Who uses Cantabile live (and no synth hardware)

[TiUser]

I am comfortable with my statement as it refers to principles.

It's absolutely right that you need to look at the context for practical use. Maybe an pipe organ simulation is spot on with 512samples buffer @44.1kHz and that might cause very little practical problems on todays PC's

"low latency" needs to be specified if we want to discuss this.

It depends on the sampling rate which buffer sizes result in which latencies. Additionally there is some midi latency. I personally am comfortable with 44.1kHz stereo output (which does not mean I see no need for internal oversampling). I still think it's very cpu inefficient to run buffer sizes below 128 samples (~3ms) on a pc. If we could easily get at 64samples (~1.5ms) I would start to say the topic gets less interesting to be discussed. But I think unless windows scheduling changes in some way that will not be possible yet (without wasting a lot of cpu power).

It's true that you can run more plugins on modern pc's - but you can't break the barriers to run them really much faster audio wise. If you are practially comfortable with this is another story.

Again, I think most musicians are no computer geeks. It's not the point wether musicians are serious about what they do - I guess everyone who performs on stage is serious about what he is doing - amateurs and pros alike.

My point is not you can manage things if you are careful. My point is finally that if a non tech geek buys a pc and installs any live host software and some plugins there is no automatic and perfect control that avoids cracks and pops - the traditional way by reducing polyphny in a more or less clever way - keeping audio out a foolproof thing.

So the answer to the question who uses cantabile live and does it work reliable can not be answered without looking at all detail facets. I really think we agree in general - just we point on different facets.

[pinkcanaru]

I am comfortable with my statement as it refers to principles.

So, in principle, "One thing is "latency". This is rarely discussed with (todays) dedicated music HW but a usual topic in a windows software based setups. If you configure small audio buffers - which is welcome for live playing - you can easily outperform any windows based pc."that is still true in your book.
I guess it's time to say I am in complete disagreement with that principle, as much as I, and I am sure many others, are grateful for the excellent input and help you provide to Cantabile users here. I hope you will take that in the spirit of moving this to a resolution for those people who find themselves asking the same question as the OP.

Again, I think most musicians are no computer geeks

If that's true, and I don't know that it is in this day and age, that's why forums like this should be able to 'de-geekify' the process and, along the way, provide a few basic do's and dont's.

"low latency" needs to be specified if we want to discuss this.

Ha ha... I can feel the goalposts moving, my friend.
OK... As good as, or better than hardware. I had hardware samplers from Roland which used to take 40 ms to process the last of the available 24 notes. I can beat that. :_)
Many high end users are running octo-cores with buffers of 32 samples... all the way through to completed mix. It's a fact, not a theory or principle. I think we might able to leave those goalposts where they are. Low latency is fact, not a principle.

My point is finally that if a non tech geek buys a pc and installs any live host software and some plugins there is no automatic and perfect control that avoids cracks and pops - the traditional way by reducing polyphny in a more or less clever way - keeping audio out a foolproof thing

.

That's different from saying it can't be done... isn't it?
I'm gonna assume that any musician who buys a laptop for live use isn't an idiot. Uninformed? Maybe. Able to learn? He'd better be or it wouldn't matter if it was hardware or not. I'm not as concerned about an individual's capacity to comprehend as I am about the simple, demonstrable fact of running nearly anything from an i3 to an i7 and getting great results up to limits which each individual should test, and even then, only if you intend to push it to the limit as Tony and others achieving the dream are doing.

So.... in light of the above... are there PCs in 2013.... which can run well at the generally accepted standards of low latency? (512 is high latency IMO)
A resounding YES from this corner - and it can be done less expensively than at any time in computing history.

[TiUser]

Hey, I never said things can not work practically, never.

Well, there is a lot that can be done - but isn't for reasons that might make another threat.

Using hex core, octo core or multi cpu server pc's isn't what I understand by a mainstream pc and I am not sure if there are any notebooks out with such specs. ok, Let's use a Cray and all cpu hassle is gone...


I still think using Windows OS for realtime audio is a waste of cpu power concerning what you get - for principle reasons. Even if you have a 100 core system this does not change from principle - also it might get practically neglegible.

But ok, if wasting cpu gets so cheap that it doesn't matter any more I will switch to your opinion - even if the principles underneath do still not change - unless MS does change things. Maybe MS does this when machines are much faster. Scheduling context switch of 0.1ms instead of > 10ms would make programming MUCH easier and relaxing - even without resource control - because using cpu efficiently would be much easier.

