41 replies to this topic

[Paradox Uncreated]

Renoise, on the Ubuntu 64-bit distribution of Linux. (Note. 32-bit is recommended, if you need compatibility with non-opensource.)

As you know, there has been a shortage of fully professional mastering-limiters on Linux.
However, I can happily announce the availability of one now. And it is opensource and licenced under the GPL.
(So if you want to compile it on SPARC, you can!)

Please visit www.sourceforge.com/projects/pxu

A track demonstrating the plugins can be downloaded here: Space Marley 2010.xrns

Edited by Paradox Uncreated, 27 February 2010 - 18:05.

[Luka P. (Nova deViator)]

As you know, there has been a shortage of fully professional mastering-limiters on Linux.
However, I can happily announce the availability of one now. And it is opensource and licenced under the GPL.
(So if you want to compile it on SPARC, you can!)

Please visit www.sourceforge.com/projects/pxu

A track demonstrating the plugins can be downloaded here: Space Marley 2010.xrns

hey, looking lovely, thanks for gpl code!
can you explain in what way does this one differ or is better from compressor, bus compressor or maximizer?

it would be nice to have the names shorter so it doesn't take so much space. "pxu limiter" would be enough.

bigups!

Edited by Nova deViator, 03 March 2010 - 10:54.

[Paradox Uncreated]

hey, looking lovely, thanks for gpl code!
can you explain in what way does this one differ or is better from compressor, bus compressor or maximizer?

it would be nice to have the names shorter so it doesn't take so much space. "pxu limiter" would be enough.

bigups!

Praised Be We, in correct Ubuntu spirit.
The maximizer in Renoise isn't quite up to professional tasks, IMO. There is some noise in it, and I would not allow that in my mixes. You can set the buffering lenght also, with my limiter, and also it uses a gaussian filter, so the whole thing is optimal, plus other things. People have various ways of doing things, and I do limiting very generally, so to avoid any colorization, or signature. If you are of the opionion that analog is better than digital, you probably will not quite understand this, and expect certain "effects". This is not what this limiter is about. It is simply about optimal design, and low noise. The purest sound possible.

And please, names.. I got used to long names a long time ago, with the Amiga, so please don't drag that up. It is already shortened in the mixview, Which the dosname fans will also like.

Edited by Paradox Uncreated, 04 March 2010 - 03:00.

[Paradox Uncreated]

I noticed that the thread name was modified also. Moderator also shortened the name, however know that that is not the correct and full name. However to shorten it to Pxu plugins, once they are known, is natural and fine. Read the text in the archive for the reason of the name.

I did this limiter to release a fully professional limiter to the GPL world, so that it would never need to be done again. No engineer should have to solve this task again and again, and if you would like to modify it, for experimental, or "analog" (one pole) simulations, you can. "Libre software", or intellectual freedom, as I would call it, or simply progress of society. It helps no-one that this task is solved time and time again, suboptimally and wasting everyone time, and hindering the progress of society.

Incidentally it has nothing to do with the plugs in renoise, renoise is an great sequencer. Although while we are on the subject of opensource, I would feel better if it was GPL'd.

[BreemiX]

Thanxs a lot... I will test the plugins...

[Paradox Uncreated]

Thanxs a lot... I will test the plugins...

Good, give me feeback, whether installation and testing went ok, I haven't had anyone test them yet, officially.Also very soon, coming up, is the v1.0 release, which includes a 4way (band) limiter, so this is going to be extremely good. I am not going to do optimized bandsplitting for v1.0 though, so it will take some cpu, but multibands are usually only used on the mix anyway. The limiter is extremely optimized, (without sacrificing audioquality), so it can be used probably on 20 tracks already though.

On a sidenote, I also compiled a kernel with the make localmodconfig + some tweaks, which was very nice. Highly recommended. So I am going to be able to run as many plugins as possible here.. If there should be a performance problem with any of the plugins, if it should be fixed sometime probably, however I am not going to optimize the bandsplitters in the 4-way in v1.0, because I need it now, and haven't bothered to look into reverse TIIR, yet.

Peace Be With You.

[BreemiX]

Good, give me feeback, whether installation and testing went ok, I haven't had anyone test them yet, officially.Also very soon, coming up, is the v1.0 release, which includes a 4way (band) limiter, so this is going to be extremely good. I am not going to do optimized bandsplitting for v1.0 though, so it will take some cpu, but multibands are usually only used on the mix anyway. The limiter is extremely optimized, (without sacrificing audioquality), so it can be used probably on 20 tracks already though.

