On Fri, May 21, 2010 11:30, James Morris wrote:
> Just a quick update. I eventually settled on a 64bit 2D array, with each
> 64bit element in the array representing 64 spaces in the grid merely for
> tracking the state of space being used or unused. Turns out much faster
> this way. There will still be a list of blocks (or windows if it was a
> window manager) placed but this is an entirely separate issue. For example
> in the test code I keep track of the coordinates (returned by the
> algorithm) and dimensions of placed blocks in an array.
>
> code/demo here: http://jwm-art.net/boxyseq/freespace_grid.tar.bz2
not very portable: uses GCC builtins ctzl, and clzl, and probably other
stuff not so portable.
>
> video/demo here: http://jwm-art.net/art/freespace_grid.ogv
>
>
>
> running 'make' in the src, builds a demo/test program which produces
> similar output to the video - this is where the good old xterm wins - set
> font to 'tiny' and maximize vertically and widen it to atleast 128 chars
> and xterm displays fast enough it looks like an animation.
>
>
> I'm hoping this will work well enough for real-time placement, though the
> worst case scenario has changed from just being 255 windows placed, to
> something like.... placing a 1x1 block in a non-empty grid full of (any
> dimensioned window... it's difficult to work out the worst case here.
>
> james.
>
>
> On Fri, April 30, 2010 23:10, James Morris wrote:
>> Hello,
>>
>> I'm still trying to develop my boxy-sequencer idea. Basically, I've
>> tried
>> out a couple of algorithms for window-manager-like box placement within
>> a
>> grid and run into big problems.
>>
>> Because I need help/advice/ideas, I'm posting here as it has some
>> relevance, and I've also posted to
>>
>> http://stackoverflow.com/questions/2746590/fast-block-placement-algorithm-a…
>>
>> A quick run down of what's happened so far, the two algorithms:
>>
>> The first I directly based on Fluxbox's Row Smart placement algorithm.
>> Testing only the data structures, (and with gcc -O0) jack kicked out my
>> client after around 200 boxes were placed in a 128x128 grid. I then
>> counted how many times the algorithm looped through the entire list of
>> boxes and discovered (not in RT for this) that the 256th box placement
>> required around 14000 scans through the list of 255 previously placed
>> boxes (i've decided 256 boxes will be the worst-case-scenario).
>>
>> The second algorithm consists of a freebox data structure which
>> represents
>> a rectangular area of free unused space. It also forms a node in a
>> doubly
>> linked list, as well as four directional links to other freeboxes
>> touching
>> it (with a maximum of one freebox touching each side of a freebox)...
>> turns out rather complex to implement fully: 700+ loc for a partial
>> implementation of row-smart left-to-right top-to-bottom placement -
>> theres
>> also col-smart, r-to-l, b-to-t and all combinations.
>>
>> Through stackoverflow, I've come across spatial hashing.
>>
>> I wondered if anyone here uses this technique for the display of data
>> (would scrollable window views use such a thing?). worth it for a
>> 128x128
>> grid etc?
>>
>> I'm just after any general thoughts/insights you might have as to what
>> might or might not work. Anything worth pursuing.
>>
>> I'm afraid there's all sorts of little things to be considered so if you
>> do have something to suggest, please read the stackoverflow post first.
>>
>> Cheers for any help, if possible,
>> James.
>>
>>
>>
>
>
> _______________________________________________
> Linux-audio-dev mailing list
> Linux-audio-dev(a)lists.linuxaudio.org
> http://lists.linuxaudio.org/listinfo/linux-audio-dev
>
Just a quick update. I eventually settled on a 64bit 2D array, with each
64bit element in the array representing 64 spaces in the grid merely for
tracking the state of space being used or unused. Turns out much faster
this way. There will still be a list of blocks (or windows if it was a
window manager) placed but this is an entirely separate issue. For example
in the test code I keep track of the coordinates (returned by the
algorithm) and dimensions of placed blocks in an array.
code/demo here: http://jwm-art.net/boxyseq/freespace_grid.tar.bz2
video/demo here: http://jwm-art.net/art/freespace_grid.ogv
running 'make' in the src, builds a demo/test program which produces
similar output to the video - this is where the good old xterm wins - set
font to 'tiny' and maximize vertically and widen it to atleast 128 chars
and xterm displays fast enough it looks like an animation.