I assume the sampler HW you refer to is quite old, isn't it? I belive that in the old days also HW audio devices were not perfect. In the old days and still today transportting a single note on midi command via traditional midi ports takes about 1ms... and in the old days you didn't see any standard pc being capable of something better than doing some video game noises a la SID...

[pinkcanaru]

Hey, I never said things can not work practically, never.

I think that you and I are a couple of real pedantic guys. Ha ha.
TiUser... I think when you say that no PC can be used at low latency, any casual reader, the kind of 'non- geek' that you refer to... would read that as saying, it can't be done to a standard I can rely on. Because you have said, clearly and unequivocally, it can't be done. If, as you wrote, one can easily outperform ANY PC at low latency then where does your acknowledgement of the practical enter the conversation? I know this appears to be going in circles but that is only a symptom of trying to reconcile your comments.

Using hex core, octo core or multi cpu server pc's isn't what I understand by a mainstream pc

You are, quite simply, wrong. Octo cores are common in professional facilities.
Quad cores are common EVERYWHERE.

and I am not sure if there are any notebooks out with such specs

.
There are laptops that can do exactly what you say cannot be done. Please tell me again that my i5 can't play at low latency. LOL.

ok, Let's use a Cray and all cpu hassle is gone...

I think this might be the source of the issue. There is too much hyperbole. In fact, you're saying it again... 'It can't be done without a supercomputer'.
Well, by the standards of computers just 5 years old, the current generation may well look and feel like supercomputers.

I still think using Windows OS for realtime audio is a waste of cpu power

Yes.... and so what? If the power is there then your position becomes a non-issue. All I want to do is play music in real time on my laptop without clicks and pops. Can I do that? Yes. Can Tony? Yes. Can any person with an i5 and upwards get remarkable performance if they take the time to click a few buttons on the internet and find out how? Yes. ... Or use a company that builds audio PCs and let them take the worry out of it. OK, it may cost a little money but someone's expertise is worth paying for if you are one of those 'non-geeks'.
(FWIW, I had PC Audiolabs build my main studio PC - not ashamed to say it.)

But ok, if wasting cpu gets so cheap that it doesn't matter any more I will switch to your opinion

I know one cannot discount the economics, but the fact is, that day is here. The question posted is not 'can I afford a low latency system?' The question is, 'who is using Cantabile live and no hardware?
You're arguing a different point. If you ain't got the dough, you ain't got the dough - this forum can't solve that... but, FWIW, an HP i5 can be had for $700. A superb i7 with 8 gig of ram can be had for $675!! Is this not cheap? Even when you factor in the Euro cost, the cost has come DOWN. You can't say that it's prohibitively expensive.

- even if the principles underneath do still not change - unless MS does change things. Maybe MS does this when machines are much faster. Scheduling context switch of 0.1ms instead of > 10ms would make programming MUCH easier and relaxing - even without resource control - because using cpu efficiently would be much easier.

You're still arguing economics. You're essentially saying, "I have a computer that should be capable of doing all this stuff if only MS wasn't in the way - and I resent being forced buy another computer to compensate for the inefficiency. TiUser.... that's not the question that was asked.

I assume the sampler HW you refer to is quite old, isn't it?

Yep, and thanks for making my point even clearer. A 1990 Roland sampler with 16 meg of ram, and which cost 5.5k British pounds and was junk by today's standards. Even better than that was my 1984 purchase of an EMUII, with 17 seconds of 8 bit sampling. That ran about 10k British pounds. Accounting for inflation, you're probably looking at close to $20k at today's prices for an instrument that could not play in time if it received more than 3 midi notes. LOL! But - if you're doing the work, and you need the contemporary gear of the day, it makes no economic sense NOT to buy that stuff.

I belive that in the old days also HW audio devices were not perfect.

Ha! See above. And they ain't perfect now either. As Tony said, that's a myth.

In the old days and still today transportting a single note on midi command via traditional midi ports takes about 1ms

Yes. At least today the playback can be sample accurate. But for live players, that's still an issue, notwithstanding that running status means that you can potentially play 3 notes and only take 1ms. The best thing is, playing live you can't really play that many LIVE notes.

... and in the old days you didn't see any standard pc being capable of something better than doing some video game noises a la SID...

And the old days come a lot quicker these days.
That's why it is safe to say that low latency, reliable performance, to a standard which retires hardware, is a reality in 2013, if one is prepared to make the relatively modest investment. Certainly WAY less than hardware.
Peace.

