My only hesitation with that is the different versions of x264 and the lack of avisynth
My only hesitation with that is the different versions of x264 and the lack of avisynth
Interesting... you're right about that. If you wanna compile this version and test it out, I'd be willing to give it a look. To your question: I obtained this version from x264.nl... I believe they make the source available therein. It's version 0.55.663
Okay, I assume that means core version 55, subversion revision 663. There's not actually much in it - i was using "0.56.674", with today's latest version as "0.56.677"
building revision 663 isn't an issue though, if it'd make the tests more to your liking. there's even the possibility for compiling up binaries for distribution as an "official" linux benchmark
I think we'd have a tough time identifying a market for them... LINUX users are pretty rare in my experience; if you are going to compile it, please do the LINUX/Win32 comparison
@directhex - can you try it both ways? Would make a cool comparison I think.
can do. i'm also building against as few libraries as i can get away with (e.g. disabling as much functionality as possible) to save space and end-user woe, against a relatively old distro. in *theory*, the x264 and mplayer binaries i'm producing should run on anything vaguely recent
x264 r663 built for i386, at any rate. building mplayer now.
Okay, built - the benchmark will automatically run a 32-bit or 64-bit version depending on the host CPU (and can be overridden with the TARGETARCH environment variable).
My benchmark can be downloaded from http://apebox.org/wordpress/linux/32 - be gentle, it's a large download.
Here's output from my Pentium-M 1.6GHz above, with my Linux compile - remember, this has exactly the same version of x264 as the Windows benchmark, and uses the same compiler, so you're pretty much doing a 1:1 comparison:
Intel Pentium M 730, stock 133 x 12, i915GMS/i910GML, 4-4-4-11 @ 266MHz, Ubuntu GNU/Linux 7.10 i386:
Intel Pentium M 730, stock 133 x 12, i915GMS/i910GML, 4-4-4-11 @ 266MHz, Microsoft Windows Vista Business i386:Code:encoded 1771 frames, 20.92 fps, 1855.00 kb/s encoded 1771 frames, 20.93 fps, 1855.00 kb/s encoded 1771 frames, 20.93 fps, 1855.00 kb/s encoded 1771 frames, 20.97 fps, 1855.00 kb/s encoded 1771 frames, 20.87 fps, 1855.00 kb/s encoded 1771 frames, 4.91 fps, 1825.28 kb/s encoded 1771 frames, 4.91 fps, 1825.28 kb/s encoded 1771 frames, 4.92 fps, 1825.28 kb/s encoded 1771 frames, 4.91 fps, 1825.28 kb/s encoded 1771 frames, 4.91 fps, 1825.28 kb/s
Code:encoded 1749 frames, 20.65 fps, 1854.10 kb/s encoded 1749 frames, 20.53 fps, 1854.10 kb/s encoded 1749 frames, 20.71 fps, 1854.10 kb/s encoded 1749 frames, 20.72 fps, 1854.10 kb/s encoded 1749 frames, 20.73 fps, 1854.10 kb/s encoded 1749 frames, 4.99 fps, 1825.89 kb/s encoded 1749 frames, 5.04 fps, 1825.89 kb/s encoded 1749 frames, 5.04 fps, 1825.89 kb/s encoded 1749 frames, 5.04 fps, 1825.89 kb/s encoded 1749 frames, 5.03 fps, 1825.89 kb/s
Hang on, i missed all AMD optimizations from the mplayer compiles (Intel optimized only). Recompiling...