On a sidenote, I also compiled a kernel with the make localmodconfig + some tweaks, which was very nice. Highly recommended. So I am going to be able to run as many plugins as possible here.. If there should be a performance problem with any of the plugins, if it should be fixed sometime probably, however I am not going to optimize the bandsplitters in the 4-way in v1.0, because I need it now, and haven't bothered to look into reverse TIIR, yet.

Peace Be With You.

I have tested the PXU Limiter Stereo :
- When the buffer is down under 0.02 the sound in Renoise is cut...then no way to have it back.
- Need to restart Renoise to have the sound again.

[Paradox Uncreated]

I have tested the PXU Limiter Stereo :
- When the buffer is down under 0.02 the sound in Renoise is cut...then no way to have it back.
- Need to restart Renoise to have the sound again.

Ok, avoid low buffer settings for the moment. About the sound disappearing, are you sure you haven't got autogain on?

Btw 0.32 = 1.5ms buffer, which is typical for many limiters, including L2.

Edited by Paradox Uncreated, 04 March 2010 - 17:13.

[BreemiX]

Ok, avoid low buffer settings for the moment. About the sound disappearing, are you sure you haven't got autogain on?

Btw 0.32 = 1.5ms buffer, which is typical for many limiters, including L2.

No Autogain enabled on master...

It is followed by 3 Renoise #Send Devices


By the way, the sound quality of the plugins is very good...

[Paradox Uncreated]

No Autogain enabled on master...

It is followed by 3 Renoise #Send Devices


By the way, the sound quality of the plugins is very good...

Hehehe, it should be the best. I'm going to limit the buffer minimum to some value, to avoid this in the next version.

Peace Be With You!

Btw, if anyone knows if Renoise on Linux, supports latency-compensation for ladspa's let me know, and I'll implement that as well.

Edited by Paradox Uncreated, 04 March 2010 - 17:44.

[BreemiX]

Hehehe, it should be the best. I'm going to limit the buffer minimum to some value, to avoid this in the next version.

Peace Be With You!

Btw, if anyone knows if Renoise on Linux, supports latency-compensation for ladspa's let me know, and I'll implement that as well.

Will test the others later...

Peace Bro !

[BreemiX]

Will test the others later...

Peace Bro !

Tested Stereo Quantize and Dither...

- Same problem with the "Zoom" cursor...(sound is going away)

Removing the plugin gives the sound back... (don't know if it is really a bug...)

[Paradox Uncreated]

Tested Stereo Quantize and Dither...

- Same problem with the "Zoom" cursor...(sound is going away)

Removing the plugin gives the sound back... (don't know if it is really a bug...)

Oh yes, I noticed a bug in dither myself aswell, I'll try and fix it for the next release.

[kazakore]

Help a complete n00b get these installed in Linux (Ubuntu Studio 9.10)

Also are they likely to be appearing in the repositories for 10.4?

[Paradox Uncreated]

Help a complete n00b get these installed in Linux (Ubuntu Studio 9.10)

Also are they likely to be appearing in the repositories for 10.4?

I will update this repository, with the current version, in a weeks time.

https://launchpad.ne...xu/ archive/pxu

Edited by Paradox Uncreated, 14 March 2010 - 14:09.

[Paradox Uncreated]

I will update this repository, with the current version, in a weeks time.

https://launchpad.ne...xu/ archive/pxu

Btw, the archive should contain enough information for anyone to compile the plugins now.

[kazakore]

Btw, the archive should contain enough information for anyone to compile the plugins now.

Still says there's no files in the PPA.

[a773]

Trying to compile the plugins (ubuntu 9.10), I get these errors:

