Current Patches, Where to get them, How they affect speed/output

[XhmikosR]

x264 r1347 x86 ICC11 + GCC (http://www.mediafire.com/?sharekey=3f33c77c2cf9ce25ab1eab3e9fa335ca2f2536f3517cf1e6).
Patches used:
x264_win_zone_parse_fix_06.diff
x264-r1347-nal_hrd-pic_struct-v11.patch

I've built an ICC QaxSSSE3 version which is supposed to work on AMD CPUs also, and a QxSSSE3 which is for Intel CPUs only.

@Boulder: I never got able to use imk's package so here (http://www.mediafire.com/?y3l4ezedjtm) is my package which uses MasterNobody's ICC patch and some stuff from imk's package (libgpac and inttypes.h). It's much easier to compile (no need for Cygwin or MSYS). I've put a Readme inside with some directions.

[Boulder]

x264 r1347 x86 ICC11 + GCC (http://www.mediafire.com/?sharekey=3f33c77c2cf9ce25ab1eab3e9fa335ca2f2536f3517cf1e6).
Patches used:
x264_win_zone_parse_fix_06.diff
x264-r1347-nal_hrd-pic_struct-v11.patch

I've built an ICC QaxSSSE3 version which is supposed to work on AMD CPUs also, and a QxSSSE3 which is for Intel CPUs only.

@Boulder: I never got able to use imk's package so here (http://www.mediafire.com/?y3l4ezedjtm) is my package which uses MasterNobody's ICC patch and some stuff from imk's package (libgpac and inttypes.h). It's much easier to compile (no need for Cygwin or MSYS). I've put a Readme inside with some directions.
Thank you, I've got a vacation this week so I'm able to do some comparison on my E6850 regarding performance. My initial test on your previous ICC builds showed that the Qx version had a noticable edge over the x264.nl build.

[LoRd_MuldeR]

I've built an ICC QaxSSSE3 version which is supposed to work on AMD CPUs also

Only if they support SSE2 at least. If you use /QaxSSSE3, the Intel compiler will still default to /arch:SSE2 for the "base" code path.

[XhmikosR]

@L_M: I don't have an AMD CPU so I don't really care if it works.:) The package is there, everyone can build everything they want.

[LoRd_MuldeR]

Yeah, sure.

But if you did use "/QaxSSSE3 /arch:SSE" then you'd get a binary that runs on any x264-capable CPU (x264 with ASM enabled requires SSE) and still uses a special "optimized" path on suitable Intel CPU's.

[imk]

Yeah, sure.

But if you did use "/QaxSSSE3 /arch:SSE" then you'd get a binary that runs on any x264-capable CPU (x264 with ASM enabled requires SSE) and still uses a special "optimized" path on suitable Intel CPU's.

I did some experiments with this and came to the conclusion not to mess with it.

Using /Qax<level> will only generate a codepath for the specified SSE level. If you use /QaxSSSE3, then only processors with SSSE3 will benefit, and not those with SSE2 or SSE3. The documentation specifies that if you would like to target more than one, then specify it multiple times on the same switch, i.e. /QaxSSSE3,SSE3,SSE2. However, this will only work correctly if you do not profile the binary. If you do profile it, then problems arise and the binary may crash for people, or produce awkward results.

Since this only benefits the C functions and not the ASM (which will likely be used anyway), I found it best to just limit it to SSE2 processors via /QxSSE2. /arch defaults to SSE2 so no change is needed there for me.


Also, Windows build of x264 r1347 built with ICC:
x264-r1347M-imk-win.7z (http://imk.cx/pc/x264/x264-r1347M-imk-win.7z)
win_build_info.txt (http://imk.cx/pc/x264/win_build_info.txt)

Added in x264-r1347-nal_hrd-pic_struct-v11.patch

[LoRd_MuldeR]

I found it best to just limit it to SSE2 processors via /QxSSE2. /arch defaults to SSE2 so no change is needed there for me.

One remark: The /Qx and /arch options are mutually exclusive. Only /Qax can be used in addition to /arch.

But if you use the /Qx option, then /arch is replaced and the resulting binary will run on suitable Intel CPU's only. It will abort with error message on Non-Intel CPU's.

Only if neither /Qx is used nor /arch is set explicitly, the Intel compiler will default to /arch:SSE2, which runs on any SSE2 capable CPU.

[imk]

One remark: The /Qx and /arch options are mutually exclusive. Only /Qax can be used in addition to /arch.

But if you use the /Qx option, then /arch is replaced and the resulting binary will run on suitable Intel CPU's only. It will abort with error message on Non-Intel CPU's.

Only if neither /Qx is used nor /arch is set explicitly, the Intel compiler will default to /arch:SSE2, which runs on any SSE2 capable CPU.

Yeah, you're right. I knew that, too. I just got mixed up in what I was writing. Good for other people to know, though.