I'm hoping this will work well enough for real-time placement, though the
worst case scenario has changed from just being 255 windows placed, to
something like.... placing a 1x1 block in a non-empty grid full of (any
dimensioned window... it's difficult to work out the worst case here.
james.
On Fri, April 30, 2010 23:10, James Morris wrote:
> Hello,
>
> I'm still trying to develop my boxy-sequencer idea. Basically, I've tried
> out a couple of algorithms for window-manager-like box placement within a
> grid and run into big problems.
>
> Because I need help/advice/ideas, I'm posting here as it has some
> relevance, and I've also posted to
>
> http://stackoverflow.com/questions/2746590/fast-block-placement-algorithm-a…
>
> A quick run down of what's happened so far, the two algorithms:
>
> The first I directly based on Fluxbox's Row Smart placement algorithm.
> Testing only the data structures, (and with gcc -O0) jack kicked out my
> client after around 200 boxes were placed in a 128x128 grid. I then
> counted how many times the algorithm looped through the entire list of
> boxes and discovered (not in RT for this) that the 256th box placement
> required around 14000 scans through the list of 255 previously placed
> boxes (i've decided 256 boxes will be the worst-case-scenario).
>
> The second algorithm consists of a freebox data structure which represents
> a rectangular area of free unused space. It also forms a node in a doubly
> linked list, as well as four directional links to other freeboxes touching
> it (with a maximum of one freebox touching each side of a freebox)...
> turns out rather complex to implement fully: 700+ loc for a partial
> implementation of row-smart left-to-right top-to-bottom placement - theres
> also col-smart, r-to-l, b-to-t and all combinations.
>
> Through stackoverflow, I've come across spatial hashing.
>
> I wondered if anyone here uses this technique for the display of data
> (would scrollable window views use such a thing?). worth it for a 128x128
> grid etc?
>
> I'm just after any general thoughts/insights you might have as to what
> might or might not work. Anything worth pursuing.
>
> I'm afraid there's all sorts of little things to be considered so if you
> do have something to suggest, please read the stackoverflow post first.
>
> Cheers for any help, if possible,
> James.
>
>
>
On 05/20/2010 09:08 AM, Jörn Nettingsmeier-6 wrote:
> hi *!
>
> in the light of recent timer discussions (was it here or on
> jack-devel?), lwn.net has interesting coverage about the time stamp
> counter and its oddities in the recent weekly edition:
> http://lwn.net/SubscriberLink/388188/62e8027425e224f6/
> (free link, the rest of the issue will become available to
> non-subscribers in 14 days or so.)
>
> best,
>
> jörn
>
> _______________________________________________
> Linux-audio-dev mailing list
> Linux-audio-dev@...
> http://lists.linuxaudio.org/listinfo/linux-audio-dev
Thanks for the link!
Interesting read.
So this is the monster that's being used when you start jack with -c
cycle...
Greetings,
lieven
hi *!
in the light of recent timer discussions (was it here or on
jack-devel?), lwn.net has interesting coverage about the time stamp
counter and its oddities in the recent weekly edition:
http://lwn.net/SubscriberLink/388188/62e8027425e224f6/
(free link, the rest of the issue will become available to
non-subscribers in 14 days or so.)
best,
jörn
Hey guys!
In many DAWs it is not possible to use a vocoder, because it requires two
inputs - a modulator and a carrier. So in order for this to work you either
have to allow routing in the mixer (which none of the linux daws are capable
of, as far as I know) or else use vocoder as an external programm.
While the latter is a partial solution, it is not always desirable. Having
Vocoder as just a plugin inside a DAW is something that I personally always
needed.
I wanted to suggest a Vocoder with built-in Modulator.
This can be accomplished in two methods:
a. The user loads a wave file from the plugin's GUI, a wave file that would
act as a modulator.
b. The plugin has a set of predefined waves and/or a set of oscillators with
basic waveforms (like the famous Orange Vocoder VST).
There are two Vocoders on Linux that I know of - the LADSPA Vocoder (and its
LV2 port, done by a different programmer) and the Vocoder, created by the
guys of the Rakarrack team.
Both vocoders are definetely very capable tools.
I wanted to ask whether
1. modifying them in a manner I specified above is a difficult task
2. and whether someone would want to take it up.
Additionally, how difficult is it to include several features that would
make this vocoder more like an instrument rather than just a plugin that
does only one job, namely filter, fine tune for the modulator and LFO.
Louigi Verona.
================
| FAUST 0.9.24 |
================
GRAME - Centre National de Creation Musicale - is happy to announce
the release of FAUST 0.9.24. This version fixes several bugs,
and introduces some new possibilities in the language.
-------------
About FAUST :
-------------
FAUST (Functional Audio Stream) is a functional programming
language specifically designed for real-time signal processing and
synthesis. A distinctive characteristic of FAUST is to be fully
compiled. The FAUST compiler translates DSP specifications into
very efficient C++ code that works at sample level. It targets
high-performance signal processing applications, libraries and
audio plug-ins for a variety of audio platforms and standards. A
same FAUST specification can be used to easily generate native
JACK or ALSA applications, as well as CSOUND, LADSPA, MAX/MSP, PD,
Q, SC and VST plugins.
The Faust distribution can be downloaded at:
http://sourceforge.net/projects/faudiostream
Two mailing lists are available:
https://lists.sourceforge.net/lists/listinfo/faudiostream-develhttps://lists.sourceforge.net/lists/listinfo/faudiostream-users
In order to test FAUST without installing it, please refer to the
Online Faust Compiler:
http://faust.grame.fr
------------
What's new :
------------
- Explicit substitutions. The language has been extended with new
expressions type : exp[x1=def1; x2=def2; ...] allowing explicit
substitutions in the lexical environment of an expression. This
extension allows for instance, to customize an existing component
by replacing some of its internal definitions without having to
modify its source code. This extension is particularly useful to
promote better code reuse.
- Improved mathematical description (--mathdoc option) and support
for two new languages : German (-mdlang de) and Italian (-mdlang
it)
- Support for floating point numbers in scientific notation and
better precision for floating point constants. The precision used
to print a floating point constant in the generated C++ code is no
more limited to 6 digits. It is now dynamically adjusted to find
the minimal number of digits that will produce the same internal
representation when read back. This approach guarantees accuracy
without sacrificing for readability.
- All expressions are now systematically represented in polynomial
forms. For example x*x will be replaced by x^2. If x is a complex
expression the later form has several advantages, in particular to
limit CSE.
- Lazy semantics to select2 and select3 : the code generated for
select2 and select3 is now based on conditional expressions
((cond)?exp1:exp0 ) instead of tables. The resulting code is more
efficient as the stateless parts of the branches are not computed
every time but only when really needed.
- new --task-graph option. It produces a graphical representation
of the internal DAG of task in dot format (Graphviz
http://www.graphviz.org/). This DAG is useful for example to
understand the potential parallelism of a program as analyzed by
the Faust compiler
- Two new tools : faust2graph and faust2graphviewer. These tools
make use of the --task-graph option in order to produce the
graphical representation, as a PDF file, of the internal DAG of
tasks of a Faust program (require Graphviz).
- new reduce.lib library. It provides various operations on block
of samples based on a high order 'reduce(op, n)' fold-like
function. Moreover the music.lib library has been extended with
break-point functions and multiple decorrelated random and noise
generators. New flanger and stereowidth control have been added to
the effect.lib library.
- new iPhone architecture. It consists in a iphone-cocoa.cpp
architecture file and an Xcode template project to be used to
produce the applications. Use "make iphone" in the example folder
to build the examples for the iPhone.
- improved cross plateform compatibility and brand new visual
studio 2008 project to build Faust on windows machines.
----------
Bug Fixes:
----------
- Report error when non-integer table size is detected during
compilation
- Corrected partial application of power operator. Now ^(n) is
equivalent to \(x).(x^n) and not anymore to \(x).(n^x)
- Added missing faustpower definition when power function is
used only in table content.
- Fixed lock-free implementations of PopHead and PopTail functions on
work stealing queues in --scheduler mode
- Corrected missing dependencies in the internal DAG of tasks
- Added missing cache code to slow shared expressions used delayed
- Added missing cache code to foreign functions
----------------
Acknowledgments:
----------------
Many persons have been contributing to the FAUST project by
providing code for the compiler, architecture files, libraries,
examples, documentation, scripts, bug reports, ideas, etc.
