Ok, some informal tests yielded the following:

* -nosleep causes lots of drop outs and actually locks up X windows, I needed to ssh in order to kill pd
* --rt audiobuf 6 -blocksize 64 are about as low as I could go without dropouts ... but really, *no* dropouts & very smooth. I'm happy!

On Jun 6, 2014, at 4:50 PM, Dan Wilcox <danomatika@gmail.com> wrote:

I just setup my UDOO Quad with the Debian hf image, the latest Pd vanilla, and a number of externals.

For realtime, you should do the following to add rt audio support:

sudo su -c 'echo @audio - rtprio 99 >> /etc/security/limits.conf'
sudo su -c 'echo @audio - memlock 250000 >> /etc/security/limits.conf'
sudo su -c 'echo @audio - nice -10 >> /etc/security/limits.conf'

This should get the -rt pd flag working. If you start pd with "pd -rt", you should now see:

    priority 96 scheduling enabled.
    priority 98 scheduling enabled.

These are my pd start options ("Unit-Control" is my main audio & playlist patch):

    pd -nogui -rt -alsa -audiodev 5 -channels 2 ../../Unit-Control.pd &

I also renice pd to -10 in my run scripts to have the scheduler give it some more cpu.

    renice -10 $(pidof pd)
 
With multiple cores, the system runs pretty great with a blocksize of 64 and a 10ms delay. I'll do some tests and check in later.

On Jun 6, 2014, at 5:28 AM, Pierre Massat <pimassat@gmail.com> wrote:

Hi,
I wonder whether this relatively poor performance has something to do with the fact that this soundcard is powered through USB. I know that this is a big issue on the RPi, so perhaps should get a powered USB hub and give it another try.
Cheers,

Pierre.


2014-06-06 11:18 GMT+02:00 Alexandros Drymonitis via Pd-list <pd-list@lists.iem.at>:
Coming back to this thread after some time, but thought it's better than starting a new thread.
I'm testing Pd-0.43-2 (from Debian's repos) on the Debian hard floating point image provided by Udoo's website, with a Focusrite Scarlett 2i4 sound card (class compliant). I've been doing some tests to see what's the optimal performance I can get, and for now I'm getting no drop outs with a 44.1kHz sampling rate, 30ms delay and 128 block size.
I tried to follow this wiki page to set up low latency, but it's a bit over my head. Since ALSA is already built, I started from here, but I can only understand the first step, below that I'm at loss. Has anyone tried to set this up, has anyone had better results (higher sampling rates, smaller delay)?


On Sun, Apr 6, 2014 at 1:06 AM, Dan Wilcox <danomatika@gmail.com> wrote:
Ah cool thanks! This is actually the same issue I saw when I briefly tried the hard float image. I'll pick up another sd card and try again ...

On Mar 24, 2014, at 5:01 PM, Simon Iten <itensimon@gmail.com> wrote:

nevermind, found the solution. here it is..

http://lists.puredata.info/pipermail/pd-list/2012-09/097892.html

cheers


On Mon, Mar 24, 2014 at 2:43 PM, Simon Iten <itensimon@gmail.com> wrote:
hey dan and list,

i gave pd on the debian hardfloat image a try and can not run it. it compiles just fine. this is what i get on the console...

debian@udoo-debian-hfp:~$ pd -verbose
Pd-0.45.4 ("") compiled 14:17:43 Mar 24 2014
port 5400
TCL_LIBRARY="/usr/local/lib/pd/lib/tcl/library" TK_LIBRARY="/usr/local/lib/pd/lib/tk/library"   wish "/usr/local/lib/pd/tcl//pd-gui.tcl" 5400
Waiting for connection request...
/usr/local/lib/pd/bin/pd-watchdog
watchdog: signaling pd...
watchdog: signaling pd...
watchdog: signaling pd...
watchdog: signaling pd...
watchdog: signaling pd...



then a watchdog-signaling loop...

this also happens with the version installed via synaptic.

any thoughts?

cheers


On Sat, Mar 15, 2014 at 4:50 PM, Dan Wilcox <danomatika@gmail.com> wrote:
Yeah, I wanted to use the hard float image but I was under time pressure and more things seemed to work out of the box with the Linaro one. I'll have more time to revisit it later.