atte@vestbjerg:~/software/pxu_plugin_suite/the_pxu_linux_opensource_gpl_plugin_suite/src$ make
cc  -march=native -O2 -pipe -fomit-frame-pointer -mfpmath=sse,387 -masm=intel -mpc64 -mcx16 -mno-push-args -maccumulate-outgoing-args -momit-leaf-frame-pointer -ftracer -fmodulo-sched -fmodulo-sched-allow-regmoves -fno-branch-count-reg -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload -funsafe-loop-optimizations -fmerge-all-constants -fsched-spec-load -fsched-spec-load-dangerous -fsched2-use-superblocks -fsched2-use-traces -fsee -fipa-struct-reorg -fwhole-program -combine -ftree-loop-linear -floop-interchange -floop-strip-mine -floop-block -ftree-loop-distribution -ftree-loop-im -ftree-loop-ivcanon -fivopts -ftree-parallelize-loops=2 -pthread -fvect-cost-model -fvariable-expansion-in-unroller -freorder-blocks-and-partition -ffast-math -freciprocal-math -ffinite-math-only -fno-signed-zeros -fno-trapping-math -frtl-abstract-sequences -fsingle-precision-constant -frename-registers -fpeel-loops -funswitch-loops -fbranch-target-load-optimize -fbranch-target-load-optimize2 -fsection-anchors -o plugins/pxu_filter.o -c plugins/pxu_filter.c
plugins/pxu_filter.c:1: warning: this target does not support ‘-fsection-anchors’
plugins/pxu_filter.c: In function ‘connectPortToPxu’:
plugins/pxu_filter.c:166: warning: branch target register load optimization is not intended to be run twice
{standard input}: Assembler messages:
{standard input}:216: Error: junk `PTR [eax+56]' after expression
{standard input}:271: Error: junk `PTR [ebx+56]' after expression
{standard input}:272: Error: junk `PTR [esp+144]' after expression
{standard input}:331: Error: junk `PTR [ebx+116]' after expression
{standard input}:333: Error: junk `PTR [ebx+104]' after expression
{standard input}:339: Error: junk `PTR [esp+176]' after expression
{standard input}:341: Error: junk `PTR [ebx+68]' after expression
{standard input}:343: Error: junk `PTR [esp+144]' after expression
{standard input}:367: Error: junk `PTR [ebx+80]' after expression
{standard input}:369: Error: junk `PTR [ebx+92]' after expression
{standard input}:390: Error: junk `PTR [esp+176]' after expression
{standard input}:481: Error: junk `PTR [esp+176]' after expression
{standard input}:572: Error: junk `PTR [esp+176]' after expression
{standard input}:673: Error: junk `PTR [ebx+68]' after expression
{standard input}:675: Error: junk `PTR [ebx+80]' after expression
{standard input}:677: Error: junk `PTR [ebx+92]' after expression
{standard input}:679: Error: junk `PTR [ebx+104]' after expression
{standard input}:680: Error: junk `PTR [ebx+116]' after expression
make: *** [../plugins/pxu_filter.so] Error 1

Any idea what I'm doing wrong?

[Antischocken]

cc1: error: unrecognized command line option "-floop-interchange"
cc1: error: unrecognized command line option "-floop-strip-mine"
cc1: error: unrecognized command line option "-floop-block"
cc1: error: unrecognized command line option "-ftree-loop-distribution"
plugins/pxu_limiter.c:1: warning: this target does not support ‘-fsection-anchors’
make: *** [../plugins/pxu_limiter.so] Error 1

Ubuntu 9.04 32bit

[vadarfone]

Quality!

Looking forward to testing this out tonight.

At last it seems people are starting to work with Linux to make decent gear.

Very good work mate!

[Paradox Uncreated]

These are the switches that I am using now:

HOSTCFLAGS = -w -march=native -O2 -pipe -fomit-frame-pointer -mfpmath=sse,387 -fgcse-lm -fgcse-sm -fgcse-las -fgcse-after-reload -fpredictive-commoning -ftree-vectorize -fvect-cost-model -floop-optimize -fivopts -freorder-blocks-and-partition -ftracer -fbranch-target-load-optimize -fbranch-target-load-optimize2 -fbtr-bb-exclusive -fmodulo-sched -fno-branch-count-reg -ftree-loop-linear

You may want to edit the makefile if you are having problems. Just replace everything with -O2, if you are unsure about what to use.

[Antischocken]

I replaced these 4 with -O2:

cc1: error: unrecognized command line option "-floop-interchange"
cc1: error: unrecognized command line option "-floop-strip-mine"
cc1: error: unrecognized command line option "-floop-block"
cc1: error: unrecognized command line option "-ftree-loop-distribution"

after that I get this message with a "hanging" Terminal:

plugins/pxu_limiter.c:1: warning: this target does not support ‘-fsection-anchors’
plugins/pxu_limiter.c: In function ‘connectPortToPxu’:
plugins/pxu_limiter.c:198: warning: branch target register load optimization is not intended to be run twice

Only limiter.o has been created that far.

[Paradox Uncreated]

I replaced these 4 with -O2:

cc1: error: unrecognized command line option "-floop-interchange"
cc1: error: unrecognized command line option "-floop-strip-mine"
cc1: error: unrecognized command line option "-floop-block"
cc1: error: unrecognized command line option "-ftree-loop-distribution"

after that I get this message with a "hanging" Terminal:

plugins/pxu_limiter.c:1: warning: this target does not support ‘-fsection-anchors’
plugins/pxu_limiter.c: In function ‘connectPortToPxu’:
plugins/pxu_limiter.c:198: warning: branch target register load optimization is not intended to be run twice

Only limiter.o has been created that far.

Redownload, btw I am not going to have the time to packaging aswell. So forget about the PPA.

[fluxBOX]

wow thanx Paradox Uncreated... going to try to compile these right now.