Linux-audio-dev

linux-audio-dev@lists.linuxaudio.org

2 participants
7501 discussions

[linux-audio-dev] Fixed vs Floating point comparison.

by Juan Linietsky

Benno Sennoner and I were discussing today on IRC about the usual fixed point vs floating point (regarding to some resampling code) We developed some tests and ran them on a variety of computers. It would be interesting if ladders here could run them on different computers (and specially non x86, like amd64. or the Gx processors) so we can see what performance can we expect on each , and how things seem to be shaping for the future. It will also be a key factor on how our projects will develop in the future. The code is available at: http://reduz.dyndns.org/resamp_fixp.c // fixed point version http://reduz.dyndns.org/resamp_float.c // floating point version, portable http://reduz.dyndns.org/resamp_float_fistl.c // X86 VERSION ONLY!! Uses fistl instruction The results dont mean just int vs float performance. They also test for float->int conversion, which is common in most algorithms that work with buffers. It is a linearly interpolated resampling with volume control. please use GCC options - -O3 -ffast-math -march=<yourcpu> so it's a fair comparison results from other compilers is also appreciated! Here's some results for reference **************************************************************** vendor_id : AuthenticAMD model name : AMD-K6(tm) 3D processor cpu MHz : 412.508 cache size : 64 KB flags : fpu vme de pse tsc msr mce cx8 pge mmx syscall 3dnow k6_mtrr bogomips : 822.47 resamp_fixp - 0m8.460s resamp_float_fistl - 0m27.390s **************************************************************** vendor_id : AuthenticAMD model name : AMD Duron(tm) Processor cpu MHz : 951.701 cache size : 64 KB flags : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat pse36 mmx fxsr syscall mmxext 3dnowext 3dnow bogomips : 1900.54 resamp_float - 0m11.180s resamp_float_fistl - 0m5.810s resamp_fixp_optimized - 0m2.790s ************************************************ Benno gave me some results: Intel p4 1800 celeron resamp_float_fistl - 4.00user resamp_fixp - 4.51user (float faster?) -------------------------------- VIA nehemiah 1GHz resamp_float_fistl - 0m21.079s resamp_fixp - 0m7.129s ************************* Some results on SPARC: 2x 125MHz HyperSPARC resamp_fixp - user 2m59.010s resamp_float - user 0m44.290s (float faster! :) **************************** Intel 2.4 GHZ P4: resamp_float - 0m3.440s resamp_float_fistl - 0m2.960s resamp_fixp - 0m1.450s ************************ 1.25GHz G4 (laptop) resamp_float - 0m11.170s resamp_fixp - 0m2.130s ----------------------------------- Conclusions SO FAR. My own conclusions about the subject is that the float -> int conversion is STILL the biggest bottleneck on most common architectures. And until this is sorted out, fixed point is still the best solution for some specific cases, and I dont see any problem mixing it with floating point code. If you look at that algorithm closely in the source, you could replace "counter/ increment" for purely fixed point values, and do the rest (managing the samples) in float. This will undoubly speed it up.. Cheers! Juan Linietsky

21 years, 6 months

Re: [linux-audio-dev] Re: POSIX caps/realtime/root processes

by Kjetil Svalastog Matheussen

On Thu, 27 Nov 2003, Martijn Sipkema wrote: > [...] > > So how is the low-latency situation for 2.6? I did install 2.6 on > > my private machine, but was not able to get better performance > > than 2.4 with ll+pre (kicked out of jack-graph pretty soon with 128 > > frames period). Is there a trick to get better lowlatency performance with > > 2.6 I don't know about? > > Did you compile with CONFIG_PREEMPT enabled? > Yes. :) --

21 years, 6 months

[linux-audio-dev] Re: simple jack-client [jack-net-cat]

by vanDongen/Gilcher

Have you seen this: http://gsd.ime.usp.br/~lago/masters/ It is a networked LADSPA plugin. Maybe it could be made into a jack system. What would be nice is to have something like your net client on one computer, and a network jackdriver running the jackgraph on a different computer. You do get one period extra latency between the two graphs. G -- electronic & acoustic musics-- http://www.xs4all.nl/~gml

21 years, 6 months

[linux-audio-dev] simple jack-client [jack-net-cat]

by Florian Schmidt

Hi, i have this silly idea of implementing a very simple jack client. It would do nothing else, but offer a number of writable and readable ports [lets assume 2 of each for now]. Now, in the background, this jack-client opens a number of UDP sockets [me is not a network expert, so bear with me here :)]. 2 UDP ports on which it listens to incoming data, which it then feeds into the readable ports. Any data that comes in on its writable ports will be made available as simple. It opens 2 UDP sockets, specifies a target adress and sends out the audio data. The mechanism is basically completely connection-less. packets that belong to the same stream just come from the same ip-adress, maybe carrying an index number to denote the port number. This way it would be vey easy to stream data from one jack graph to another on a different computer. Latencies will not be considered at all. These connections must be considered to be async. So a send to another computer, there feeding into an effect and then sending back will not be really a "realtime" thing [hmm, well, depends on the network speed and how big then network buffers are, etc.. i assume both on a private local lan here] Also, there probably must be some basic/minimal network protocol be implemented. At least a sequential numbering or timestamp, so the receiver can maintain a buffer and sort packets that come in out of order.. Again i'm no expert on this. But it sounds failry simple. What do you think? This could of course be done arbitrarily complex [with timing stuff, connection handling, authentification, etc], but i look for something dead simple. Flo -- music: http://www.soundclick.com/bands/9/florianschmidt.htm