On Mar 15, 2014, at 12:46 PM, Simon Iten <itensimon@gmail.com> wrote:

hmm,

well to me 12ms is way to much. but then again i play a lot of fast attack notes in up-tempo pieces :-)

thanks for your notes anyway, they helped a lot! and write back when you tried with the debian hardfloat image. i tried it for a short time and it was not very stable with pd. but then again i did not try a lot of things to fix this.

cheers
On 15 Mar 2014, at 13:10, Dan Wilcox <danomatika@gmail.com> wrote:


I was wrong, the guitar to amp latency at 1 meter away is roughly 3 ms.

The accumulation of a monitors and an effect or two gets you to 8ms. Acceptable latency is 12 ms.

Again, I haven't measured my rig or the latency of my old wearable rig, both both were responsive to me, so they must e at least around 12 ms.

Sorry for being unscientific about it.

enohp ym morf tnes
--------------
Dan Wilcox

On Mar 15, 2014, at 5:49 AM, Simon Iten <itensimon@gmail.com> wrote:

dan, no 15 ms is in no way tolerable for live use (if you have effects that should react in realtime) it is of course ok for delay and reverb stuff. the latency from an amp because of cable length and stuff is totally different, since your ear actually hears where the sound comes from and can adapt.
but for studio use for example, 15 ms on a headphone is really two attacks for evey attack. heck even 10ms is evil :-) of course your/anyones mileage may vary. but i only wanted the box to output delay and reverb soundscape stuff away, so i might be good. i will add analog circuitry that mixes the dry and effect part of the signal, so i get no (very very little) latency on the unaffected part of the signal.

no worries as far as the script goes. i had no problems at all to follow it. but i worked with linux a lot before. i was just suggesting, that the typical ubuntu user would not get some of the steps in between your steps :-)


On 15 Mar 2014, at 02:10, Dan Wilcox <danomatika@gmail.com> wrote:

I haven't run any latency tests, so that might be what I'm getting. If so, it's acceptable for what I do. From what I've read, guitar -> effects -> amp latencies are already closer to 20ms.

Sorry I haven't gotten back to the UDOO and pulled the relevant scripts etc off of it yet. I'm trying to get a few things done before I head out of town for work the next 2 weeks. I might be abel to get to it Sunday, but no promises.

On Mar 14, 2014, at 8:41 PM, Simon Iten <itensimon@gmail.com> wrote:

hi dan,

tried your setup/instructions. thanks, it now works down to 15ms. at 12ms i start to get clicks here and there…

your script has some “errors” (missing instructions a novice would not understand how to deal with). do you want me to post them, or do you overdo it anyway?

thanks again

simon

On 13 Mar 2014, at 05:21, Dan Wilcox <danomatika@gmail.com> wrote:

Thanks. I was just waiting to redo my website, edit the video, put the pics together, etc etc but life and freelance work get in the way. Man, I could use a clone right about now :P
 
On Mar 13, 2014, at 12:19 AM, Richie Cyngler <glitchpop@gmail.com> wrote:

Also interested in the UDOO setup instructions so thank you. A bit OT but, Dan, love your work (that "onward to mars patch" is awesome) thanks for the links. I think people should post more of this sort of thing to the list, celebrate what we make. =)


On Thu, Mar 13, 2014 at 11:09 AM, Dan Wilcox <danomatika@gmail.com> wrote:
FWIW: here's a picture of my UDOO setup inside my Mars space suit backpack: http://www.flickr.com/photos/danomatika/13115604285/

Media of the backpack in use https://twitter.com/danomatika/status/433273394122207232/photo/1 & https://vimeo.com/86670103 (not my video, I'll put out a different edit soon)

On Mar 12, 2014, at 10:28 AM, Dan Wilcox <danomatika@gmail.com> wrote:

I will do that later tonight when I boot the udoo and pull my run scripts off of it. I'll post everything to GitHub so we can share resources.

On Mar 12, 2014, at 5:44 AM, Jamie Bullock <jamie@jamiebullock.com> wrote:


Hi Dan,

Thanks for sharing these notes. They arrived in my inbox to coincide nicely with the delivery of my quad Udoo this morning! It would be great to see a full writeup of your Udoo setup at some point as I think many people will want to be doing a similar thing.

