Skip to main content ↑ ↓

Notice

If you are using a Hotmail or Outlook email address, please change it now, as Microsoft is rejecting all email from our service outright.
Topic: aoTuV Patches, Vorbis 1.3.5 and Lancer (Read 38038 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #25 –
OK, I've located the problem with the resampling and I have fixed and tested it. It will take a while to generate new compiles as all windows compiles of oggdropXPd and oggenc2 are affected. For those who are interested, it turned out to be simply a matter of not feeding the resampler with a sufficient number of samples for the 'best' quality resampler to 'do its stuff'!

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #26 –
The oggenc2 Lancer builds have been updated with the version number bumped to avoid confusion. I'll get to all other compiles as soon as I'm able.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #27 –
Quote from: john33 on
You're both right, they probably won't work on XP.
As there is no difference in the quality of the output, I'm not too sure that the effort involved in producing XP compatible compiles is worthwhile, particularly as I don't have an XP based system for testing.

Isn't it just a flag to enable in the build script ?

http://forum.videohelp.com/threads/357754-...l=1#post2362275

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #28 –
Excellent work john33, and a big thanks for the prompt fix. It is interesting how the bug came to be, as i remember oggenc to encode maybe a couple of samples at the beginning of the file,
irrespective of the "ignorelength" setting.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #29 –
Quote from: Brazil2 on
Quote from: john33 on
You're both right, they probably won't work on XP.
As there is no difference in the quality of the output, I'm not too sure that the effort involved in producing XP compatible compiles is worthwhile, particularly as I don't have an XP based system for testing.

Isn't it just a flag to enable in the build script ?

http://forum.videohelp.com/threads/357754-...l=1#post2362275

Different compiler, different build environment.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #30 –
It is in ICC15.

But if john33 is using ICC14 you have compile in VS2010 to get XP support.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #31 –
Quote from: edwardar on
These aren't working for me on Windows XP 32 bit - am I missing something, or is it finally time I upgrade!!

I get a "not a valid Win32 application" error.

All three builds of oggenc2 2.88 work fine, including resampling, on Windows XP SP2 or later if I change the MajorSubsystemVersion from 6 to 5. This header field is at 0140h in these particular files. Just overwrite 06000000 with 05000100. oggenc2 doesn't seem to require any Vista/Seven-specific system functions.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #32 –
Quote from: j7n on
This header field is at 0140h in these particular files. Just overwrite 06000000 with 05000100.

Nice find! It's working fine with this edit.
Thanks for the tip

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #33 –
Quote from: j7n on
All three builds of oggenc2 2.88 work fine, including resampling, on Windows XP SP2 or later if I change the MajorSubsystemVersion from 6 to 5. This header field is at 0140h in these particular files. Just overwrite 06000000 with 05000100. oggenc2 doesn't seem to require any Vista/Seven-specific system functions.

I have modified the 32 bit builds as indicated above and re-uploaded. Perhaps someone could kindly check that they do indeed work on XP SP2, or greater? They certainly still work with no impact on anything on Win 10 Pro 64 bit.

Thanks for the tip.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #34 –
Quote from: john33 on
Perhaps someone could kindly check that they do indeed work on XP SP2, or greater?

The oggenc2.88-aoTuVb6.03_2015*.zip are working fine on XP, however the oggenc2.88-aoTuVb6.03_2015-LancerSSE*.zip builds are strictly identical to the previous ones (which is confirmed using CloneSpy) hence still return the "not a valid Win32 application" error message.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #35 –
Hmmm, thanks. Let me check again!

Edit: Hopefully I've got that right, this time!!    Fresh uploads of the SSE, SSE2 and SSE3 compiles.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #36 –
Quote from: john33 on
Fresh uploads of the SSE, SSE2 and SSE3 compiles.

Perfect, all x86 builds are running fine on XP now
Thanks for your work!

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #37 –
Phew!! Got it right, finally.  Thanks for checking them out.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #38 –
It seems the aotuv source were updated to vorbis 1.3.5. I've tweaked the patches to work with the newest aotuv version, but while it compiles fine, make check fails with:
Code: [Select]
aotuv-b6.03_20110424-20150808-lancer-test/lib/.libs/libvorbis.so: undefined reference to `mdct_butterflies_backward'


I'm not sure if it's because I messed up the patches, or maybe the initial ones had this problem as well?

edit: nvm, seems I broke something somewhere, it's working now. Thanks for the patches!

