AutoVAQ patch testing

[MuLTiTaSK]

32 bit thanks :) for me, the 64 bit workaround is actually slower!

try with vfw4x264.exe instead it's faster for me and others it also doesnt corrupt MPEG2 streams loaded with MPEG2Source("") like avs4x264.exe does;)

[Terranigma]

Sorry for being annoying, but could someone patch the newest build (r1416*) with the new AQ Experimental? I don't care being x86 or x64 :) (http://forum.doom9.org/showthread.php?p=1369567#post1369567)

I'd like to see this as well. :thanks:

[julius666]

Here you are :)

http://uploading.com/files/c834mfd1/x264.exe/

Toolchain:
gcc-core-4.4.3 (built by Komisar)

Patches used:
x264_AQ_experiments_v4.diff

Optimization:
--extra-cflags="-march=core2"
fprofiled VIDS="C:/git/station2_720p.yuv"

Platform: x86
System: MINGW
asm: yes
avs input: yes
lavf input: yes
ffms input: no
mp4 output: yes
pthread: yes
debug: no
gprof: no
PIC: no
shared: no
visualize: no

[Terranigma]

Optimization:
--extra-cflags="-march=core2"
fprofiled VIDS="C:/git/station2_720p.yuv"

Platform: x86
System: MINGW
asm: yes
avs input: yes
lavf input: yes
ffms input: no
mp4 output: yes
pthread: yes
debug: no
gprof: no
PIC: no
shared: no
visualize: no

Thanks, but your builds crashes for me. Perhaps you can compile another build without the optimizations i have highlighted?

[julius666]

Thanks, but your builds crashes for me. Perhaps you can compile another build without the optimizations i have highlighted?

Of course, a build withouth -march=core2 (and not fprofiled):
http://uploading.com/files/7d1c63a3/x264.exe/

[jpsdr]

Hello.
Thanks already for all the people who are making compiles.
Can someone nevertheless do the avq patch + nal_hrd beta 1.22 with 14xx (last one) ?

Thanks.

[Terranigma]

Of course, a build withouth -march=core2 (and not fprofiled):
http://uploading.com/files/7d1c63a3/x264.exe/

OK, it works perfect now, so it could be just one of the patches that causes it to crash on my system. If you noticed by now, there has been some additional changes (now rev. 1419), so if you wouldn't mind compiling with your normal configuration (with march=core2/fprofiled) + 3 Additional builds would be great. :o
1. One without the march=core2 patch (and not fprofiled).
2. One without the march=core2 patch (and fprofiled).
3 And one with the march=core2 patch (and not profiled).
?

Trying to figure out which one of the 2 patches (possibly both) crashes my system, which i'm suspecting is more than likely the "march=core2" patch.

:thanks: In advance.

[masterkivat]

Hello guys.
I learned last night how to compile using this guide (http://doom10.org/index.php?topic=26.msg1230#msg1230) and with some help from the user @julius666.
I'd like to know if someone could run some tests to see if my version was compiled without any kind of error.
Here we go:

http://www.megaupload.com/?d=QLDYZSQ2

Toolchain:
gcc-core-4.4.3 (built by Komisar)

Patches used:
x264_AQ_experiments_v4.diff

Optimization:
--extra-cflags="-march=core2" && fprofiled VIDS="C:/home/ICE_352x288_30_orig_02.yuv"

Platform: x86
System: MINGW
asm: yes
avs input: yes
lavf input: no
ffms input: no
mp4 output: yes
pthread: yes
debug: no
gprof: no
PIC: no
shared: no
visualize: no

Just a observation: since I'm still learning to do it with care, I didn't add lavf neither ffms suport. If someone could teach me, I'll appreciate.

[julius666]

OK, it works perfect now, so it could be just one of the patches that causes it to crash on my system. If you noticed by now, there has been some additional changes (now rev. 1419), so if you wouldn't mind compiling with your normal configuration (with march=core2/fprofiled) + 3 Additional builds would be great. :o
1. One without the march=core2 patch (and not fprofiled).
2. One without the march=core2 patch (and fprofiled).
3 And one with the march=core2 patch (and not profiled).
?

Trying to figure out which one of the 2 patches (possibly both) crashes my system, which i'm suspecting is more than likely the "march=core2" patch.

:thanks: In advance.

of course :)
as far as I know fprofiled mode shouldn't cause any trouble, but I could be wrong.
Anyway, I made two fprofiled builds, if neither of them works then I'll make the not fprofiled ones.

Oh, and a question: what CPU do you have?

rev1419:

Toolchain:
gcc-core-4.4.3 (built by Komisar)

Patches used:
x264_AQ_experiments_v4.diff

Platform: x86
System: MINGW
asm: yes
avs input: yes
lavf input: yes
ffms input: no
mp4 output: yes
pthread: yes
debug: no
gprof: no
PIC: no
shared: no
visualize: no

1. Optimization:
--extra-cflags="-march=core2"
fprofiled VIDS="C:/git/station2_720p.yuv"

http://uploading.com/files/9533a5fd/x264.aq4.core2.fporf.exe/

2. Optimization:
fprofiled VIDS="C:/git/station2_720p.yuv"

http://uploading.com/files/854ma677/x264.aq4.gen.fprof.exe/

(for some reason the one without core2 optimization was compiled without lavf input, because of that it is just 1.1 MB)

[Terranigma]

a question: what CPU do you have?


Link (http://forum.doom9.org/showthread.php?p=1364469#post1364469)


rev1419:

Toolchain:
gcc-core-4.4.3 (built by Komisar)

Patches used:
x264_AQ_experiments_v4.diff

Platform: x86
System: MINGW
asm: yes
avs input: yes
lavf input: yes
ffms input: no
mp4 output: yes
pthread: yes
debug: no
gprof: no
PIC: no
shared: no
visualize: no

1. Optimization:
--extra-cflags="-march=core2"
fprofiled VIDS="C:/git/station2_720p.yuv"

http://uploading.com/files/9533a5fd/x264.aq4.core2.fporf.exe/

2. Optimization:
fprofiled VIDS="C:/git/station2_720p.yuv"

http://uploading.com/files/854ma677/x264.aq4.gen.fprof.exe/

(for some reason the one without core2 optimization was compiled without lavf input, because of that it is just 1.1 MB)


Thanks, the first one crashed, but the second one works perfectly. The first one is obviously crashing because i'm not on a dual-core system. :)

[kemuri-_9]

Thanks, the first one crashed, but the second one works perfectly. The first one is obviously crashing because i'm not on a dual-core system. :)

and here's another reason people need to drop the -march=core2 compilation placebo.

[julius666]

and here's another reason people need to drop the -march=core2 compilation placebo.

It gives me 5-6% speedup on my core2.

[aBra]

It gives me 5-6% speedup on my core2.
No speedup on my core2 at all.

[nurbs]

Is the new --aq-mode 2 the same as --aq-mode 4 in the patched builds?

[komisar]

nurbs, yes

[nurbs]

Cool. No waiting for patched builds for me anymore. :thanks: