View previous topic :: View next topic |
Author |
Message |
wrc1944 Advocate
Joined: 15 Aug 2002 Posts: 3435 Location: Gainesville, Florida
|
Posted: Sat Aug 04, 2012 1:04 am Post subject: KLANG: A New Linux Audio System For The Kernel |
|
|
Just ran across this item.
KLANG: A New Linux Audio System For The Kernel
Anyone have more info or thoughts on this, and is it something to look forward to?
As one who's sought a really great Linux pro audio solution for a long time now, It sure sounds intriguing to me.
http://www.phoronix.com/scan.php?page=news_item&px=MTE1MDc
http://klang.eudyptula.org/ _________________ Main box- AsRock x370 Gaming K4
Ryzen 7 3700x, 3.6GHz, 16GB GSkill Flare DDR4 3200mhz
Samsung SATA 1000GB, Radeon HD R7 350 2GB DDR5
OpenRC Gentoo ~amd64 plasma, glibc-2.36-r7, gcc-13.2.1_p20230304
kernel-6.8.4 USE=experimental python3_11 |
|
Back to top |
|
|
PaulBredbury Watchman
Joined: 14 Jul 2005 Posts: 7310
|
Posted: Sat Aug 04, 2012 1:21 pm Post subject: |
|
|
Interesting discussion.
Until I can patch it into the kernel and play with it, I shall remain a reluctant ALSA fanboi. |
|
Back to top |
|
|
Ant P. Watchman
Joined: 18 Apr 2009 Posts: 6920
|
|
Back to top |
|
|
John R. Graham Administrator
Joined: 08 Mar 2005 Posts: 10590 Location: Somewhere over Atlanta, Georgia
|
Posted: Tue Aug 07, 2012 7:10 pm Post subject: |
|
|
Moved from Multimedia to Unsupported Software. Not in Portage so it fits better here.
- John _________________ I can confirm that I have received between 0 and 499 National Security Letters. |
|
Back to top |
|
|
VinzC Watchman
Joined: 17 Apr 2004 Posts: 5098 Location: Dark side of the mood
|
Posted: Thu Aug 09, 2012 3:43 pm Post subject: |
|
|
I followed some bits of the discussion between KLANG guy and Lennart Poettering at Linux Expo. I wouldn't pay much attention to it as it looks like the guy behind it (Wolfgang Draxinger) doesn't really have thorough understanding of real-time sound and Linux internals, as pointed to by Paul Davis (JACK developer) on the dedicated Phoronix discussion. _________________ Gentoo addict: tomorrow I quit, I promise!... Just one more emerge...
1739! |
|
Back to top |
|
|
krinn Watchman
Joined: 02 May 2003 Posts: 7470
|
Posted: Thu Aug 09, 2012 5:39 pm Post subject: |
|
|
Great discoveries were made by mistakes and hazard.
So, even if he fail, that guy might do something everyone with high understanding of sound and real-time linux will never do because it appears silly for them, and maybe discover something that is finally useful and a great step for them. |
|
Back to top |
|
|
VinzC Watchman
Joined: 17 Apr 2004 Posts: 5098 Location: Dark side of the mood
|
Posted: Thu Aug 09, 2012 7:03 pm Post subject: |
|
|
krinn wrote: | Great discoveries were made by mistakes and hazard.
So, even if he fail, that guy might do something everyone with high understanding of sound and real-time linux will never do because it appears silly for them, and maybe discover something that is finally useful and a great step for them. |
Maybe. But the major argument I retain is about using floating point for audio. It is currently used in major audio software and wanting to port that to kernel is the reason why it's a blocker: floating point in the (Linux) kernel is prohibited. So as far as I can see it, KLANG is likely not going to happen otherwise at least one of its goals (precision and/or speed for integer instruction uses more clocks than SSE, for instance) is going to fail. _________________ Gentoo addict: tomorrow I quit, I promise!... Just one more emerge...
