Do I really need an RT(Real Time) kernel

[l0wt3ch]

You have already stated out clearly what you think about non native English speakers and that imply your interest in a international conversation. Well, I know, that those statements from you are already deleted by the forum moderation. But you sign for it it in this post as well. http://www.linuxmusicians.com/viewtopic ... =60#p27424

Are you really that mad because I used the word "[sic]" in one of your quotes? I didn't know you were going to be all butt-hurt about it. For the record, I have plenty of interest in international conversation.

But this thread isn't about me, brummer. It's about realtime kernels, and you still haven't answered why you've held out for so long on this subject. Why don't you spell it out for all the new users out there, and tell us why you insist on only using RT kernels?

[i2productions]

I think we are all now talking in circles. Are there any more arguments for or against? If not, than maybe we should drop this before it gets out of hand again.

[l0wt3ch]

I, for one, would like to hear a response from brummer. (Which is why I asked the question...)

You're certainly welcome to exit the discussion, however.

[brummer]

Are you really that mad because I used the word "[sic]" in one of your quotes? I didn't know you were going to be all butt-hurt about it. For the record, I have plenty of interest in international conversation.

First, it is a sign of disrespect, as I am used to from you, to edit quotes !

But this thread isn't about me, brummer. It's about realtime kernels, and you still haven't answered why you've held out for so long on this subject. Why don't you spell it out for all the new users out there, and tell us why you insist on only using RT kernels?

Second, it seems you didn't understand me, I wrote clearly what I mean. I only jump in this discussion, because you try to imply wrong conclusions. I use rt-kernels only, but, that didn't mean that rt-kernels must be the best choice for every one who like to produce audio on linux.
The kernels from the 3.) generation suite very well the needs of pro audio users, also without a recent rt-patch.
Thanks to the rt-kernel team that have managed to plug the rt patch more and more in to the mainline kernel.

(edit: minor censorship by raboof)

[l0wt3ch]

I only jump in this discussion, because you try to imply wrong conclusions. I use rt-kernels only, but, that didn't mean that rt-kernels must be the best choice for every one who like to produce audio on linux.

So you were content to have new users believe the kernel you yourself use is unnecessary, possibly even worse for pro audio work (according to i2productions)?

Ok then!

[brummer]

However, AFAIK their target audience isn't really the pro audio crowd. I think they actually are doing it for controling laser beams, other machinery and stock exchanges.

Right, the current maintainer of the rt-kernel team is Thomas Gleixner, he is also the Manager of the Linutronix company, witch provide linux based solutions for industries.
So latest development of the rt-patch have more the needs from industrial devices in mind, but that didn't mean that linux pro audio don't benefit from it any more.

That was my first comment in this thread.
I jump in to let users know that you, lowtech, are wrong.
And other then you, I'm not meaning that it is the best for all, just because it is the best for me.
And other then you, I'm not make users believe they can't get a rt-kernel from there favourite distribution.
Also other the you, I didn't recommend the distribution I use to everyone. I know very well that others may be suite better with a other choice. That, it seems, is a point you need to learn, before you can become a useful member of a community.

[i2productions]

So you were content to have new users believe the kernel you yourself use is unnecessary, possibly even worse for pro audio work (according to i2productions)?

I think it's especially deterimental to new linux users(which I'm guessing most that would buy an audio distro are.) They are going to drive their latency as low as it can go without understanding the possible consequences.(I.E. Xruns)

My take on it is that "low as possible" means it will be "low enough" for more people.

And can potentially be too low for your hardware. And for the new linux user, they're going to run into those wonderful clicks and pops in their audio when recording a live guitar for instances, and not understand why when using a "real-time" kernel, they're not getting a good recording!

Not to mention when the new user unknowingly downloads a driver for their Nvidia graphics card, it screws things up. I myself use a completely generic kernel for maximum compatibility with all the hardware I want to run, and I maintain 11ms of latency with few xruns.(As an example I worked for 4 hours straight yesterday with Ardour3+Reason 5.0[WINDOWS SOFTWARE IN THE MIX EVEN!] and in the entire thing only registered 2 Xruns, and they only appeared when I first opened Reason.) But the low-latency kernel isn't bad either(as long as you truely understand latency, and how your hardware effects it.) I'm not against the RT kernel in general. But I am against it being included in an audio distro with a broad user base that has diverse needs that the RT kernel doesn't help.

[GMaq]

Latency needs to 'low enough' not as 'low as possible'. Someone happily working at 10ms reliably will not care if 2ms is theoretically possible in a real world situation.

My take on it is that "low as possible" means it will be "low enough" for more people. You don't have to use all of its capabilities (it comes with the default JACK settings!), but it's there if you want/need it.

lowt3ch you have a good product there no doubt but milking the -RT kernel reputation that hard with it's very modest latency gains vs. low-latency really comes across as a bit overwrought and certainly latency is only one of many factors potential users will be interested in. Your quote from Mr. Gleixner could really go both ways as he also concedes that IRQ-threading is a substantial improvement over the stock kernel.

GMaq, Studio 13.37 doesn't have modest latency gains. It flies! It is built on a micro-Linux that is designed to run in RAM (the whole OS), comes with a super-light desktop environment, and, yes, a realtime kernel. The whole filesystem is less than 400 MB. It's the innovative combination of different elements that make it what it is.

As for Gleixner's quote, I've never denied that IRQ-threading was an improvement. However, I've consistently maintained that anything less than an RT kernel was second-rate. And, if I remember correctly, wasn't Ubuntu Studio dropping RT-kernel support one of the reasons you originally started AVLinux?

lowt3ch,

I think we will both agree latency undoubtedly is an important part of the equation and that anyone doing what we do is striving to provide a stable reduced-latency environment. What I meant was regardless of whatever benchmarks you achieve with LEET that if you used a low-latency or RT kernel in LEET the milliseconds of difference with all else being equal probably would not reduce it's unquestioned effectiveness as a fast light system significantly enough to sway a potential user one way or the other, so lets just agree to disagree on whatever milliseconds remain

Personally I am neither 'pro-rt' nor 'anti-rt' the current stable branch Kernels we use are giving us happy users and good performance so whatevs. Obviously the landscape back in 2007 when I hung out my shingle has changed immensely and if in a year's time -RT is the most promising ticket you can bet I'll be looking into it. Everything in Linux changes much too rapidly and often to be too rigidly set in our ways...

[l0wt3ch]

lets just agree to disagree on whatever milliseconds remain

Hey, those are important milliseconds!

[i2productions]

Hey, those are important milliseconds!

Not if 1 of them is an Xrun!