Linux-audio-dev

linux-audio-dev@lists.linuxaudio.org

3 participants
7535 discussions

[LAD] Linux Audio Documentaion Effort : (Was "Question 0")

by Harry van Haaren

Following recent emails on list about documentation and learning "Linux Audio Programming", I've been thinking about an effort to help people get started with Linux Audio Coding. I think some "beginner" coding documentation on Linux Audio would be a great asset to the community, and I'm willing to contribute to such an effort. As Robin Gareus mentioned in another thread, a "FLOSS" manual is probably the best way to go for a community effort on documenting. I've been doing some "tutorial" style programming articles on my blog harryhaaren.blogspot.com and I'd have no problem sharing the examples there (some basic GTK stuff, some small JACK apps, combining the GUI / JACK stuff ). Problem is that I was learning as I was going along, and there are some *fundamental* issues with some of the tutorials (especially with regards to thread safe code & more advanced programming concepts) Personally I'd be even more enthusiastic about such an effort if some of the veteran Linux audio guys were to get on board and ensure the content is of a high quality. (As I'm a self though programmer, there are some big gaps in my knowledge, and I'd not like to provide bad sample code, or share bad concepts.) Of course some issues will arise in choosing how to document Linux Audio, and some typical "flame" topics like GUI toolkits, libraries etc will arise. I have no idea how we can best avoid that issue, except by following the "if you think it should be thought that way write the tutorial"... the downside of this is that if one tutorial uses toolkit <X> and the next toolkit <Y>, the average beginning coder is going to get lost in implementation details and that defeats the purpose of documentation :D -Harry

14 years, 2 months

[LAD] Another call for brave testers: new Guitarix LADSPA plugins

by Andreas Degert

Hi, we'd like to ask those who are interested in the new Guitarix LADSPA plugins to run them on their systems and give us some feedback. They do fine on our own machines and seem to be fit for wider testing. The plugins "Guitarix Amp" and "Guitarix Stereo Fx" wrap all of the entire sound engine of Guitarix. You can load a preset that you defined with the Guitarix program, and even define some parameters for DAW automation. It's explained in our wiki: https://sourceforge.net/apps/mediawiki/guitarix/index.php?title=How_to_use_… Our SVN: svn co http://guitarix.svn.sourceforge.net/svnroot/guitarix/trunk guitarix After checkout, build with ./waf configure && ./waf && sudo ./waf install for an installation to /usr/local. Our tracker: http://sourceforge.net/tracker/?group_id=236234 Your feedback will be welcome here or there or anywhere.. Or in our forum: http://sourceforge.net/apps/phpbb/guitarix/ so please, don't be shy and tell us your test results or if you think the concept is usable / unusable :-) ciao Andreas

14 years, 2 months

Re: [LAD] Pipes vs. Message Queues

by Kjetil Matheussen

David Robillard: > On Fri, 2011-11-25 at 15:21 +0100, Nick Copeland wrote: > [...] >> >> So if the pipe() is replaced with >> >> socketpair(PF_UNIX, SOCK_STREAM, PF_UNSPEC, pipe_fd); >> >> Then the issue I was seeing goes away. Perhaps the pipe() code has not >> been >> optimised since sockets were developed to replace them when IPC suddenly >> needed to be between hosts rather than processes? Pure conjecture. > > Added sockets: > > Normal scheduling: > > $ ./ipc 4096 1000000 > Sending a 4096 byte message 1000000 times. > Pipe recv time: 7.048740 > Pipe send time: 7.048648 > Socket send time: 2.365210 > Socket recv time: 2.365292 > Queue recv time: 2.072530 > Queue send time: 2.072494 > > SCHED_FIFO: > > $ ./ipc 4096 1000000 > Sending a 4096 byte message 1000000 times. > Pipe send time: 5.279210 > Pipe recv time: 5.279508 > Socket send time: 2.709628 > Socket recv time: 2.709645 > Queue send time: 5.228892 > Queue recv time: 5.228980 > > Interesting. I find sockets being significantly faster than the much > simpler pipes quite counter-intuitive. > > Code at the same location: > > http://drobilla.net/files/ipc.c > On my computer, sockets are sometimes slower than both pipes and queues. However, by making sure the two processes doesn't change CPU during runtime, sockets are always the fastest, and I get consistent results: Realtime, same CPU: [kjetil@ttleon c]$ ./ipc 4096 2000000 Sending a 4096 byte message 2000000 times. Pipe send time: 7.762312 Pipe recv time: 7.762423 Socket send time: 4.607025 Socket recv time: 4.607103 Queue send time: 7.278029 Queue recv time: 7.278052 Realtime, different CPUs: [kjetil@ttleon c]$ ./ipc 4096 2000000 Sending a 4096 byte message 2000000 times. Pipe recv time: 5.520099 Pipe send time: 5.519875 Socket send time: 2.856027 Socket recv time: 2.856136 Queue recv time: 3.049226 Queue send time: 3.049159 Non-realtime, same CPU: [kjetil@ttleon c]$ ./ipc 4096 2000000 Sending a 4096 byte message 2000000 times. Pipe send time: 7.331822 Pipe recv time: 7.332272 Socket send time: 4.520883 Socket recv time: 4.520959 Queue send time: 7.103408 Queue recv time: 7.103448 Non-realtime, different CPUs: [kjetil@ttleon c]$ ./ipc 4096 2000000 Sending a 4096 byte message 2000000 times. Pipe recv time: 4.305804 Pipe send time: 4.305587 Socket send time: 2.899696 Socket recv time: 2.899806 Queue recv time: 3.028210 Queue send time: 3.028151 [kjetil@ttleon c]$ diff -u ipc.c~ ipc.c --- ipc.c~ 2011-11-26 00:55:38.000000000 +0100 +++ ipc.c 2011-11-26 14:51:44.003407960 +0100 @@ -41,9 +41,10 @@ // Comment out this line to disable RT scheduling //#define SET_RT_SCHED 1 -#ifdef SET_RT_SCHED +#define _GNU_SOURCE +#define __USE_GNU #include <sched.h> -#endif +#include <pthread.h> static inline double elapsed_s(const struct timespec* start, const struct timespec* end) @@ -177,6 +178,14 @@ return time; } +static void bound_thread_to_cpu(int cpu){ + cpu_set_t set; + CPU_ZERO(&set); + CPU_SET(cpu,&set); + pthread_setaffinity_np(pthread_self(), sizeof(cpu_set_t), &set); +} + int main(int argc, char** argv) { @@ -218,6 +227,11 @@ return 1; } + if (child_pid ==0) + bound_thread_to_cpu(0); + else + bound_thread_to_cpu(1); // set to 0 to run both on the same CPU + #ifdef SET_RT_SCHED struct sched_param param; param.sched_priority = 90;

