IPB

Welcome Guest ( Log In | Register )

2 Pages V   1 2 >  
Reply to this topicStart new topic
libvorbis 1.3.4
Dukers
post Jan 24 2014, 12:53
Post #1





Group: Members
Posts: 22
Joined: 29-July 07
Member No.: 45724



Monty blogged about 1.3.4.

http://xiphmont.livejournal.com/62844.html

Thanks to the Mozilla folks, encoder lib size is lower now.

New compiles at Rarewares? rolleyes.gif
Go to the top of the page
+Quote Post
john33
post Jan 24 2014, 15:06
Post #2


xcLame and OggDropXPd Developer


Group: Developer
Posts: 3760
Joined: 30-September 01
From: Bracknell, UK
Member No.: 111



There will be, in time, but since there are no functional changes, it will not be a top priority. wink.gif


--------------------
John
----------------------------------------------------------------
My compiles and utilities are at http://www.rarewares.org/
Go to the top of the page
+Quote Post
lithopsian
post Feb 27 2014, 21:35
Post #3





Group: Members
Posts: 158
Joined: 27-February 14
Member No.: 114718



I patched this and compiled. libvorbisfile and libvorbis were roughly the same size, while libvorbisenc (which was by far the biggest) was less than half the size. Should be half again on 64 bit architectures I think. Sounds the same to me, certainly nothing horrible happened. The patch is pretty big, although blindingly simple.
Go to the top of the page
+Quote Post
145dBSPL
post Feb 28 2014, 10:55
Post #4





Group: Members
Posts: 15
Joined: 13-October 12
From: Germany
Member No.: 103833



Hello everyone,

i was also trying to compile the oggenc.exe with a MinGW/GCC toolchain but had no success so far. #1 libogg, #2 libvorbis seem to build as they should but #3 vorbis-tools always end up in 'checking for vorbis...yes | checking whether OV_ECTL_COUPLING_SET is declared...no | configure: error: Vorbis >=1.3.0 required !'
I really would appreciate if there's someone out there who could point me to the right direction. I have to admit that i have only very limited programming experience but i got a few other binaries (generic lame build (btw. faster than the ICL one of rarewares); fftw; xvid; ..) compiled succesfully after reading all the INSTALLs and READMEs and searching the forums/web.

Thank you
Martin
Go to the top of the page
+Quote Post
lithopsian
post Feb 28 2014, 12:55
Post #5





Group: Members
Posts: 158
Joined: 27-February 14
Member No.: 114718



I don't know anything about MiniGW, but that is a new option introduced in 1.3.0. Due to this new option, your libvorbis and vorbis-tools both have to have a version greater than 1.3.0 in order to be compatible. That is what the error is saying and it looks like it thinks the libvorbis version is too low.
Go to the top of the page
+Quote Post
145dBSPL
post Mar 1 2014, 16:43
Post #6





Group: Members
Posts: 15
Joined: 13-October 12
From: Germany
Member No.: 103833



I was using all the latest packages from xiph.org, so vorbis should be v1.3.4. The path settings seem to be ok as the libogg*.dll and libvorbis*.dll were created succesfully. What i also discovered in the config.log is 'fatal error: vorbis/vorbisenc.h: No such file or directory' but this header is available everywhere in the packages or copied to the include folders. The vorbis-tools leave me puzzled...

If someone experienced a comparable issue i would be really happy to get further hints.

The reason why i try to compile a few things by myself is because i want to see a little more into this process and because i can optimize the binaries to my specific platform (bypassing the suboptimal performance of the ICL for non-Intel systems).

Martin
Go to the top of the page
+Quote Post
lamedude
post Mar 3 2014, 13:33
Post #7





Group: Members
Posts: 14
Joined: 2-January 12
Member No.: 96171



ICC on AMD is still usually faster than GCC/VC. I assume john33 compiles with just /arch:SSE2 so none of that if (CPUID=GenuineIntel) use fast code; else use slow code; dispatcher stuff applies.
Go to the top of the page
+Quote Post
145dBSPL
post Mar 4 2014, 13:52
Post #8





Group: Members
Posts: 15
Joined: 13-October 12
From: Germany
Member No.: 103833



The results of my testings show something different. Of course this is only valid for this specific task of audio encoding.
Generally the Intel compiler does not optimize equally as Intel also has state on their websites
(http://software.intel.com/en-us/articles/optimization-notice#opt-en).