Edit: fixed. Benchmark on the web is ready for use. Pentium-M result above is fine (since it doesn't use any AMD optimizations, obviously). Currently doing a 3-way test on my desktop between xp32, linux32 and linux64
Well NOW it gets interesting:
AMD Athlon FX 55, stock 200 x 13, nForce 4, 3-4-4-8 @ 200MHz, Microsoft Windows XP Home:
AMD Athlon FX 55, stock 200 x 13, nForce 4, 3-4-4-8 @ 200MHz, Ubuntu GNU/Linux 7.04 AMD64 (32-bit benchmark):Code:encoded 1749 frames, 33.02 fps, 1854.10 kb/s encoded 1749 frames, 32.94 fps, 1854.10 kb/s encoded 1749 frames, 31.97 fps, 1854.10 kb/s encoded 1749 frames, 33.00 fps, 1854.10 kb/s encoded 1749 frames, 33.04 fps, 1854.10 kb/s encoded 1749 frames, 7.83 fps, 1825.89 kb/s encoded 1749 frames, 7.80 fps, 1825.89 kb/s encoded 1749 frames, 7.85 fps, 1825.89 kb/s encoded 1749 frames, 7.88 fps, 1825.89 kb/s encoded 1749 frames, 7.88 fps, 1825.89 kb/s
AMD Athlon FX 55, stock 200 x 13, nForce 4, 3-4-4-8 @ 200MHz, Ubuntu GNU/Linux 7.04 AMD64 (64-bit benchmark):Code:encoded 1771 frames, 34.84 fps, 1855.00 kb/s encoded 1771 frames, 35.35 fps, 1855.00 kb/s encoded 1771 frames, 35.64 fps, 1855.00 kb/s encoded 1771 frames, 35.68 fps, 1855.00 kb/s encoded 1771 frames, 35.35 fps, 1855.00 kb/s encoded 1771 frames, 8.35 fps, 1825.28 kb/s encoded 1771 frames, 8.36 fps, 1825.28 kb/s encoded 1771 frames, 8.36 fps, 1825.28 kb/s encoded 1771 frames, 8.35 fps, 1825.28 kb/s encoded 1771 frames, 8.34 fps, 1825.28 kb/s
Code:encoded 1771 frames, 39.98 fps, 1854.13 kb/s encoded 1771 frames, 40.83 fps, 1854.13 kb/s encoded 1771 frames, 41.22 fps, 1854.13 kb/s encoded 1771 frames, 41.16 fps, 1854.14 kb/s encoded 1771 frames, 40.91 fps, 1854.13 kb/s encoded 1771 frames, 9.61 fps, 1825.29 kb/s encoded 1771 frames, 9.63 fps, 1825.29 kb/s encoded 1771 frames, 9.62 fps, 1825.38 kb/s encoded 1771 frames, 9.63 fps, 1825.29 kb/s encoded 1771 frames, 9.62 fps, 1825.29 kb/s
requirements:
glibc 2.3.4+
e.g. Arch 0.7+, Fedore Core 4+, Debian 4.0+, SUSE 9.3+, RHEL 4+, Ubuntu 5.10+, Slackware 10.1+, Gentoo 2006.0+, PCLinuxOS 2007+, Knoppix 4.0.2+, Mandriva 2005+
instructions (these will also work with a live cd distro - nothing's stopping you from giving it a go):
1) download x264_benchmark.tar.bz2 from above
2) extract to any location, either using a GUI tool like "ark" (KDE) or "file-roller" (Gnome), or using a command line to extract to the current directory:
3) run the benchmark - output files will be placed in your current directoryCode:tar jxvf /path/to/x264_benchmark.tar.bz2
4) (optional) to run the 32-bit version on a 64-bit OS:Code:/path/to/linuxbench/runbench.sh
Code:TARGETARCH=i686 /path/to/linuxbench/runbench.sh
Windows'll need to wait until Monday for me to be in the office, but here are remotely tested Linux results from my office PC:
Intel Mobile Core 2 Duo T7400, stock 166 x 13, i945PM, 5-5-5-15 @ 333MHz, Ubuntu GNU/Linux 7.10 AMD64 (32-bit benchmark):
Intel Mobile Core 2 Duo T7400, stock 166 x 13, i945PM, 5-5-5-15 @ 333MHz, Ubuntu GNU/Linux 7.10 AMD64 (64-bit benchmark):Code:encoded 1771 frames, 50.56 fps, 1851.50 kb/s encoded 1771 frames, 51.09 fps, 1851.50 kb/s encoded 1771 frames, 51.33 fps, 1851.50 kb/s encoded 1771 frames, 51.33 fps, 1851.50 kb/s encoded 1771 frames, 51.34 fps, 1851.50 kb/s encoded 1771 frames, 12.81 fps, 1826.69 kb/s encoded 1771 frames, 12.84 fps, 1826.69 kb/s encoded 1771 frames, 12.96 fps, 1826.70 kb/s encoded 1771 frames, 12.83 fps, 1826.69 kb/s encoded 1771 frames, 12.84 fps, 1826.70 kb/s
Code:encoded 1771 frames, 58.53 fps, 1850.63 kb/s encoded 1771 frames, 58.49 fps, 1850.63 kb/s encoded 1771 frames, 58.51 fps, 1850.63 kb/s encoded 1771 frames, 58.35 fps, 1850.63 kb/s encoded 1771 frames, 58.73 fps, 1850.63 kb/s encoded 1771 frames, 15.04 fps, 1826.02 kb/s encoded 1771 frames, 14.97 fps, 1826.02 kb/s encoded 1771 frames, 14.97 fps, 1826.91 kb/s encoded 1771 frames, 15.01 fps, 1826.02 kb/s encoded 1771 frames, 14.98 fps, 1826.19 kb/s
Cool man, these are interesting results. I'm adding them to the table now, but have a question: what are the codenames for your various processors? The P-M is probably a Dothan. Is your FX-55 a Clawhammer or San Diego? Also, does the FX-55 use DDR or DDR2 memory?
Last edited by graysky; 21-09-2007 at 08:19 PM.
Did you compile a 64bit version of x264 for win64?
There are currently 1 users browsing this thread. (0 members and 1 guests)