[olapanekala]

x264 r1347 x86 ICC11 + GCC (http://www.mediafire.com/?sharekey=3f33c77c2cf9ce25ab1eab3e9fa335ca2f2536f3517cf1e6).
Patches used:
x264_win_zone_parse_fix_06.diff
x264-r1347-nal_hrd-pic_struct-v11.patch

I've built an ICC QaxSSSE3 version which is supposed to work on AMD CPUs also, and a QxSSSE3 which is for Intel CPUs only.

@Boulder: I never got able to use imk's package so here (http://www.mediafire.com/?y3l4ezedjtm) is my package which uses MasterNobody's ICC patch and some stuff from imk's package (libgpac and inttypes.h). It's much easier to compile (no need for Cygwin or MSYS). I've put a Readme inside with some directions.

Congrats !!!! QxSSS3 version rocks..... :thanks:

[techouse]

x264_x64_r1352_unpatched (http://techouse.project357.com/builds/vanilla/x64/revision1352.7z)
GCC 4.4.2 20091019 (x86_64.core2.Komisar), unpatched, generic, fprofiled

________________________________________________________________________________

x264_x86_r1352_techouse (http://techouse.project357.com/builds/x86/x264_x86_r1352_techouse.7z) | INFO (http://techouse.project357.com/nfo/x86/x264_x86_r1352_techouse.txt)
GCC 4.4.2 20091019 (x86.core2.Komisar), fprofiled, -march=core2

x264_x64_r1352_techouse (http://techouse.project357.com/builds/x64/x264_x64_r1352_techouse.7z) | INFO (http://techouse.project357.com/nfo/x64/x264_x64_r1352_techouse.txt)
GCC 4.4.2 20091019 (x86_64.core2.Komisar), fprofiled, -march=core2

Patches used:

x264_win_zone_parse_fix_06.diff (http://kemuri9.net/dev/x264/patches/x264_win_zone_parse_fix_06.diff)
x264-r1352-nal_hrd-pic_struct-v11.patch (http://techouse.pastebin.com/fe24657)

[techouse]

x264_x64_r1353_unpatched (http://techouse.project357.com/builds/vanilla/x64/revision1353.7z)
GCC 4.4.2 20091019 (x86_64.core2.Komisar), unpatched, generic, fprofiled

________________________________________________________________________________

x264_x86_r1353_techouse (http://techouse.project357.com/builds/x86/x264_x86_r1353_techouse.7z) | INFO (http://techouse.project357.com/nfo/x86/x264_x86_r1353_techouse.txt)
GCC 4.4.2 20091019 (x86.core2.Komisar), fprofiled, -march=core2

x264_x64_r1353_techouse (http://techouse.project357.com/builds/x64/x264_x64_r1353_techouse.7z) | INFO (http://techouse.project357.com/nfo/x64/x264_x64_r1353_techouse.txt)
GCC 4.4.2 20091019 (x86_64.core2.Komisar), fprofiled, -march=core2

Patches used:

x264_win_zone_parse_fix_06.diff (http://kemuri9.net/dev/x264/patches/x264_win_zone_parse_fix_06.diff)
x264-r1352-nal_hrd-pic_struct-v11.patch (http://techouse.pastebin.com/fe24657)

[JEEB]

x264 r1353 64bit unpatched:
download (http://x264.fushizen.eu/files/revision1353/x264.exe) ; hash (http://x264.fushizen.eu/files/revision1353/x264.md5)

built on Nov 30 2009, gcc: 4.3.4 20090220 (prerelease) (x64.generic.Komisar)
fprofiled, otherwise defaults

________________________________________________________________________________

x264 r1353 32bit
download (http://x264.fushizen.eu/files/1353/x264.exe)

built on Nov 30 2009, gcc: 4.3.4 20090220 (prerelease) (x32.generic.Komisar)
fprofiled, -march=i686


x264 r1353 64bit
download (http://x264.fushizen.eu/files/1353_x64/x264.exe)

built on Nov 30 2009, gcc: 4.3.4 20090220 (prerelease) (x64.generic.Komisar)
fprofiled, -march=core2


patched with:

x264_win_zone_parse_fix_06.diff (http://kemuri9.net/dev/x264/patches/x264_win_zone_parse_fix_06.diff)
x264-r1352-nal_hrd-pic_struct-v11.patch (http://techouse.pastebin.com/fe24657)


git log: clicky (http://x264.fushizen.eu/files/git_log.txt)

A real Avisynth interface, oh jolly god this makes me yarr. Too bad I still haven't installed the 64bit Avisynth as it has far too few filters

[juGGaKNot]

x264_x86_r1353_techouse (http://techouse.project357.com/builds/x86/x264_x86_r1353_techouse.7z) | INFO (http://techouse.project357.com/nfo/x86/x264_x86_r1353_techouse.txt)
GCC 4.4.2 20091019 (x86.core2.Komisar), fprofiled, -march=core2


crashes on open

AVIsource("C:\x264\movie.avi")
Crop(0, height%2, -width%2, 0)
ConvertToYV12()

--preset veryslow --level 3.2 --ref 5 --min-keyint fps --keyint fpx10 --bframes 3 --merange 32 --aq-mode 2 --sar 1:1 --aud
works fine with 1152 from you.

