?

Plug my nose? Down some pills and halucinate?

That is what is so great with Asio Guard. It doesn't run everything at the insanely low latency. Just the things you are editing right now.

Which still doesn't work on my end. Again - consult the PDF. The "Guard Values" (which do count if it's on) are too high, if the Guard is off (then the realtime meter counts) skyrockets.

Again - compare the values with the AVG Guard Meter (On) and Realtime Meter (off) to the C6.x values. They're usually 20% higher, if not even up to 50%.

Unacceptable! (I can't emphasize that enough)

I'm not running "in the red". The test is just at the limit of not going to red.

To be honest, this test isn't even a test. It's just one... random thing to show what's working and what not - with Prologue even just one simple bassline. You still seem to fail the point of all this.

Try out the project.

Thanks, but no thanks. I've tested ENOUGH in the last couple of months, not counting the last couple of days with that EIST nonsense (nobody even commented on that!).

These "tests" clearly show that there is something not right. You don't want to believe it, I don't want to accept it as paying customer - Steinberg doesn't further comment on it either.


STALEMATE.




One final time:
Steinberg is aware of all this - they are working on it - they mentioned to me however that this might not be drastically improved or fixed within the C7.x cycle.

The accessibility of the test project is non-relevant at this point. Or do you ask "magazines" to reveal their test projects as well so that you can "confirm yourself" that they are wrong in your opinion?





Can we stop this kindergarden nonsense now?

[V0RT3X]

So what other DAW would you recommend compyfox since you aren't keen on Cubase 7?

[egbert]

I would say Compy wants Cubase but he wants it done right
I imagine many others who have paid Steiny several thousand smackers over the years would like the same.

For me, it is the same as MS effing up Windows. It is not really a practical or financially viable option for me to switch to another platform - too much invested in the platform. I want MS to focus on not screwing the pooch with the platform that supports my tools.

On Windows, there are other DAWs and I use them too but for some tasks I want scoring and sequencing and everything working on the same project. Studio One is improving will likely have a score editor soon since Presonus bought Notion, Reaper is very solid but neither are as mature as Cubase. If Steinberg focused on making Cubase run like a swiss watch many would be satisfied - all the sandboxed plugins etc I could do without.

[SJ_Digriz]

Since I didn't find videos about this, care to post some screenshots?

I'd love to at least keep the Metering on the mixer. There is no demo yet, so no testing until Febrary (where it's supposed to hit latest).

So the Meter requires you to click the meter tab (top right), which exposes different tabs at the bottom for working with the Meters. There is no "bigger" version of the control panel. That screen shot is it.

[chk071]

So what other DAW would you recommend compyfox since you aren't keen on Cubase 7?

I think the only host feasible would have to be programmed by himself. That's the only way he could be satisfied. Or not, but at least he wouldn't be moaning all day long.

[egbert]

So what other DAW would recommend compyfox since you aren't keen on Cubase 7?

I think the only host feasible would have to be programmed by himself. That's the only way he could be satisfied. Or not, but at least he wouldn't be moaning all day long.

Look - that is just shooting the messenger.

Here in Oz, we had a situation where a person was driving their VW car along a freeway in Melbourne at 60 mph or so and it suddenly decelerated drastically. A large vehicle ran into the back of the VW and the results were unpleasant shall we say. After this crash, many other drivers came forward and reported similar isolated instances of this sudden deceleration in traffic. There was nothing the drivers could do and it was very dangerous. VW said "nothing to see here, move along" and denied there was a problem until enough noise was made and it was confirmed there was a bug in some software which resulted in this problem. I gather there is/was some kind of recall to address this potentially fatal error.

Point is - standard corporate behaviour is to deny/ignore the problem until they are held to account. It is only when people gather evidence and stick to their guns that problems are acknowledged and addressed. While I don't want to read about this in every thread on Cubase, this thread was started by Compyfox and is about this problem. If you don't want to hear about it don't read this thread. But really, it makes more sense to support efforts to improve the product by giving forthright feedback on any problems.

[SJ_Digriz]

@compyfox, I'm confused ... don't you have C6 and C7? you could have popped up the control room on both. The difference is obvious and humongous.

[Tp3]

Excuse my that I interrupt, gentleman, but I have one (potentially) stupid question :

Were the meter compared 1:1 visually ? (ie. you SEE 25% load on 6.5 meter and you SEE 35% load on Cubase 7.xx meter)

Forgive if I haven't read the whole thread...

[chk071]

@ egbert: You remember the thing with the stuck brake pedals of Toyota cars in america, when they had to pull back lots of cars to repair them? Well, in the end, it turned out, that the driver made a mistake when operating the brakes when he did the accident which set loose that thing. Don't want to take that as an example for this topic, but as you can see, there's the other side too.

But anyway, i don't see how either of these examples would fit here, and i can't be bothered to be involved anyway, just saying that it's a bit over the edge, since months.

