10 Jan
2012
10 Jan
'12
1:07 a.m.
Hi Moshe,
The same here Ubuntu 10.04 with kernel 2.6.39 manually compiled, with
THREADIRQS enabled (which is the most important part of the RT kernel,
according to what I've learned here and read over there in the grid)
I can get 5.8 ms latency without xruns BUT having the WIFI disabled (no
IRQs are shared but it still cause a lot of xruns....)
I'm not a kernel expert, but I can say that you don't need to use a RT
Kernel in order to have a low latency. [1]
For instance, if you decide to give a try with Ubuntu, you can use the
default kernel and configure jack to use RT scheduling [2].
If that kernel is not suitable for you (I mean, you get a lot of xruns),
then you can use the PREEMPT or RT Kernels that are included in the
Official Ubuntu Repositories (which are configured in the apt sources
automatically) and see how it works before going to a manual kernel
compillation.
[1] http://jackaudio.org/realtime_vs_realtime_kernel
[2] http://jackaudio.org/linux_rt_config
Good luck and have fun.
2012/1/9 Moshe Werner <moshwe(a)gmail.com>
Wow!
Thank you for your knowledgeable inputs!
It indeed seems that the openSuse distro is going into the direction of
usual office use, too bad cause I already got accustomed to it.
Will look into the options you listed. (Kxstudio probably try it out with
Arch, or Tango studio)
Something little OT, but what exactly is the difference between
Windows/Mac, where one doesn't need an rt kernel to run realtime processes,
and Linux which needs one?
Please excuse my ignorance for as I said I'm more the audio guy than
computer guy.
Really appreciate your opinions!
Moshe
On Mon, Jan 9, 2012 at 11:06 PM, Jeremy Jongepier <jeremy(a)autostatic.com>wrote;wrote:
On 01/09/2012 03:24 PM, Moshe Werner wrote:
_______________________________________________
Linux-audio-user mailing list
Linux-audio-user(a)lists.linuxaudio.org
http://lists.linuxaudio.org/listinfo/linux-audio-user
Hi everyone,
after many years of studio work using the openSuse distro with the
kernel-rt from Jan Engelhard it seems that he no longer continues his
great
work on rt kernels.
Being more on the recording engineer side of things and not a Linux
expert
(user yes, expert no) I really fret at the thought of patching and
compiling my own kernel package.
I would like to hear your opinions on what distro is solid for audio work
and has a reliable rt kernel.
Also I would appreciate if you could explain the degree of difficulty and
learning curve of the specific distro.
My system:
Intel i7 950
Gigabyte motherboard
6 Gb ram
Rme HDSP 9652 audio interface
Appreciate your answers.
Moshe
P.S. I tried to use Ubuntu on the same machine I use openSuse 11.2 on and
got pretty bad results regarding latency and x runs on jack 2.
Hello Moshe,
Ubuntu 10.04 here with the Tango Studio real-time kernel. Latency-wise I
can go as low as the soundcard allows. If I have any xruns that shouldn't
be there I search just as long until it's solved. I still prefer Ubuntu
because it is one of the biggest distro's and that comes with some
advantages. I'm also very fond of the PPA system (Personal Package
Archives) and that Ubuntu is akin of Debian. I've used Fedora and Mandriva
(both RPM based) for a long time but prefer the Debian way for a lot of
things (packaging, filesystem layout). Also Ubuntu LTS releases just work,
at least, in my experience. And they're stable, especially after the first
point release (10.04 is at 10.04.3 now).
I'm also dabbling with Arch at the moment. I like it, it was a lot like
coming home but Arch also has some major drawbacks. The packaging system is
a huge security flaw, especially when you use AUR. Anyone can upload
anything (this is possible with Ubuntu PPA's too but it's a lot harder).
Other than that AUR is simply amazing, compared to Debian/Ubuntu packaging
is a breeze. Other thing are the rough edges. Ubuntu is polished,
especially when it comes to the desktop experience. I've come to appreciate
that through the years so I had a rough time getting font rendering right
for instance, and it still doesn't look and feel like on my Ubuntu install.
But Arch does have its pros. It's a rolling release so you only have to
install it once and then you're good to go for years to come. This will
also ensure that you're always running a pretty up to date system. Which
could also be a disadvantage. I'm a Jack1 user for instance but also a
seq24 user and seq24 doesn't work with Jack transport with versions >
0.118.x. So 0.121.x that is in Arch at the moment doesn't fly for me. The
biggest pros for me are the configurability and that everything is so well
documented. I LOVE opening a terminal on Arch and configure stuff that way
because it's so easy and fun.
I'm not very fond of specialized multimedia/audio distro's. I want to
configure a system the way I want, most of the distro's do things a
different way or just wrong in my opinion. Also most of these distro's are
driven by incredibly small communities or simply just one person.
Continuity is not assured. I do check them every now and then and cherry
pick the good stuff and integrate it on my own system. If I'd have to
choose a multimedia distro though I would most certainly choose AVLinux,
closely followed by Tango Studio. GMaq and Jof are simply very knowledgable
guys and listen well to what users have to say.
In your case I think Arch might be a bit too much expert. If you're
coming from OpenSuSE you might want to try Fedora with the CCRMA repo or
give Ubuntu another try, it is a Linux flavour after all so Jack should be
able to run with acceptable latencies. Or stick with OpenSuSE and hope
someone is willing to take over maintaining a real-time kernel.
Best,
Jeremy
______________________________**_________________
Linux-audio-user mailing list
Linux-audio-user@lists.**linuxaudio.org<Linux-audio-user@lists.linuxaudio.org>
http://lists.linuxaudio.org/**listinfo/linux-audio-user<http://lists.lin…