[JEEB]

How does avs2yuv like that script?

The way of loading avs scripts has changed (to a way closer to what avs2yuv does) so it might just be crashing because of errors you didn't get earlier.

[JEEB]

x264 r1354 64bit unpatched:
download (http://x264.fushizen.eu/files/revision1354/x264.exe) ; hash (http://x264.fushizen.eu/files/revision1354/x264.md5)

built on Nov 30 2009, gcc: 4.3.4 20090220 (prerelease) (x64.generic.Komisar)
fprofiled, otherwise defaults

________________________________________________________________________________

x264 r1354 32bit
download (http://x264.fushizen.eu/files/1354/x264.exe)

built on Nov 30 2009, gcc: 4.3.4 20090220 (prerelease) (x32.generic.Komisar)
fprofiled, -march=i686


x264 r1354 64bit
download (http://x264.fushizen.eu/files/1354_x64/x264.exe)

built on Nov 30 2009, gcc: 4.3.4 20090220 (prerelease) (x64.generic.Komisar)
fprofiled, -march=core2


patched with:

x264_win_zone_parse_fix_06.diff (http://kemuri9.net/dev/x264/patches/x264_win_zone_parse_fix_06.diff)
x264-r1352-nal_hrd-pic_struct-v11.patch (http://techouse.pastebin.com/fe24657)


git log: clicky (http://x264.fushizen.eu/files/git_log.txt)

And fixes a 10L in MP4 output.

[juGGaKNot]

How does avs2yuv like that script?

I'll have to download and test.

LE :

avs2yuv works fine ( avs2yuv video.avs -o out.y4m )

Your 1354 4.3.4 works fine
techouse 1353 4.4.2 crashes

[techouse]

x264_x64_r1354_unpatched (http://techouse.project357.com/builds/vanilla/x64/revision1354.7z)
GCC 4.4.2 20091019 (x86_64.core2.Komisar), unpatched, generic, fprofiled

________________________________________________________________________________

x264_x86_r1354_techouse (http://techouse.project357.com/builds/x86/x264_x86_r1354_techouse.7z) | INFO (http://techouse.project357.com/nfo/x86/x264_x86_r1354_techouse.txt)
GCC 4.4.2 20091019 (x86.core2.Komisar), fprofiled, -march=core2

x264_x64_r1354_techouse (http://techouse.project357.com/builds/x64/x264_x64_r1354_techouse.7z) | INFO (http://techouse.project357.com/nfo/x64/x264_x64_r1354_techouse.txt)
GCC 4.4.2 20091019 (x86_64.core2.Komisar), fprofiled, -march=core2

Patches used:

x264_win_zone_parse_fix_06.diff (http://kemuri9.net/dev/x264/patches/x264_win_zone_parse_fix_06.diff)
x264-r1354-nal_hrd-pic_struct-v11.patch (http://techouse.pastebin.com/fe24657)

[[ReX]]

About XhmikosR's ICC project.

They removed input/avis.c and the new files are:
input/vfw.c [moved from input/avis.c with 80% similarity]
input/avs.c
extras/avisynth_c.h

So I obviously removed input/avis.c from Input in the x264 project.
I added extras/avisynth_c.h to the libx264 project in extras/extras_h.

Now, if I add only input/vfw.c or input/avs.c in Input in the x264 project, I can compile, but I don't get AVS input.
And if I add both of them, I can't compile (I get this: avs_input already defined).

What should I do to get it working?

[kemuri-_9]

$ ./configure --intel
Platform: X86/X86_64
System: WINDOWS
asm: yes
avs input: avs
mp4 output: yes
pthread: yes
debug: no
gprof: no
PIC: no
shared: no
visualize: no

$ make fprofiled

$ ./x264.exe --version
x264 0.79.1354M 636f98f
built on Nov 30 2009, intel: 1110 (20091012)


x264_icl_x86.exe (http://kemuri9.net/dev/x264/x264_icl_x86.exe)
x264_icl_x86_64.exe (http://kemuri9.net/dev/x264/x264_icl_x86.exe)

patches used:
x264_intel_support (http://kemuri9.net/dev/x264/patches/x264_intel_support.diff)
(x264_win_zone_parse_fix_06 is unnecessary here)

[juGGaKNot]

Since techouse pm'ed me about the errors i have made a test and i decided to post it here, if anyone else can test with the same setting ...

