There's so many of these conversations that revolve around dual purpose setups.
Work and domestic.
I've split the two, with a work box, and the laptop for domestic stuff. As soon as i did that, most of the problems i had initially using linux/Ubuntu disappeared. (Plus research, 4 million user errors, learn some, make more mistakes, ask patient and talented people who knew about this stuff when i didn't, etc..)
Running a red hot audio box is like owning a ferrari. It needs setting up properly, for a specific purpose. Most of us trying to make a modest living with this stuff don't have the luxury of full time engineers and developers tweaking and maintaining our systems. (Zimmerman, Howard, Frizzell, etc....)
I reckon trying to run a dedicated audio box, with domestic use as well, is like a disturbing hybrid of ferrari and ford, one getting in the way of the other at times. (The Jack v Pulse paradigm springs to mind here.)
I used to have 5 gig boxes, going into a win box running cubase and sibelius. I had no end of challenges, and some humilitating moments in front of punters wanting to hear their masterpiece in all it's glory, only to see the system throw the teddies out of the pram without warning. It was a bloody nightmare, and when i did the time/motion bit, i was spending more time fixing, than writing. And that was a dedicated setup with no domestic stuff at all.
When i first got into linux, i resolved to learn as much as i could, as the idea of adding and subtracting apps specific to purpose appealed to me. I did the dual role bit, and discovered conflicts of interest in a dual purpose box, but when i divided the tasks, well, the light came on, and life got a lot simpler, day in day out. My old G4 laptop, once i dumped mac, and put linux on it (even better now with a shiny new Debian 5 install), does a good job of handling domestic, and the audio rig is as clean as my current knowledge level will allow. I have UBStudio Hardy 32 and 64 bit boots, soon to be pure single boot 64bit (Still deciding on UBStudio or Debian.) It's stable and works, every day. 5ms average latency dependent on task, and over 80 tracks in Ardour, with the same in RG, and over 250 tracks in Linuxsampler, which frankly, imho, completely destroys gigastudio for performance. (1 LS instance =at least 5 gig boxes, in my testing so far, and it's stable versus that nervous 'giga' state that breeds grey hairs and bad habits.)
Gutsy was even better for performance, imho, but the devs tempted me with new stuff that i wanted in my workflow. (RG's brilliant free segment positioning per track, for just one example. Great addition that reduces donkey work by a sizable chunk. Or the formidable progress being made in Denemo. And then there's inline midi editing in Ardour to look forward to...)
I'd agree that 8.0.4 is slightly less friendly than Gutsy, and takes a little more to setup, but if the task is audio only, it goes fine. (at least here.)
I wonder if we're asking too much sometimes, expecting a single computer to handle ferrari audio, and ford domestic, all in one place. so i think the discussion gets clouded sometimes, often with personal experiences based on hybrid use and expectation, that will sit uneasily at times in any OS.
Domestic is a ford, and most linux distros do that well. (Allowing for the big kernel changes we've been getting lately, and the activity required of distro builders to keep up with this.)
Specific audio/graphic use is a Ferrari that needs time, knowledge, and patience, to setup for great, stable, performance. And i've been delighted to discover the speed and performance boost i get when i work from terminal as much as possible. The more i do this, the more stable, well behaved, and strong in performance, the box seems to be.
Either domestic or dedicated purpose is an interesting discussion to have, but together may be an exercise in smoke and mirrors that clouds the real challenges, and confuses user expectations.
2 roubles worth,
Alex.
p.s. Jconv is an excellent terminal app that performs really well. Have i mentioned that? Nothing wrong with working from terminal with scripts, at all.
:)
On Mon, Feb 23, 2009 at 7:34 PM, Michal Seta
<mis@artengine.ca> wrote:
Hi all.
I have been following this thread and meant to add my own .02 units of
time and neural activity but was a bit swamped with work. I take this
opportunity because since Windblows is being dragged into the debate,
it's getting even more fun.
Disclaimer: I have never tried to use Windblows for pro audio.
I have been in recording studios that relied on windows and I have
seen frequent crashes and issues with audio latency, compatibility,
driver installation and reinstallation and stuff like that. Just last
night I went to see a friend's demo of a VJ system he's developing
which was delayed because of some video driver setting that was not
compatible with the projector. It took 2 hours of fiddling with XP to
get the demo going (including intricate tweaks and several reboots).
Far from user-friendly if you ask me. I have seen countless blue
screens of death in performances, art installations, studios, demo
kiosks, publicity panels etc.
The only time I used Windows for an installation where performance and
stability was an issue it had to be stripped down to a minimum,
offered a lot of RAM, etc. So basically we had a bare Windows XP
Professional system running only the components that were being used.
In that situation we have actually achieved the performance and
stability we required (but the machine still needs rebooting from time
to time in order to ensure perfect functioning). Since the computer
was dedicated to that installation it was not a big issue but it took
use quite a bit of effort to actually make it perform to our spec.
The only people that I have known to have stable and reliable Windows
system for audio and/or video were people who knew Windows inside-out
and performed the necessary tweaks (which often were very intricate)
to ensure the proper functioning. Also, having a dedicated windows
machine to do audio or video is a must. That's a waste of resources
if you ask me.
However, if you are referring to the fact that in windows you do not
need to hunt for video codecs or perform separate flash installation
to watch youtube, then feel free to disregard the above rant.
Now, on to Linux. I have been using Ubuntu for the past couple of
years and I have been generally happy with it. I was running 7.10
until around November or December last year. All was fine, including
-rt kernel and low latency performance. I moved on to 8.04 and
that's when my problems started. Lots of problems with -rt kernel.
The vanilla kernel unusable for realtime audio work. I upgraded to
8.10. Bunch of other problems (bluetooth stopped working, file
sharing with XP stopped working, bunch of other annoying little
problems) greeted me every day. Some of them have been corrected and
I am slowly getting back to a usable system. The -rt kernel (again
from ubuntustidio, IIRC) was total mess. I decided to give vanilla a
run. I am now running a kernel identified as 2.6.27-11-generic #1
SMP. I run jack with 11.6 ms latency, this was the best setting I
could come up with in 44.1 kHz. It's good enough for me right now. I
get an occasional xrun but no more that 3-4 a day and most of the time
they happen when loading or closing applications (or surfing the net
with firefox). I have been able to run Pd, process multichannel audio
(2-4), and simultaneously record 4-6 tracks into ardour. no glitches.
All this on a CoreDuo 1.8G laptop. I do not do any overdubs so I
don't know if my setup would be sufficient for a typical studio use.
I seem to be the only other user (besides Jostein Chr. Andersen) who
is satisfied with Ubuntu (vanilla kernel in my case!) as an audio
platform. I am not satisfied with it as a desktop, however, it is
beginning to feel like windows, but I am biased (hey, I have been
using linux since 1997 and developed nasty habits such as typing
commands into the terminal window and editing configuration files by
hand. I still find it weird to configure stuff with a GUI, sometimes
even annoying). I run xfce or awesome anyways so I don't get to see
most of the GNOME bloat.
So, why is it that vanilla kernel works for me and nobody else? Have
I done something wrong?
./MiS
On Mon, Feb 23, 2009 at 11:25 AM, mikk <
michiel33@gmail.com> wrote:
> Who can explain to me why it seems to be so hard to get a stable audio and
> video environment in Linux (not requiring recompilations and intricate
> tweaks) while this problem seems to have been solved quite satisfactorily in
> Windows? What is the fundamental reason for this?
>
> Michiel
>
--
./MiS
514-344-0726
http://www.creazone.ca
--
Parchment Studios (It started as a joke...)