As far as i remember the binaries from Rarewares by john33 are optimized also to a certain degree (as are mine).
Here are the results - partially quite historic systems i know, but nice to see how they cope with this:

AMD Richland APU (Piledriver arch.)
ICC 9.1 Rarewares: ~ 10,8
ICC 12.1 Rarewares: ~ 39,6
GCC 4.8.2 Generic SSE2: ~ 49,2
GCC 4.8.2 CPU specific: ~ 54,0

Intel PIII (Coppermine arch.)
ICC 9.1 Rarewares: ~ 4,5
ICC 12.1 Rarewares: ~ 8,0
GCC 4.8.2 Generic SSE: ~ 8,2
GCC 4.8.2 CPU specific: ~ 8,4

Intel Celeron M530 (Core arch.)
ICC 9.1 Rarewares: ~ 11,7
ICC 12.1 Rarewares: ~ 20,3
GCC 4.8.2 Generic SSE2: ~ 21,1
GCC 4.8.2 CPU specific: ~ 21,6

Intel Core i7 (Sandy Bridge arch.)
ICC 9.1 Rarewares: ~ 36,2
ICC 12.1 Rarewares: ~ 49,4
GCC 4.8.2 Generic SSE2: ~ 56,1
GCC 4.8.2 CPU specific: ~ 57,0

So my conclusion: GCC provides a more constistent performance across different architectures.
It can outperform the ICL clearly even with generic builds and especially on non-Intel platforms.
These testings led me to the try of compiling the vorbis-tools by myself and also to see remarkable speed gains.
But unfortunately i had no succes as written before...
Go to the top of the page
+Quote Post
eahm
post Mar 12 2014, 20:51
Post #9





Group: Members
Posts: 1029
Joined: 11-February 12
Member No.: 97076



No Windows binaries yet?
Go to the top of the page
+Quote Post
145dBSPL
post Mar 14 2014, 18:32
Post #10





Group: Members
Posts: 15
Joined: 13-October 12
From: Germany
Member No.: 103833



No - i had to surrender. Maybe i try to cross-compile from a Linux some day. Of course i am still curious
to see how Oggenc would behave when looking at my promising Lame results.
Go to the top of the page
+Quote Post
eahm
post Apr 27 2014, 20:21
Post #11





Group: Members
Posts: 1029
Joined: 11-February 12
Member No.: 97076



http://dropcanvas.com/sil25

Token from TAC 0.9.5 beta (from RareWares.org before it went down), I can't get it to work with foobar2000. Try people try and let me know.

Thanks.

This post has been edited by eahm: Apr 27 2014, 20:22
Go to the top of the page
+Quote Post
Brazil2
post Apr 27 2014, 21:03
Post #12





Group: Members
Posts: 143
Joined: 9-May 10
Member No.: 80499



QUOTE (eahm @ Apr 27 2014, 21:21) *
http://dropcanvas.com/sil25

Token from TAC 0.9.5 beta (from RareWares.org before it went down), I can't get it to work with foobar2000. Try people try and let me know.

Thanks.

I believe the culprit is the -Q switch: http://www.hydrogenaudio.org/forums/index....mp;#entry864337
Go to the top of the page
+Quote Post
eahm
post Apr 27 2014, 22:09
Post #13





Group: Members
Posts: 1029
Joined: 11-February 12
Member No.: 97076



QUOTE (Brazil2 @ Apr 27 2014, 13:03) *
I believe the culprit is the -Q switch: http://www.hydrogenaudio.org/forums/index....mp;#entry864337

It's not that sorry, install a portable foobar2000, copy that file and test it. It really takes 1min.

This post has been edited by eahm: Apr 27 2014, 22:15
Go to the top of the page
+Quote Post
Brazil2
post Apr 28 2014, 10:45
Post #14





Group: Members
Posts: 143
Joined: 9-May 10
Member No.: 80499



QUOTE (eahm @ Apr 27 2014, 23:09) *
It's not that sorry, install a portable foobar2000, copy that file and test it. It really takes 1min.

I already said I'm getting the same error as you with foobar.
But you're right, the culprit is not the -Q switch, I spoke too soon. It might be related to stdin because all others parameters are working fine in a command line. This needs further testing which I can't do at the moment.