1352 techouse all work
1353 techouse all fail ( withous avs script also )
1354 techouse all fail ( withous avs script also )

jeebs all work

AVIsource("C:\x264\movie.avi")
Crop(0, height%2, -width%2, 0)
ConvertToYV12()

@echo off
x264_52.exe --pass 1 --preset veryslow --level 4.0 --bitrate 2000 --frames 300 --ref 5 --bframes 3 --merange 32 --aq-mode 2 --sar 1:1 --aud --output NUL x.avs
x264_52.exe --pass 2 --preset veryslow --level 4.0 --bitrate 2000 --frames 300 --ref 5 --bframes 3 --merange 32 --aq-mode 2 --sar 1:1 --aud --output x264_52.264 x.avs
echo.
echo.
x264_53.exe --pass 1 --preset veryslow --level 4.0 --bitrate 2000 --frames 300 --ref 5 --bframes 3 --merange 32 --aq-mode 2 --sar 1:1 --aud --output NUL x.avs
x264_53.exe --pass 2 --preset veryslow --level 4.0 --bitrate 2000 --frames 300 --ref 5 --bframes 3 --merange 32 --aq-mode 2 --sar 1:1 --aud --output x264_53.264 x.avs
echo.
echo.
x264_54.exe --pass 1 --preset veryslow --level 4.0 --bitrate 2000 --frames 300 --ref 5 --bframes 3 --merange 32 --aq-mode 2 --sar 1:1 --aud --output NUL x.avs
x264_54.exe --pass 2 --preset veryslow --level 4.0 --bitrate 2000 --frames 300 --ref 5 --bframes 3 --merange 32 --aq-mode 2 --sar 1:1 --aud --output x264_54.264 x.avs
echo.
echo.
x264_52.exe --pass 1 --preset veryslow --level 4.0 --bitrate 2000 --frames 300 --ref 5 --bframes 3 --merange 32 --aq-mode 2 --sar 1:1 --aud --output NUL x.avs
x264_52.exe --pass 2 --preset veryslow --level 4.0 --bitrate 2000 --frames 300 --ref 5 --bframes 3 --merange 32 --aq-mode 2 --sar 1:1 --aud --output x264_52.mp4 x.avs
echo.
echo.
x264_53.exe --pass 1 --preset veryslow --level 4.0 --bitrate 2000 --frames 300 --ref 5 --bframes 3 --merange 32 --aq-mode 2 --sar 1:1 --aud --output NUL x.avs
x264_53.exe --pass 2 --preset veryslow --level 4.0 --bitrate 2000 --frames 300 --ref 5 --bframes 3 --merange 32 --aq-mode 2 --sar 1:1 --aud --output x264_53.mp4 x.avs
echo.
echo.
x264_54.exe --pass 1 --preset veryslow --level 4.0 --bitrate 2000 --frames 300 --ref 5 --bframes 3 --merange 32 --aq-mode 2 --sar 1:1 --aud --output NUL x.avs
x264_54.exe --pass 2 --preset veryslow --level 4.0 --bitrate 2000 --frames 300 --ref 5 --bframes 3 --merange 32 --aq-mode 2 --sar 1:1 --aud --output x264_54.mp4 x.avs
pause

avis [info]: 1280x540 @ 30.00 fps (1472 frames)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Slow
x264 [info]: profile Main, level 4.0
x264 [info]: frame I:2 Avg QP:13.13 size:133654
x264 [info]: frame P:221 Avg QP:13.40 size: 11130
x264 [info]: frame B:77 Avg QP:19.00 size: 25881
x264 [info]: consecutive B-frames: 57.7% 16.8% 20.1% 5.4%
x264 [info]: mb I I16..4: 50.1% 0.0% 49.9%
x264 [info]: mb P I16..4: 12.6% 0.0% 0.0% P16..4: 15.1% 0.0% 0.0% 0.0% 0
.0% skip:72.4%
x264 [info]: mb B I16..4: 15.1% 0.0% 0.0% B16..8: 24.4% 0.0% 0.0% direct:
16.2% skip:44.3% L0:37.3% L1:36.8% BI:25.9%
x264 [info]: final ratefactor: 18.00
x264 [info]: direct mvs spatial:72.7% temporal:27.3%
x264 [info]: coded y,uvDC,uvAC intra: 51.5% 45.7% 25.5% inter: 12.6% 7.4% 2.2%
x264 [info]: i16 v,h,dc,p: 27% 26% 21% 25%
x264 [info]: i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 15% 28% 7% 5% 7% 7% 12% 5% 13%
x264 [info]: Weighted P-Frames: Y:7.7%
x264 [info]: kb/s:3775.83

encoded 300 frames, 16.67 fps, 3775.83 kb/s
avis [info]: 1280x540 @ 30.00 fps (1472 frames)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Slow
x264 [info]: profile High, level 4.0
x264 [info]: frame I:2 Avg QP:20.26 size: 65236
x264 [info]: frame P:221 Avg QP:23.71 size: 6298
x264 [info]: frame B:77 Avg QP:25.41 size: 13172
x264 [info]: consecutive B-frames: 57.7% 16.8% 20.1% 5.4%
x264 [info]: mb I I16..4: 50.0% 40.6% 9.4%
x264 [info]: mb P I16..4: 2.6% 8.2% 0.6% P16..4: 6.7% 2.2% 1.2% 0.0% 0
.0% skip:78.4%
x264 [info]: mb B I16..4: 1.9% 7.3% 0.9% B16..8: 26.2% 3.0% 2.4% direct:
8.2% skip:50.1% L0:47.5% L1:43.2% BI: 9.3%
x264 [info]: 8x8 transform intra:70.0% inter:86.2%
x264 [info]: direct mvs spatial:80.5% temporal:19.5%
x264 [info]: coded y,uvDC,uvAC intra: 64.8% 52.4% 11.5% inter: 10.9% 7.4% 0.3%
x264 [info]: i16 v,h,dc,p: 26% 14% 4% 56%
x264 [info]: i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 12% 13% 4% 7% 11% 11% 18% 9% 16%
x264 [info]: i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 10% 11% 3% 6% 14% 11% 21% 7% 16%
x264 [info]: Weighted P-Frames: Y:7.7%
x264 [info]: ref P L0: 50.0% 9.6% 27.1% 4.9% 5.8% 2.4% 0.4%
x264 [info]: ref B L0: 79.8% 10.4% 6.2% 3.5%
x264 [info]: kb/s:2029.29