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #39 –
Quote from: akapuma on
Hello john33,

will you update libvorbis.dll, too?

best regards

akapuma

All Lancer dlls have now been updated.  Should all work on XP SP2, or greater.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #40 –
Encoding a CD  in tmpfs on an AMD Phenom II X4 970@3.5GHz using caudec on arch linux x64, vorbis@q5:

Code: [Select]
caudec -c vorbis -O /tmp/vorbis-lancer *.wav
17.43 seconds (read: 1209.9 MB/s, write: 4.1 MB/s, rate: 200.0x)
17.44 seconds (read: 1253.2 MB/s, write: 4.1 MB/s, rate: 200.0x)
17.41 seconds (read: 1260.0 MB/s, write: 4.1 MB/s, rate: 200.3x)

WAV:                586.8 MiB    100.0%
Source:             586.8 MiB    100.0%    1411 kbps
Ogg Vorbis:          67.7 MiB     11.5%     162 kbps

caudec -c vorbis -O /tmp/vorbis-default *.wav
20.73 seconds (read: 1301.0 MB/s, write: 3.2 MB/s, rate: 168.2x)
20.69 seconds (read: 1279.6 MB/s, write: 3.2 MB/s, rate: 168.5x)
20.72 seconds (read: 1280.1 MB/s, write: 3.2 MB/s, rate: 168.3x)

WAV:                586.8 MiB    100.0%
Source:             586.8 MiB    100.0%    1411 kbps
Ogg Vorbis:          63.5 MiB     10.8%     152 kbps

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #41 –
Quote from: ChronoSphere on
It seems the aotuv source were updated to vorbis 1.3.5.

Thanks for hinting me at this!

The new official aoTuV sources are identical to my aoTuV-1.3.5 patch except for one additional null pointer check in aoyumi's sources. I'll post updated Lancer patches soon.

Edit: Done! See first post.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #42 –
again, thanks a lot for your efforts, enzo and john - really appreciated.

here is a short sse3 x64 foobar2000 speed comparison:

oggenc2.87-aoTuVb6.03-Lancer_ICC12.1 (lvqcl) 2012/02:

Total encoding time: 0:29.829, 180.62x realtime


oggenc2.88-aoTuVb6.03_2015-LancerSSE3_x64 (john33) 2015/09:

Total encoding time: 0:32.094, 167.87x realtime

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #43 –
Quote from: john33 on
All Lancer dlls have now been updated.  Should all work on XP SP2, or greater.


Thank you very much.

I tested the libvorbis.dll. The player MPC-HC shows "Writing library: Lancer 20110424 (UTC 2011-04-24)". Is this old date intended?

Best regards

akapuma

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #44 –
Quote from: akapuma on
Quote from: john33 on
All Lancer dlls have now been updated.  Should all work on XP SP2, or greater.


Thank you very much.

I tested the libvorbis.dll. The player MPC-HC shows "Writing library: Lancer 20110424 (UTC 2011-04-24)". Is this old date intended?

Best regards

akapuma

Assuming you are using the dll that was created on 10 September, yes, that is correct. The full 'ENCODE_VENDOR_STRING' is "BS; LancerMod(SSE2) (based on aoTuV [20110424][1.3.5])", or SSE/SSE3, but it looks as though MPC-HC is only picking part of the string. The aoTuV version is of the date shown. If you look at the dll in a hex editor, you will be able to find the string in full.
 

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #45 –
Is rarewares.org down for anyone else?  I'm getting a 404 on the domain.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #47 –
Hmmm, I'll try and find out what's going on.

Having just had a look via FileZilla, I see that our main folder is marked: rarewares.org_DISABLED_FOR_MALWARE_CONTACT_DREAMHOST_SUPPORT_cp. I've emailed Roberto to try to get it sorted.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #48 –
We're back. There was concern about a file that's been there for years and someone suddenly decided to raise an alarm!! I compiled the file in question, so there is definitely no real problem.

aoTuV Patches, Vorbis 1.3.5 and Lancer

Reply #49 –
Quote from: enzo on
Quote from: ChronoSphere on
It seems the aotuv source were updated to vorbis 1.3.5.

Thanks for hinting me at this!

The new official aoTuV sources are identical to my aoTuV-1.3.5 patch except for one additional null pointer check in aoyumi's sources. I'll post updated Lancer patches soon.

Edit: Done! See first post.


Thanks Enzo.  Under debian I built the code with the patches for SSE3 without a problem.  Encoding works like a charm.  But when decoding I get a segfault in the function ov_read_filter of vorbisfile. Could it be because the new function "ov_read_float2pcm" is not referenced ?

This is what I have:
strings libvorbisfile.so | grep ov_read
ov_read_float
ov_read_filter
ov_read

 
SimplePortal 1.0.0 RC1 © 2008-2021