1739! |
|
Back to top |
|
|
earthy Tux's lil' helper
Joined: 07 Jul 2006 Posts: 147
|
Posted: Tue Aug 21, 2012 3:36 pm Post subject: |
|
|
Plz no more audio engines for linux.Rather improve the existing ones. |
|
Back to top |
|
|
ulenrich Veteran
Joined: 10 Oct 2010 Posts: 1480
|
Posted: Tue Aug 21, 2012 4:12 pm Post subject: |
|
|
VinzC wrote: | the major argument I retain is about using floating point for audio. It is currently used in major audio software and wanting to port that to kernel is the reason why it's a blocker: floating point in the (Linux) kernel is prohibited. So as far as I can see it, KLANG is likely not going to happen otherwise at least one of its goals (precision and/or speed for integer instruction uses more clocks than SSE, for instance) is going to fail. |
He might not be the kernel expert, but on phoronix he made the point:
floating point is about large scales at the price of precision. You may want to mesure big stars and bacteria on a common scale, but did you once hear any sound above 30 thousand Hertz? |
|
Back to top |
|
|
VinzC Watchman
Joined: 17 Apr 2004 Posts: 5098 Location: Dark side of the mood
|
Posted: Wed Aug 22, 2012 8:27 am Post subject: |
|
|
VinzC wrote: | the major argument I retain is about using floating point for audio. It is currently used in major audio software and wanting to port that to kernel is the reason why it's a blocker: floating point in the (Linux) kernel is prohibited. So as far as I can see it, KLANG is likely not going to happen otherwise at least one of its goals (precision and/or speed for integer instruction uses more clocks than SSE, for instance) is going to fail. |
ulenrich wrote: | He might not be the kernel expert, but on phoronix he made the point:
floating point is about large scales at the price of precision. You may want to mesure big stars and bacteria on a common scale, but did you once hear any sound above 30 thousand Hertz? |
That's wrong, really: floating points are not meant for large scales but also for small numbers, between 0 and 1, for instance, which might result from signal processing (although I don't really know the internals of a DSP); floats do quite make sense. The point in using floats is that the precision (e.g. the number of significant digits or the mantissa) is constant. With integer numbers that precision decreases when the number is small, potentially leading to more noise with small values. But even then, most audio processing software are said to use floats. Also floating point operations are faster than integer operations so it makes no sense to use integers in the process. _________________ Gentoo addict: tomorrow I quit, I promise!... Just one more emerge...
1739! |
|
Back to top |
|
|
Herring42 Guru
Joined: 10 Mar 2004 Posts: 373 Location: Buckinghamshire
|
Posted: Wed Aug 22, 2012 10:31 am Post subject: |
|
|
Why are floats not allowed in kernel code? _________________ "The problem with quotes on the internet is that it is difficult
to determine whether or not they are genuine." -- Abraham Lincoln |
|
Back to top |
|
|
xaviermiller Bodhisattva
Joined: 23 Jul 2004 Posts: 8710 Location: ~Brussels - Belgique
|
Posted: Wed Aug 22, 2012 10:43 am Post subject: |
|
|
Hello,
Do the audio hardware support floating point buffers ? I don't think so.
So, there is at least two layers :
- the hardware support, with integer buffer
- the audio engine, as an application, which communicate with the kernel, and converting the floating point streams. _________________ Kind regards,
Xavier Miller |
|
Back to top |
|
|
Fitzcarraldo Advocate
Joined: 30 Aug 2008 Posts: 2038 Location: United Kingdom
|
|
Back to top |
|
|
PaulBredbury Watchman
Joined: 14 Jul 2005 Posts: 7310
|
Posted: Wed Aug 22, 2012 3:28 pm Post subject: |
|
|
Herring42 wrote: | Why are floats not allowed in kernel code? |
They are allowed when justifiable, but it's complicated - has different subtleties, and potential slowness of switching in and out of floating-point mode, depending on the architecture.
Code: | $ find /usr/src/linux-3.4.9/ -name \*.c | xargs grep kernel_fpu_begin\( | wc -l
35 |
|
|
Back to top |
|
|
Ant P. Watchman
Joined: 18 Apr 2009 Posts: 6920
|
Posted: Wed Aug 22, 2012 5:27 pm Post subject: |
|
|
I don't see why people are up in arms about this using fixed-point because it's in-kernel. How do you think ALSA already does it? |
|
Back to top |
|
|
VinzC Watchman
Joined: 17 Apr 2004 Posts: 5098 Location: Dark side of the mood
|
Posted: Thu Aug 23, 2012 7:00 am Post subject: |
|
|
VinzC wrote: | Also floating point operations are faster than integer operations |
My bad: I only checked the clock cycles of a few opcodes on Intel CPU. Might indeed not be the same on *every* supported architecture. _________________ Gentoo addict: tomorrow I quit, I promise!... Just one more emerge...
1739! |
|
Back to top |
|
|
|