encoded 300 frames, 3.10 fps, 2029.29 kb/s


avs [info]: 1280x540 @ 30.00 fps (1472 frames)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Slow
x264 [info]: profile Main, level 4.0
avs [info]: 1280x540 @ 30.00 fps (1472 frames)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Slow
x264 [info]: profile High, level 4.0


avs [info]: 1280x540 @ 30.00 fps (1472 frames)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Slow
x264 [info]: profile Main, level 4.0
avs [info]: 1280x540 @ 30.00 fps (1472 frames)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Slow
x264 [info]: profile High, level 4.0


avis [info]: 1280x540 @ 30.00 fps (1472 frames)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Slow
x264 [info]: profile Main, level 4.0
x264 [info]: frame I:2 Avg QP:13.13 size:133654
x264 [info]: frame P:221 Avg QP:13.40 size: 11130
x264 [info]: frame B:77 Avg QP:19.00 size: 25881
x264 [info]: consecutive B-frames: 57.7% 16.8% 20.1% 5.4%
x264 [info]: mb I I16..4: 50.1% 0.0% 49.9%
x264 [info]: mb P I16..4: 12.6% 0.0% 0.0% P16..4: 15.1% 0.0% 0.0% 0.0% 0
.0% skip:72.4%
x264 [info]: mb B I16..4: 15.1% 0.0% 0.0% B16..8: 24.4% 0.0% 0.0% direct:
16.2% skip:44.3% L0:37.3% L1:36.8% BI:25.9%
x264 [info]: final ratefactor: 18.00
x264 [info]: direct mvs spatial:72.7% temporal:27.3%
x264 [info]: coded y,uvDC,uvAC intra: 51.5% 45.7% 25.5% inter: 12.6% 7.4% 2.2%
x264 [info]: i16 v,h,dc,p: 27% 26% 21% 25%
x264 [info]: i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 15% 28% 7% 5% 7% 7% 12% 5% 13%
x264 [info]: Weighted P-Frames: Y:7.7%
x264 [info]: kb/s:3775.83

encoded 300 frames, 17.19 fps, 3775.83 kb/s
avis [info]: 1280x540 @ 30.00 fps (1472 frames)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Slow
x264 [info]: profile High, level 4.0
mp4 [info]: initial delay 1 (scale 30)
x264 [info]: frame I:2 Avg QP:20.26 size: 65236
x264 [info]: frame P:221 Avg QP:23.71 size: 6298
x264 [info]: frame B:77 Avg QP:25.41 size: 13172
x264 [info]: consecutive B-frames: 57.7% 16.8% 20.1% 5.4%
x264 [info]: mb I I16..4: 50.0% 40.6% 9.4%
x264 [info]: mb P I16..4: 2.6% 8.2% 0.6% P16..4: 6.7% 2.2% 1.2% 0.0% 0
.0% skip:78.4%
x264 [info]: mb B I16..4: 1.9% 7.3% 0.9% B16..8: 26.2% 3.0% 2.4% direct:
8.2% skip:50.1% L0:47.5% L1:43.2% BI: 9.3%
x264 [info]: 8x8 transform intra:70.0% inter:86.2%
x264 [info]: direct mvs spatial:80.5% temporal:19.5%
x264 [info]: coded y,uvDC,uvAC intra: 64.8% 52.4% 11.5% inter: 10.9% 7.4% 0.3%
x264 [info]: i16 v,h,dc,p: 26% 14% 4% 56%
x264 [info]: i8 v,h,dc,ddl,ddr,vr,hd,vl,hu: 12% 13% 4% 7% 11% 11% 18% 9% 16%
x264 [info]: i4 v,h,dc,ddl,ddr,vr,hd,vl,hu: 10% 11% 3% 6% 14% 11% 21% 7% 16%
x264 [info]: Weighted P-Frames: Y:7.7%
x264 [info]: ref P L0: 50.0% 9.6% 27.1% 4.9% 5.8% 2.4% 0.4%
x264 [info]: ref B L0: 79.8% 10.4% 6.2% 3.5%
x264 [info]: kb/s:2029.29

encoded 300 frames, 3.29 fps, 2029.29 kb/s


avs [info]: 1280x540 @ 30.00 fps (1472 frames)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Slow
x264 [info]: profile Main, level 4.0
avs [info]: 1280x540 @ 30.00 fps (1472 frames)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Slow
x264 [info]: profile High, level 4.0


avs [info]: 1280x540 @ 30.00 fps (1472 frames)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Slow
x264 [info]: profile Main, level 4.0
avs [info]: 1280x540 @ 30.00 fps (1472 frames)
x264 [info]: using SAR=1/1
x264 [info]: using cpu capabilities: MMX2 SSE2Slow
x264 [info]: profile High, level 4.0
mp4 [info]: initial delay 1 (scale 30)
Press any key to continue . . .