21 years, 6 months

[linux-audio-dev] Re: POSIX caps/realtime/root processes

by Kjetil Svalastog Matheussen

Paul Davis: > >Since mainstream capabilities support seems always to be somewhere > >over the horizon, I am interested in the patch Paul and Steve > >mentioned. IIUC, it defines a control file in /proc which, if > >enabled, allows any process access to scheduling and memory locking > >privileges. No other capabilities are provided. I would love to see > >a copy of this patch to study exactly what it does. > > its a very simple patch, IIRC. it just short-circuits the checks on > uid==0 and/or capabilities when assigning SCHED_FIFO and/or locking > memory. > > i'm looking for it in my archives. i'm a bit worried i may have I couldn't wait til you found it, so I wrote one from scratch instead. :) The url below point to a hackish patch againt 2.4.23-rc1, and yes, it is very simple. Works by setting /proc/sys/kernel/setschedandmlock to 1. http://www.notam02.no/arkiv/src/schedmlockpatch-2.4.23-rc1 --

21 years, 6 months

[linux-audio-dev] Announcing the ultra-mega-alpha release of Specimen

by Peter Bessman

Specimen is a midi controlled sampler, and it's very first public release is available from http://www.gazuga.net/specimen.tar.gz In all honesty, I advise against downloading this in it's current state unless you intend to hack on it or play with it for sheer amusement value only. There are a few TODO items that must be taken care of before this can be considered usable software (as the next release will be). The main reason for this release is because I said I was going to make a release in a couple of weeks 15 days ago. It's very nascent stuff. You have been warned. [pb]

21 years, 6 months

[linux-audio-dev] Linux Audio article in DATOR MAGAZIN 2003-12 (swedish)!

by Roger Larsson

7 pages with written by Albert Veli Take a look at the front page :-) http://www.datormagazin.se/ssjs/aktuellt_nr.html It mentions, with varying levels of detail: * Alsa (three pages) * GNOME ALSA Mixer (with screen dump? - it look old and boring compared to KMix in KDE 3.2 IMHO) * Jack * Qjackctl (with screen capture?) * Fluidsynt * Timidity++ * LADSPA * Jack-rack * Ecamegapedal (with screen capture) * Rosegarden-4 (with screen capture) * ReZound (with screen capture) * Muse * Audacity * Ardour ["problematic to compile, steep learning curve" the article suggests looking at www.djcj.org/LAU/quicktoots * Ecasound * AlsaModularSynth * Legasynth * SpiralSynth * Pd * Soundtracker, cheesetracker * Alsaplayer * xmms /RogerL -- Roger Larsson Skellefteå Sweden

21 years, 6 months

Re: [linux-audio-dev] plugin GUIs .. Mediastation, LADSPA ?

by Juhana Sadeharju

>From: Benno Senoner <sbenno(a)gardena.net> > >the method I use to communicate between the GUIs is SYSV message queues >because they can be multiclient but the API is abstracted so the >underlying transport model can be chosen arbitrarily. I have written and I'm writing a message system where client sends, e.g., the following msg to the server: read <file id> <loc> <len> <addr> <return value> which reads a block of audio from a file to a (shared) memory location. I did not see the return value in your system (and don't figure out how one goes without it). The server sends both the result and the given return value back to the client. The return value could be something which only the client knows, e.g., array index, memory address, text. Something which associates the result with the request. The implementation is too complex to my taste as the messages are combination of raw byte data and string data. Each message starts with <int> which tells the length of the message. I have written a set of routines to assemble and disassemble these messages. I send these messages via pipes/sockets, via shared memory, and whatever. In any case, such message system should be general enough, and the return value brings it to that direction. Hmm... what do you mean by multiclient? Is the writing to the message queue an "atomic" operation for long messages? I use one pipe/socket per client to one server because I suspect only the server can make the long messages "atomic" by interleaving them correctly. Likewise, I use one lock-free shared-memory FIFO per client to one server. Regards, Juhana

21 years, 6 months

[linux-audio-dev] free MPEG-4 AAC Encoder

by Richard Spindler

Have you seen this? http://faac.sourceforge.net/ I was browsing sourceforge, and found this, it might be of interest to some people on this list. -Richard

21 years, 6 months

[linux-audio-dev] Retro speech synthesis

by David Olofson

Inspired by the Vocaloid thread, an old idea sprung back to mind; retro "chip style" speech synthesis. Is there a simple, minimalistic, Free/Open Source phoneme-to-audio synthesiser out there? I'm not terribly interested in the text-to-phonemes part and other higher level stuff, as I intend to use this for sound effects in games and (other) toys. Doesn't hurt if CPU and memory requirements are very low, but I'm probably going to render words and phrases off-line anyway (at install or load time), for later processing as normal sound effects. //David Olofson - Programmer, Composer, Open Source Advocate .- Audiality -----------------------------------------------. | Free/Open Source audio engine for games and multimedia. | | MIDI, modular synthesis, real time effects, scripting,... | `-----------------------------------> http://audiality.org -' --- http://olofson.net --- http://www.reologica.se ---

21 years, 6 months

Jump to page:

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