[SJ_Digriz]

Excuse my that I interrupt, gentleman, but I have one (potentially) stupid question :

Were the meter compared 1:1 visually ? (ie. you SEE 25% load on 6.5 meter and you SEE 35% load on Cubase 7.xx meter)

Forgive if I haven't read the whole thread...

It's not quite that simple either. There was no ASIO Guard in C6. It changes the loading characteristics significantly. It should scale better, but costs you calculation overhead in various states. In practice it works in a few scenarios if you manage it very closely. Evidently more so on Mac than on Windows. But I guess it has other not so good side affects. But, I just leave it off as I have not had performance issues to solve.

[Tp3]

Again, I was asking a simple question really : were the results based on a "WYSIWYG" basis ? (that is : Meter show 40% load ? then 40% it is. and for the sake of simplicity and comparability - with ASIO Guard OFF)

EDIT : OK, I can see that is exactly the method of checking. FWIW : I remember vividly one of Steinberg mods saying that C6 meter and C7 meter results and NOT interchangeable (that is : you CANNOT pitch one result against the other. it is based on a completely different set of rules).

In MY experience, DAWBench (the ONLY bench I count on) gave me the same results with cubase 7.5 but that was not the real surprise. what struck me dead was the fact that I can achieve the SAME RESULTS with HALF THE LATENCY (128 <=> 64)

[Compyfox]

There is no "bigger" version of the control panel. That screen shot is it.

@compyfox, I'm confused ... don't you have C6 and C7? you could have popped up the control room on both. The difference is obvious and humongous.

And here I thought, we got a new Control Room. That is the one I'm used to now. Since that screen is from 7.5 - they did change something for the better visually - but the font size is still funked up. And I think the meters as well (plain RMS meter "inline" with the Peak Meter, EBU R-128 meter still not able to change the scale or the color codes).

Never really got used to the C6 concept, but... it was at least cool to look at.

Were the meter compared 1:1 visually ? (ie. you SEE 25% load on 6.5 meter and you SEE 35% load on Cubase 7.xx meter)

Forgive if I haven't read the whole thread...

The meters can't be compared 1:1, since the new meter is not only based upon a new "engine" (which is where this issues boil down to), they are also hard to read.

I filed several FR's since v7.02 at least to reincorporate thin lines to show at least 10% increments, or a numeric value. Instead it's in 25% steps and really hard to read. After a weeks of testing however, my mind adjusted to the meters and I could find a "somewhat" value that fit to the meter readout.

So yes - it's a WYSIWYG readout comparison.

In MY experience, DAWBench (the ONLY bench I count on) gave me the same results with cubase 7.5 but that was not the real surprise. what struck me dead was the fact that I can achieve the SAME RESULTS with HALF THE LATENCY (128 <=> 64)

Sorry, but these "benchmark tests" use too many demo versions for my liking. Why can#t this test be performed with freeware or host internal stuff.

I did, with the project I got by Steinberg, and the results could be found in the PDF on page 28. It's not about "how much you can run" and "how much can you abuse the new system" (sticking to the reds). It's about "how former Cubase versions respond compared to C7.x".

In my case (opinion), it performs worse. And I only used what Steinberg used - host internal stuff. So that rules out a user error or stuff involved with third parties.

Still, it is not okay to have a VST or even VSTi run fine in C6 with (say) 3-5% ASIO load before, but twice as much in C7. As is the case with Slate Digital and Variety of Sound for example. Kontakt also performs worse on my end. And I never had a load of over 30% with Musiclab RealGuitar - only in C7.

But anyway, i don't see how either of these examples would fit here, and i can't be bothered to be involved anyway, just saying that it's a bit over the edge, since months.

Of course it's a two sided medal. But it's still hard to understand for me that so many people simply said "oh... performance is lower - cool", while I say "folks - I can not work as you advertised - and I won't give in to you forcing me to update and keep up with your nonsense - or upgrade my hardware to compensate". Which turned out to be not working btw (see the PDF).

I don't want to constantly drive things into the "reds" - I want a healthy headroom. I mentioned that to Steinberg months ago, I pointed that out in here on KVR. Nobody wanted to believe me. Now that there are results I can post (I didn't want to keep this confidental anymore - and I wanted to back up my findings), it's still down to "you're doing it wrong, mate".

I showed these results to two professional rig builders by now as well - and the only response I got was a clear headshake. On top of some nice "behind the hand" horror stories they have encountered.


In the end, you as user have to decide if you keep up with this or not. I don't want to, so I made the users aware of this.

Just like SJ_Digriz made the users aware of missing features and some other hickups, SODDI of the lazy initialisation bug, TheoM of the Maverick issue (shortly before it was publically announced), etc.


Maybe my i7 920 is too outdated after all (which shouldn't be with extreme low CPU tools like MeldaProduction, Variety of Sound, Kontakt 4, Slate Digital drasically optimized their ASIO load as well, etc). In this case, I'm open for sponsorship to test further. Else, I'm done with this.

Currently not even interested in updating Cubase - since I'm also not cool with the idea to "pay even further" what they messed up and disimproved in the first place. Just to even "get" updates now as well.