[TiUser]

To be pedantic isn't a bad thing - it's the source of discussion and progress.

You seem to get me wrong.

You say current machines have enough cpu for what you are doing. You have also quoted the plugins you're using. That's important practical information. I am not about to discuss that.

My statements about windows, scheduling and "reliability" are based on principles. Unless you present me with different technical facts I stick with my current opinion. If all this is a complete non issue why then a thread about how to control cpu spikes... obviously there is something you wish to understand, control and better.

I am not about "can it be done" when I criticize a system behavior. I just point to things that are not under control, are "best guess" and inefficient. If a user is experienced just raises the bar but does not change what is going on behind.

Look at DPC. We know it shall be low. But if it is not it's out of control to change it. It's a matter of hope if the related drivers will ever be updated. Wouldn't it be better to have a system where this problem did not exisit because of different system design?

All that does not mean that certain configurations can not work. It only means there is a potential risk. It also means one can spin in circles to get working what he wants.

[pinkcanaru]

To be pedantic isn't a bad thing - it's the source of discussion and progress.

Great I'm all for it LOL.

You say current machines have enough cpu for what you are doing. You have also quoted the plugins you're using. That's important practical information. I am not about to discuss that.

OK - but it's the only thing that means anything when you consider the question of the OP. I don't see how there can be any debate about that.

My statements about windows, scheduling and "reliability" are based on principles.

Yes, and I believe we have moved way beyond that. Principles mean nothing to someone who wants to walk on stage and have it work. Dare I say that is actually way more 'geeky' than required for the question. The simple non geeky answer is that anyone (especially your non-geek muso) who wants to run a serious system live should use the fastest computer affordable and test it, hopefully after having taken some advice here first. We can provide real world examples of configurations that work. If there were a poll taken by 'non-geeks'as to which information is the most useful, principle or practical, I bet I know which would win.

Unless you present me with different technical facts I stick with my current opinion.

Then your opinion is that real world results, based on tried and tested combinations of hardware and software, mean less than a discussion about theoretical problems. Once again, I don't think that anyone who is truly interested in getting up and running is going to take much away from that. What the OP wanted to know is 'who is DOING it'. Not who is 'THEORISING' about it.

If all this is a complete non issue why then a thread about how to control cpu spikes... obviously there is something you wish to understand, control and better.

You are conflating two issues which are tangentially linked but not the same subject. I can prove that the sheer power of a an i5 compared to a core duo relegates peaks to a level which cannot harm what I need to produce live.
However, I am very interested in how the 'principles' you are drawing attention to impact an OS - and so I purposely started that spikes topic to try and see if other users here had information, opinions and experiences that may benefit all us Cantabile users. Maybe we can ascertain whether some plugins can induce peaks. There are certainly 'bad' plugins out there that consume more than they produce. Peaks which have little to do with OS calls per se, but are caused by badly written code. Maybe we can clue in on some other processes which can be stopped and get a few cycles back. That, certainly, would be of interest to those who are looking to get a bit more life out of older hardware.

I am not about "can it be done" when I criticize a system behavior.

Respectfully, I can't work out why that belongs in a 'can it be done' thread. If you disagree, can you tell me why it does?

I just point to things that are not under control, are "best guess" and inefficient.

Which we have determined is irrelevant to this topic. When we determine safe thresholds we can be assured that we are going to achieve the required results, even in the presence of peaks.

If a user is experienced just raises the bar but does not change what is going on behind.

Only raises the eyebrows of geeks like you and me. I am all for 'de-geeky- fying/'

All that does not mean that certain configurations can not work. It only means there is a potential risk. It also means one can spin in circles to get working what he wants.

But it becomes a distraction for non-geeks who just wanna know the answer to the question.
I think we now have a geek thread going. I'll meet you there for more 'principles'.
Cheers for the lively debate.

[TiUser]

I respect your standpoint - you know I'm not about flame.

We are pointing to different things and I admit you can look at my point as "geeky". What remains to me is that audio on windows is relatively inefficient and does by far not reveal all the cpu power we have paid for. I think it is important to know this regardless how you conclude and decide for yourself.

Maybe it's irrelevant to the OP's question but I think economical aspects are not irrelevant too. I remember well the times when I was young and money was an issue. Making an uninformed decision and surprises can turn out into frustration, even desaster. Today I think getting the fastest and latest is not really economical too - unless you absolutely need it for commercial reasons.

I am not in a counter position to your statements, I just tried to add more information as I always try to, in the spirit of enabling people to make decisions with better background.