BTW this also looks weird:
CODE
Writing library                          : libVorbis (Turpakäräjiin) (20140122 (Turpakäräjiin))
Go to the top of the page
+Quote Post
lvqcl
post Apr 28 2014, 15:26
Post #15





Group: Developer
Posts: 3325
Joined: 2-December 07
Member No.: 49183



It is possible to test stdin with command line:

CODE
oggenc2.exe -q 4 - -o out.ogg < input.wav

...and yes, it crashes.
Go to the top of the page
+Quote Post
eahm
post Apr 28 2014, 16:28
Post #16





Group: Members
Posts: 1029
Joined: 11-February 12
Member No.: 97076



QUOTE (Brazil2 @ Apr 28 2014, 02:45) *
BTW this also looks weird:
CODE
Writing library                          : libVorbis (Turpakäräjiin) (20140122 (Turpakäräjiin))

Exactly, I saw that too but ozok said it was from RareWares so I didn't even say anything.
Go to the top of the page
+Quote Post
lvqcl
post Apr 28 2014, 16:48
Post #17





Group: Developer
Posts: 3325
Joined: 2-December 07
Member No.: 49183



Vendor string for 1.3.3 is "Xiph.Org libVorbis I 20120203 (Omnipresent)", for 1.3.4 is "Xiph.Org libVorbis I 20140122 (Turpakrjiin)"
Go to the top of the page
+Quote Post
john33
post Apr 28 2014, 16:52
Post #18


xcLame and OggDropXPd Developer


Group: Developer
Posts: 3760
Joined: 30-September 01
From: Bracknell, UK
Member No.: 111



Directly from info.c:
CODE
#define ENCODE_VENDOR_STRING "Xiph.Org libVorbis I 20140122 (Turpakrjiin)"


--------------------
John
----------------------------------------------------------------
My compiles and utilities are at http://www.rarewares.org/
Go to the top of the page
+Quote Post
eahm
post May 10 2014, 17:56
Post #19





Group: Members
Posts: 1029
Joined: 11-February 12
Member No.: 97076



QUOTE (lvqcl @ Apr 28 2014, 07:26) *
It is possible to test stdin with command line:

CODE
oggenc2.exe -q 4 - -o out.ogg < input.wav

...and yes, it crashes.

Brazil2, lvqcl, thanks. Any way around it?
Go to the top of the page
+Quote Post
lvqcl
post May 10 2014, 18:41
Post #20





Group: Developer
Posts: 3325
Joined: 2-December 07
Member No.: 49183



QUOTE (eahm @ May 10 2014, 20:56) *
Any way around it?

a) use custom encoding settings such as "-s %r --ignorelength -Q -q4 %s -o %d"
b) use older oggenc
c) ask john33 to fix the bug...

This post has been edited by lvqcl: May 11 2014, 10:52
Go to the top of the page
+Quote Post
john33
post May 11 2014, 10:41
Post #21


xcLame and OggDropXPd Developer


Group: Developer
Posts: 3760
Joined: 30-September 01
From: Bracknell, UK
Member No.: 111



oggenc2 compiles updated at Rarewares to fix STDIN bug. smile.gif


--------------------
John
----------------------------------------------------------------
My compiles and utilities are at http://www.rarewares.org/
Go to the top of the page
+Quote Post
lvqcl
post May 11 2014, 14:06
Post #22





Group: Developer
Posts: 3325
Joined: 2-December 07
Member No.: 49183



aoTuV Beta6.03 (2014):

QUOTE
The latest aoTuV beta6.03 was unified with Xiph.Org's libvorbis1.3.4. The part related to sound quality is not changed from previous version.
Go to the top of the page
+Quote Post
eahm
post May 11 2014, 14:13
Post #23





Group: Members
Posts: 1029
Joined: 11-February 12
Member No.: 97076



QUOTE (lvqcl @ May 11 2014, 06:06) *
aoTuV Beta6.03 (2014):

QUOTE
The latest aoTuV beta6.03 was unified with Xiph.Org's libvorbis1.3.4. The part related to sound quality is not changed from previous version.


Wow from 3.6MB to 750kB.

This post has been edited by eahm: May 11 2014, 14:13
Go to the top of the page
+Quote Post
lithopsian
post May 11 2014, 14:23
Post #24





Group: Members
Posts: 158
Joined: 27-February 14
Member No.: 114718



QUOTE (eahm @ May 11 2014, 14:13) *
QUOTE (lvqcl @ May 11 2014, 06:06) *
aoTuV Beta6.03 (2014):

QUOTE
The latest aoTuV beta6.03 was unified with Xiph.Org's libvorbis1.3.4. The part related to sound quality is not changed from previous version.


Wow from 3.6MB to 750kB.

Almost all down to data types being changed in the codebooks tables.
Go to the top of the page
+Quote Post
lvqcl
post May 11 2014, 14:33
Post #25





Group: Developer
Posts: 3325
Joined: 2-December 07
Member No.: 49183



Previous venc.exe was ~1.7 MB, so only 1MB size difference.
Go to the top of the page
+Quote Post

2 Pages V   1 2 >
Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 24th July 2014 - 13:19