I would like to thank them and especially: Fons Adriaensen, Tiziano
Bole, Baktery Chanka, Thomas Charbonnel, Damien Cramet, Etienne
Gaudrin, Albert Graef, Stefan Kersten, Victor Lazzarini, Matthieu
Leberre, Mathieu Leroi, Kjetil Matheussen, Remy Muller, Sampo
Savolainen, Nicolas Scaringella, Stephen Sinclair, Travis Skare,
Julius Smith, as well as my colleagues at GRAME, in particular :
Dominique Fober, Stephane Letz and Karim Barkati, and from the
ASTREE project : Jerome Barthelemy (IRCAM), Alain Bonardi (IRCAM),
Raffaele Ciavarella (IRCAM), Pierre Jouvelot (Ecole des
Mines/ParisTech), Laurent Pottier (U. Saint-Etienne)
Yann Orlarey
GRAME
Hi LADs,
Following up on the LAC Tools round-table, I've started a wiki page:
http://wiki.linuxaudio.org/wiki/tools_comparison
Since I've not taken part in the discussion, I'm missing a few footnotes
and explanation for keys in the context (eg. batch, sync). Could you
please enlighten me, or just fill in the missing content there.
IMHO it'd also make sense to do a 2nd table with orthogonal practical
information, for instance:
audio: JACK,ALSA,..
midi: JACK,ALSA,ALSA-seq
audio-file-formats: pcm,mp3,gig,mid,..
control: OSC,TCP,pipe,MMC,..
interop/sync: jack-transport,MTC
plugins (if applicable): LV2,LADSPA,VST,AU,..
Come to think of it, those could be integrated in the apps-database as
tags for each tool, similar to:
http://wiki.linuxaudio.org/apps/categories/jack_transport
but let's get the categories/tags straightened out before starting on an
implementation and posting it to LAU. What do you think?
Cheers!
robin
Greetings;
Can someone go look at lists.linuxaudio.org? The bottom line of the message
is timing out.
Thanks.
--
Cheers, Gene
"There are four boxes to be used in defense of liberty:
soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
"Live or die, I'll make a million."
-- Reebus Kneebus, before his jump to the center of the earth, Firesign
Theater
Hi everyone,
We have a new post in our department, which might interest some of you.
http://www.publicjobs.ie/publicjobs/en/star/goToJobDetails.do?id=419
Unfortunately, given my involvement in the selection process, I can't
answer any queries related to this post, but
my colleague Gordon Delap will be most happy to do so.
Maynooth is an interesting place to be at the moment, so I hope some
of you will find this worth a go.
All the best
Victor
guitarix is a simple Linux Rock Guitar amplifier and is designed
to achieve nice thrash/metal/rock/blues guitar sounds.
guitarix uses the Jack Audio Connection Kit as its audio backend
and brings in one input and two output ports to the jack audio graph.
guitarix provide a jack midi input port to connect a midi controller
(midi learn) and a (3 channel) jack midi output port, feed by a
(scalable) mix of the tuner and a beat-detector.
Release 0.08.0 comes with a bunch of major changes :
* new portmap window
* new BiQuad Filter
* new Moog Filter
* new Flanger
* new post processing tube3
* new two 10 band parametric EQ's
* new oversample mode selector(1x -8x)
* fix some bugs in midi out
* reworked tuner interface (show frequency(hz))
* new configure options (see ./waf --help for more info)
have fun
________________________________________________________________________
guitarix is licensed under the GPL.
Project page with screenshots:
http://guitarix.sourceforge.net/
download:
http://sourceforge.net/projects/guitarix/
please report bugs and suggestions in our forum here:
http://sourceforge.net/apps/phpbb/guitarix/
________________________________________________________________________
For capture, guitarix uses the great 'jack_capture'
(version >= 0.9.30) written by Kjetil S. Matheussen.
If you don't have it installed,
you can look here:
http://old.notam02.no/arkiv/src/?M=D
For extra Impulse Responses, guitarix uses the
zita-convolver library, and,
for up/down sampling we use zita-resampler,
both written by Fons Adriaensen.
If you don't have it installed, get it here:
http://www.kokkinizita.net/linuxaudio/index.html
We use the marvellous faust compiler to build the amp and effects and will say
thanks to
: Julius Smith
http://ccrma.stanford.edu/realsimple/faust/
: Albert Graef
http://q-lang.sourceforge.net/examples.html#Faust
: Yann Orlary
http://faust.grame.fr/
________________________________________________________________________
For faust users :
All used Faust dsp files are included in /guitarix/src/faust,
the resulting cc files are in /guitarix/src/faust-cc
The tools we use to convert (post-processing and plot)
the resulting faust cpp files to the needed include format,
stay in the /guitarix/tools directory.
________________________________________________________________________
regards
Hermann Meyer, James Warden, Andreas Degert