[techouse]

As I see you only tested the patched x86 builds. Could you please test the patched x64 and the unpatched x64?

[juGGaKNot]

windows xp x86, amd x2 4000+

I'll try unpatched, if that fails than maybe compiler problem, i'll try komisar's 4.4.2 also

LE : ahh its 64bit, i'll test komisar than k ?

[kemuri-_9]

windows xp x86, amd x2 4000+

I'll try unpatched, if that fails than maybe compiler problem, i'll try komisar's 4.4.2 also

LE : ahh its 64bit, i'll test komisar than k ?

the -march=core2 part is your enemy here.
it successfully prevented you from being able to use x264 for barely, if any speedup on ssse3 supported cpus.
as your cpu is does not have ssse3.

[juGGaKNot]

x264_x86_r1352_techouse (http://techouse.project357.com/builds/x86/x264_x86_r1352_techouse.7z) | INFO (http://techouse.project357.com/nfo/x86/x264_x86_r1352_techouse.txt)
GCC 4.4.2 20091019 (x86.core2.Komisar), fprofiled, -march=core2


Works

x264_x86_r1353_techouse (http://techouse.project357.com/builds/x86/x264_x86_r1353_techouse.7z) | INFO (http://techouse.project357.com/nfo/x86/x264_x86_r1353_techouse.txt)
GCC 4.4.2 20091019 (x86.core2.Komisar), fprofiled, -march=core2

Does not work, same for 1354.

[Brazil2]

x264_x86_r1354_techouse (http://techouse.project357.com/builds/x86/x264_x86_r1354_techouse.7z)
Works: x264 -o testAVS.mp4 alex.avi

Crashes: x264 --crf 22 -o testAVS.mp4 alex.avi

Looks like it crashes as soon as an additional parameter is used such as:
ref 4
subme 10
me umh

Windows XP 32 bits, Core2Duo.

[rack04]

x264_x86_r1354M

Built by rack04 on November 30, 2009, 1:20:09 PM CST
GCC 4.3.4 (x86.core2.Komisar)
--extra-cflags="-march=core2"
make fprofiled

Patched with:

x264-r1352-nal_hrd-pic_struct-v11.patch (http://techouse.pastebin.com/fe24657)
x264_win_zone_parse_fix_06.diff (http://kemuri9.net/dev/x264/patches/x264_win_zone_parse_fix_06.diff)


EDIT: I get crashes with this build so I have removed the download link.

[alexins]

x264 1354 x86 Unpatched - Works fine
x264 1354M x86 with x264-r1352-nal_hrd-pic_struct-v11.patch - crashes (my build, JEEB, rack04)
x264 1354M x86 with x264_hrd_pd_interlace.16_r1301.diff - Works fine

[techouse]

Did you use interlaced or progressive stuff to encode, since x264-r1352-nal_hrd-pic_struct-v11.patch may be problematic with interlaced videos.

[rack04]

Did you use interlaced or progressive stuff to encode, since x264-r1352-nal_hrd-pic_struct-v11.patch may be problematic with interlaced videos.

Progressive

[alexins]

Did you use interlaced or progressive stuff to encode, since x264-r1352-nal_hrd-pic_struct-v11.patch may be problematic with interlaced videos.

Progressive

[rack04]

x264 1354 x86 Unpatched - Works fine
x264 1354M x86 with x264_hrd_pd_interlace.16_r1301.diff - Works fine

Confirmed!

[techouse]

x264_x86_r1354_vanilla_techouse (http://techouse.project357.com/builds/vanilla/x86/x264_x86_r1354_vanilla_techouse.7z)
GCC 4.4.2 20091019 (x86.generic.Komisar), unpatched, generic, fprofiled

x264_x64_r1354_vanilla_techouse (http://techouse.project357.com/builds/vanilla/x64/x264_x64_r1354_vanilla_techouse.7z)
GCC 4.4.2 20091019 (x86_64.generic.Komisar), unpatched, generic, fprofiled

________________________________________________________________________________

x264_x86_r1354_v2_techouse (http://techouse.project357.com/builds/x86/x264_x86_r1354_v2_techouse.7z) | INFO (http://techouse.project357.com/nfo/x86/x264_x86_r1354_v2_techouse.txt)
GCC 4.4.2 20091019 (x86.generic.Komisar), generic, fprofiled

x264_x64_r1354_v2_techouse (http://techouse.project357.com/builds/x64/x264_x64_r1354_v2_techouse.7z) | INFO (http://techouse.project357.com/nfo/x64/x264_x64_r1354_v2_techouse.txt)
GCC 4.4.2 20091019 (x86_64.generic.Komisar), generic, fprofiled

Patches used:

x264_win_zone_parse_fix_06.diff (http://kemuri9.net/dev/x264/patches/x264_win_zone_parse_fix_06.diff)
x264-r1352-nal_hrd-pic_struct-v11.patch (http://techouse.pastebin.com/fe24657)

[laserfan]

x264 1354 x86 Unpatched - Works fine
x264 1354M x86 with x264-r1352-nal_hrd-pic_struct-v11.patch - crashes (my build, JEEB, rack04)
x264 1354M x86 with x264_hrd_pd_interlace.16_r1301.diff - Works fine
JEEB's 1354M x86 crashes for me also. His version 1352M is running OK.