14 years, 2 months

[LAD] Pipes vs. Message Queues

by David Robillard

I got curious, so I bashed out a quick program to benchmark pipes vs POSIX message queues. It just pumps a bunch of messages through the pipe/queue in a tight loop. The results were interesting: $ ./ipc 4096 1000000 Sending a 4096 byte message 1000000 times. Pipe recv time: 6.881104 Pipe send time: 6.880998 Queue send time: 1.938512 Queue recv time: 1.938581 Whoah. Which made me wonder what happens with realtime priority (SCHED_FIFO priority 90): $ ./ipc 4096 1000000 Sending a 4096 byte message 1000000 times. Pipe send time: 5.195232 Pipe recv time: 5.195475 Queue send time: 5.224862 Queue recv time: 5.224987 Pipes get a bit faster, and POSIX message queues get dramatically slower. Interesting. I am opening the queues as blocking here, and both sender and receiver are at the same priority, and aggressively pumping the queue as fast as they can, so there is a lot of competition and this is not an especially good model of any reality we care about, but it's interesting nonetheless. The first result really has me thinking how much Jack would benefit from using message queues instead of pipes and sockets. It looks like there's definitely potential here... I might try to write a more scientific benchmark that better emulates the case Jack would care about and measures wakeup latency, unless somebody beats me to it. That test could have the shm + wakeup pattern Jack actually uses and benchmark it vs. actually firing buffer payload over message queues... But I should be doing more pragmatic things, so here's this for now :) Program is here: http://drobilla.net/files/ipc.c Cheers, -dr

14 years, 2 months

Re: [LAD] sliders/fans

by Jeff McClintock

> In other words, you'd expect such a system to behave as if you > had two faders in series. > > Now if the DSP code only sees the sum of the two values (as it > should, having a VCA group is just a user interface issue), Ah, You just contradicted yourself. If you expect the system to behave like two sliders, then the model must represent 2 sliders. The user widgets should not embody decision making, that implies intelligence in the GUI, which is not strictly model-view-controller. Imagine your device supplemented with a 'dumb' pair of MIDI controllers complementing the two GUI sliders, they could not correctly implement the complex interaction between the two gain settings, therefore you need to go back and move that 'intelligence' into the model. Best Regards, Jeff > Message: 18 > Date: Thu, 24 Nov 2011 20:45:09 +0000 > From: Fons Adriaensen <fons(a)linuxaudio.org> > Subject: Re: [LAD] sliders/fans > To: linux-audio-dev(a)lists.linuxaudio.org > Message-ID: <20111124204509.GA14316(a)linuxaudio.org> > Content-Type: text/plain; charset=us-ascii > > On Thu, Nov 24, 2011 at 02:21:25PM -0500, David Robillard wrote: > > > Agreed. Everything here is about the *view*. How that maps to > actual > > parameter values is an underlying model issue. > > Not always. Consider the case of 'VCA' groups for faders. That > is: you have a slider that controls the gain of a group of > channels (without those being mixed). The effective channel > gain (in dB) is the sum of the per channel fader value and > the one from the group fader. The model sees only this sum. > > Now a fader has to go down to zero gain (-inf dB). So you would > map the lowest possible (finite) value of the widget to something > that the model (the DSP code) would translate to 'off'. > > The question is then: is this exception handled by the widget and > the DSP code, or by the DSP code only ? > > Suppose the minimum value of the widget would correspond to say > -100 dB if not handled specially. If you just have a single fader > per channel, you could arrange for the model or the DSP code to > translate that to 'off'. That is no longer the case if you have > 'VCA' faders. > > There are two thing you'd expect from such a system: > > * If either the channel or the group fader is at minimum, then > the channel must be off (zero gain). > > * If the channel fader is at -50 dB, and the group at -60 dB > you don't want zero gain, but -110 dB. Becaus either fader is > still in a position where you'd expect that moving it makes a > difference. > > In other words, you'd expect such a system to behave as if you > had two faders in series. > > Now if the DSP code only sees the sum of the two values (as it > should, having a VCA group is just a user interface issue), then > that implies that the mapping of the minimum fader position (e.g. > -100 dB) to something that would be interpreted as 'off' by the > DSP code (e.g. -9999999 dB) _must be done by each individual > fader_. > > > Ciao, > > > -- > FA > > Vor uns liegt ein weites Tal, die Sonne scheint - ein Glitzerstrahl. > > >

14 years, 2 months

[LAD] Question 0

by C Net

Hi at all, I am interested in learning to develop linux audio. Do you know any links to get started? Thanks in advance.

14 years, 2 months

[LAD] RFC: LV2 state

by David Robillard

Here is the current version of the LV2 state extension, which defines the model for plugin state, and a mechanism for saving and restoring it: http://lv2plug.in/ns/ext/state It's time to tag this one as stable, unless anyone can see any issues (i.e. speak now, or forever hold your peace). If anyone has the time to give it a quick read-through, feedback would be appreciated. I have done a lot of work on the documentation lately, hopefully everything should be clear. This is currently implemented in Ardour 3 SVN, QTractor SVN, and a patch for LinuxSampler SVN is available here: http://drobilla.net/files/linuxsampler_lv2_state_0_4.diff Thanks, -dr

14 years, 2 months

[LAD] jack transport change accuracy for looping

by Iain Duncan

Just wondering if I understand this correctly. I making a loop based app for step sequencing. When I previously did this in Csound, I clocked it off a phasor, so the timing was sample accurate ( but that brought all it's own issues to be sure ). I'm wondering whether I should do the same thing in jack app, or use the jack transport clock, or some hybrid. My question, am I correct in understanding that if I use the jack transport position to rewind in time, I'll get: C) any other clients with running audio looping back to ( may or may not be desirable ) B) a jitter based on the amount of time left between when the loop should end and the end of the frame buffer in which the loop length runs out? Has anyone solved B? Could it be done by some complex tempo cheating trick? Does anyone have any methods they've used for tight timing of looping in a jack app? Pointers at code appreciated of course. =) thank! Iain

14 years, 2 months

Re: [LAD] tutorial or example for using jack ring buffer between threads?

by harryhaaren＠gmail.com

On , Iain Duncan <iainduncanlists(a)gmail.com> wrote: > Thanks! Did you just write it? Yup. As in literally just there. And I was reading your post in the new RAUL thread as you were typing that :D All the best, -Harry

14 years, 2 months

[LAD] Fw: Re: "bleeding edge html5" has interesting Audio APIs

by Fons Adriaensen

On Tue, Nov 22, 2011 at 09:13:37PM +0100, Nick Copeland wrote: > If you are using a toolkit that has a data flow of the following: > > pointer motion->graphical display->values->application->output > > Well, basically that is broken as you have a flow that is > > input->output->input->application->output > > invariably that is going to lead to issues. The tail (the toolkit) is wagging the dog > (the application) as it imposes restrictions on the values the application is allowed > to see. > > In my opinion (ok, it is only opinion) is that the correct flow is > > input->application->output Yes, I see your point, and it makes a lot of sense. So what would be required is * compute the new parameter value from - a stored state in 'paramater space' rather than 'widget space' - and pointer (mouse) gestures, * update the widget according to that value. This is more or less what I do in the rotary controls used in e.g. zita-at1 and zita-rev1. It's possible because the mouse movement and the visual representation of the value (the angle of the line on the rotary knob) are not directly related anyway. But this is not how most (all) toolkits work. You could probably use them in the way you suggest with some extra effort. But in many cases (e.g. linear sliders) the pointer and widget would have to remain in sync visually, which then means that your resolution in paramater space can't be better than the visual one. Unless you allow the pointer to move faster than the visual object it is controlling (which is what I do in the 2-D panner, but it's possible only because the widget is so small). Ciao, -- FA Vor uns liegt ein weites Tal, die Sonne scheint - ein Glitzerstrahl.

14 years, 2 months

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

Linux-audio-dev