All best,

Jamie


On 11 Mar 2014, at 14:14, Dan Wilcox <danomatika@gmail.com> wrote:

Heres a trim of my notes:

Enable realtime audio priority (if you haven't done it already):

sudo su -c 'echo @audio - rtprio 99 >> /etc/security/limits.conf'
  sudo su -c 'echo @audio - memlock 250000 >> /etc/security/limits.conf'
  sudo su -c 'echo @audio - nice -10 >> /etc/security/limits.conf'

I disable pulseaudio. Make sure pulseaudio does not respawn itself (from http://voices.canonical.com/david.henningsson/2012/07/13/top-five-wrong-ways-to-fix-your-audio):

    echo autospawn=no > ~/.pulse/client.conf

Also add the following to ~/.bash_login to kill pulse audio if it's running on login:

    # kill pulse audio if it was spawned
    pulseaudio -k

I'm not looking at the udoo run script, but I'm pretty sure I'm using the following with the US-25EX USB soudcard:

pd -rt -nogui -alsa -audiodev 5

Use pd -listdev to get the device list from alsa. I chose 5 as the first 4 (from memory) are 1-2 (built in hardware & plugin) & 2-3 (HDMI audio hardware & plugin). 5 is the USB hardware alsa dev.

On Mar 11, 2014, at 4:05 AM, Simon Iten <itensimon@gmail.com> wrote:

without jack i should add...
On 11 Mar 2014, at 08:55, Simon Iten <itensimon@gmail.com> wrote:

hey dan,

unfortunately i’m in switzerland :-)

would be great if you could post your setup somewhere or send the infos! i compiled pd from source as well. i start it from console (with -rt) and it works without problems with the builtin sound card. maybe the cheap card from dx.com just does not work properly with udoo.
but please post your setup.

thanks

On 09 Mar 2014, at 22:26, Dan Wilcox <danomatika@gmail.com> wrote:

I've tried my both Roland Edirol UA-25 & UA-25EX and both work great. The dedicated USB controller makes these guys work as compared to an RPI where I can't get full duplex without tons of dropouts. I'm using a Linaro install which boots to the console and runs the PD through scripting. The speed is great as compared to my old wearable computer. 4 cores makes a difference.

I had to recompile my kernel to add midi support, but that's working great. It's not too bad, actually. I also built Pd-vanilal from source which was pretty easy using ./configure + make. I also have a script which fetches externals and builds/installs the agains vanilla so I have the few externals I need.

As with my previous experience running Pd + embedded Ubuntu, I get great performance with RT permissions, the -rt startup flag, and ALSA. Jack is needless overheard unless you want to work with other Jack-enabeld apps. Same with X windows, although my setup was running great in X with pd + ALSA in testing.

From your description, it sounds like your main issue is jack & pd are probably not running in realtime.

I can sent you my install notes if you want (or put them online, as I've been meaning to). Also, are based in the NE within driving distance to Pittsburgh? We could do a patching circle/UDOO setup afternoon :D

On Mar 9, 2014, at 5:14 PM, pd-list-request@iem.at wrote:

From: Miller Puckette <msp@ucsd.edu>
Subject: Re: [PD] udoo board sound issues
Date: March 9, 2014 at 5:07:54 PM EDT
To: Simon Iten <itensimon@gmail.com>
Cc: PD list <pd-list@iem.at>


H iSimon -

I haven't tried any but the built-in yet but I have a few USB interfaces
around here that I can try.  I'm about to go on an intense trip but should
be able to do some tests when I get back, assuming nobody else has figured
this out first.

cheers
Miller

On Sun, Mar 09, 2014 at 09:57:45PM +0100, Simon Iten wrote:
hey list,

does anybody that uses an udoo board have any recommendations on a usb-soundcard? should be very compact.
i tried a cheap one from dx and i could not get any good results (loads of xruns even with periods 3 and 1024 and up frames in qjackctl) this is on the ubuntu version from udoo.
are there some tweaks i can do to improve usb-sound capabilities? i thought maybe recompile the kernel with only usb-1 support since there is no option as on the pi to disable usb-2 via config, or am i missing something?
is it better to use the usb-soundcard without jack (pd does not like the usb-soundcard either when i tried briefly)?

the internal sound-input is to noisy for my application (guitar effect)

cheers. 
_______________________________________________
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list

--------
Dan Wilcox
@danomatika








--------
Dan Wilcox
@danomatika





_______________________________________________
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list


--------
Dan Wilcox
@danomatika






--------
Dan Wilcox
@danomatika






_______________________________________________
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list




-- 
Richie


--------
Dan Wilcox
@danomatika





_______________________________________________
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list


--------
Dan Wilcox
@danomatika








--------
Dan Wilcox
@danomatika








--------
Dan Wilcox
@danomatika






_______________________________________________
Pd-list@iem.at mailing list
UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list



_______________________________________________
Pd-list@lists.iem.at mailing list
UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list



--------
Dan Wilcox
@danomatika






--------
Dan Wilcox
@danomatika