[XhmikosR]

x264 x86 r1354 ICC + x264 x86 r1354 GCC unpatched profiled

http://www.mediafire.com/?sharekey=3f33c77c2cf9ce25ab1eab3e9fa335cafbdae63b01ff035e

[juGGaKNot]

Bla bla bla and bla + bla

v2 crashes, unpatched works.
Try building only with x264_win_zone_parse_fix_06.diff

unpatched WORKS, TESTED.
patched with x264_win_zone_parse_fix_06.diff AND x264-r1352-nal_hrd-pic_struct-v11.patch FAILS, TESTED.
pachted only with x264_win_zone_parse_fix_06.diff NOT TESTED.
patched with x264_win_zone_parse_fix_06.diff AND x264_hrd_pd_interlace.16_r1301.diff NOT TESTED.

JEEB's 1354M x86 crashes for me also. His version 1352M is running OK.

x264-r1352-nal_hrd-pic_struct-v11.patch is the same as it was @ 1352 so the error is from x264 revision 1353, why would it work with 1352 and not with 1353.

[JEEB]

Anyone who has gdb can try and test it with this (http://x264.fushizen.eu/files/1354/x264_1354_nhrd11_debug.exe) debug build. Has win_zone_parse_fix for mingw GCC and the nal-hrd patch in it.


$ patch -p1 < ../x264_win_zone_parse_fix_06.diff
patching file `encoder/ratecontrol.c'
Hunk #1 succeeded at 837 (offset 111 lines).

$ patch -p1 < ../x264-r1352-nal_hrd-pic_struct-v11.patch
patching file `common/common.c'
patching file `common/common.h'
patching file `common/frame.c'
patching file `common/frame.h'
patching file `common/osdep.h'
patching file `common/set.h'
patching file `encoder/encoder.c'
Hunk #9 succeeded at 2044 with fuzz 1.
Hunk #10 succeeded at 2174 (offset 6 lines).
Hunk #12 succeeded at 2248 (offset 6 lines).
patching file `encoder/ratecontrol.c'
Hunk #1 succeeded at 1568 (offset 23 lines).
patching file `encoder/ratecontrol.h'
patching file `encoder/set.c'
patching file `encoder/set.h'
patching file `x264.c'
Hunk #1 succeeded at 362 (offset 2 lines).
Hunk #3 succeeded at 546 (offset 2 lines).
patching file `x264.h'
Hunk #6 succeeded at 478 (offset 2 lines).

It could be the fuzzy thing, or then again -- something else.

[VFR maniac]

x264_x86_r1354_with_Trahald's_nal-hrd_16_and_ts-ctrl.rar (http://www.mediafire.com/?zrmytmldwqj)

gcc: 4.4.2-dw2
-march=core2
patch:
x264_hrd_pd_interlace.16_r1330.diff
x264_ts_ctrl_17_r1352.diff
x264_fix_extended_zones_r1330.diff

[LoRd_MuldeR]

If patches apply with "fuzz" that's not good! Offsets usually are okay, but "fuzz" means that the lines around the place where the patch used to apply have been changed and problems are likely.

So if you see the "succeeded at X with fuzz" message, you should carefully check the resulting code and, after a manual correction (if necessary), make a new DIFF file...

[JEEB]

Yeah, it has mostly been a'okay, though -- so I didn't pay much attention to it. Esp. since I wasn't the first one to build 1353 with it.

But certainly, will need checking. Too bad I'll have to leave for Sweden in an hour or so, which limits my options to actually debug it myself.

[CpT]

x264 x86 r1354 ICC + x264 x86 r1354 GCC unpatched profiled

http://www.mediafire.com/?sharekey=3f33c77c2cf9ce25ab1eab3e9fa335cafbdae63b01ff035e

TY - :thanks:

[kemuri-_9]

Anyone who has gdb can try and test it with this (http://x264.fushizen.eu/files/1354/x264_1354_nhrd11_debug.exe) debug build. Has win_zone_parse_fix for mingw GCC and the nal-hrd patch in it.

./x264_1354_nhrd11_debug.exe...(no debugging symbols found)...done
you kind of missed the whole debug symbol part there JEEB ^^

and everyone: post command lines that you get crashes/issue with,
there's nothing more of a pain than to get vague bug reports to try to track down and fix because you can't reproduce it.

anyways, Alex already stated on the mailing list that the v11 patch was broken since
r1348 when x264-r1352-nal_hrd-pic_struct-v11.patch posted to the ML and you should wait for v12

[techouse]

The fuzz is due to this.

This is the hunk in question:

x264-r1352-nal_hrd-pic_struct-v11.patch, from line 445 to line 473

@@ -1933,6 +2044,28 @@ int x264_encoder_encode( x264_t *h,
x264_frame_expand_border_mod16( h, fenc );

fenc->i_frame = h->frames.i_input++;
+ // fenc->i_pic_struct = 0;
+
+ if ( h->param.vui.b_timing_info_present )
+ {
+ if ( fenc->i_pic_struct == 0 ) // don't override if an app has set it before
+ {
+
+ if ( h->param.i_pulldown != X264_PULLDOWN_NONE )
+ {
+ fenc->i_pic_struct = h->param.pulldown.pattern[ fenc->i_frame % h->param.pulldown.mod];
+ }
+ else
+ {
+ if ( h->param.b_interlaced ) fenc->i_pic_struct = ( fenc->b_top_field_first ) ? 3 : 4;
+ }
+ }
+ }
+
+ // add number of "displayed" fields
+ // fixme: case of field frames handled incorrectly
+ fenc->i_field_cnt = h->i_disp_fields;
+ h->i_disp_fields += DeltaTfiDivisor[fenc->i_pic_struct];

if( h->frames.b_have_lowres )
x264_frame_init_lowres( h, fenc );


Now in encoder.c the last 2 lines, that the hunk refferences to are different:

ORIGINAL ENCODER.C, from line 1937 to line 1942

if( h->frames.b_have_lowres )
{
if( h->param.analyse.i_weighted_pred )
x264_weight_plane_analyse( h, fenc );
x264_frame_init_lowres( h, fenc );
}


The fuzzy patched output then looks like this:

PATCHED ENCODER.C, from line 2044 to line 2075

x264_frame_expand_border_mod16( h, fenc );

fenc->i_frame = h->frames.i_input++;
// fenc->i_pic_struct = 0;

if ( h->param.vui.b_timing_info_present )
{
if ( fenc->i_pic_struct == 0 ) // don't override if an app has set it before
{

if ( h->param.i_pulldown != X264_PULLDOWN_NONE )
{
fenc->i_pic_struct = h->param.pulldown.pattern[ fenc->i_frame % h->param.pulldown.mod];
}
else
{
if ( h->param.b_interlaced ) fenc->i_pic_struct = ( fenc->b_top_field_first ) ? 3 : 4;
}
}
}

// add number of "displayed" fields
// fixme: case of field frames handled incorrectly
fenc->i_field_cnt = h->i_disp_fields;
h->i_disp_fields += DeltaTfiDivisor[fenc->i_pic_struct];

if( h->frames.b_have_lowres )
{
if( h->param.analyse.i_weighted_pred )
x264_weight_plane_analyse( h, fenc );
x264_frame_init_lowres( h, fenc );
}

[Trahald]

Ive had fuzz put code in not even remotely right place. and you miss it because it compiles. but very wrong. definitely stop if you see that.

[kemuri-_9]

i'm not touching the v11 patch again,
I already sent Alex a revision of it before to fix a majority of the cosmetic issues it has.

wait for v12 or don't use it at all. it's as simple as that.

[techouse]

Yep...

[laserfan]

Alex already stated on the mailing list that the v11 patch was broken since r1348 when x264-r1352-nal_hrd-pic_struct-v11.patch posted to the ML...Sorry to interrupt, but to those of us non-programmers, not on the ML, but use --nal-hrd in our command lines, should we revert to a pre-r1348 build? How might the "broken" patch manifest itself if the x264 command gets to a completed output file (which looks good & plays well)?

[techouse]

Sorry to interrupt, but to those of us non-programmers, not on the ML, but use --nal-hrd in our command lines, should we revert to a pre-r1348 build? How might the "broken" patch manifest itself if the x264 command gets to a completed output file (which looks good & plays well)?
In that case just use r1352 till it's fixed.

[laserfan]

In that case just use r1352 till it's fixed.
:thanks:

[techouse]

x264_x86_r1354_v3_techouse (http://techouse.project357.com/builds/x86/x264_x86_r1354_v3_techouse.7z) | INFO (http://techouse.project357.com/nfo/x86/x264_x86_r1354_v3_techouse.txt)
GCC 4.4.2 20091019 (x86.generic.Komisar), generic, fprofiled

x264_x64_r1354_v3_techouse (http://techouse.project357.com/builds/x64/x264_x64_r1354_v3_techouse.7z) | INFO (http://techouse.project357.com/nfo/x64/x264_x64_r1354_v3_techouse.txt)
GCC 4.4.2 20091019 (x86_64.generic.Komisar), generic, fprofiled

Patches used:

x264_win_zone_parse_fix_06.diff (http://kemuri9.net/dev/x264/patches/x264_win_zone_parse_fix_06.diff)

[shon3i]

Sounds to me we back to old good HRD 16 :)