Fedora 13 vs. Ubuntu 10.04 LTS Benchmarks

[phoronix]

Phoronix: Fedora 13 vs. Ubuntu 10.04 LTS Benchmarks

Fedora 13 was released last week with a number of new features like Btrfs system rollback support and support for easily using the open-source 3D NVIDIA driver, but how does its performance compare to that of Ubuntu 10.04 LTS, which was released a month earlier? We have the benchmarks today from three different systems to tell you.

http://www.phoronix.com/vr.php?view=14986

[glasen]

Hi Michael,

Thanks for the test. I had basically the same results when i benchmarked Fedora 13 and Ubuntu 10.04 on my system (Dell Latitude D820, GMA950 graphics).

Mesa 7.8 delivers about 15% more FPS compared to Mesa 7.7. When using the Xorg-Edgers-PPA, the performance for both system is basically the same.

But :

The x264-test is a bit misleading, because it compiles its own version of x264. When using the binaries delivered by each system, the situation is very bad for Ubuntu 10.04 :

When using the system binaries i get about 15fps for Fedora 13 and (*sigh*) about 3fps for Ubuntu 10.04.

So, if anybody wants to use x264 on a daily basis, you have to compile your own version of x264. Otherwise you will not be very happy.

[Sacha]

Why is the system x264 so bad? Is it because it has no SSE optimisations? Do you compile for SSE4.2?

[deanjo]

Why is the system x264 so bad? Is it because it has no SSE optimisations? Do you compile for SSE4.2?

The configure detects the cpu's capability. If the CPU supported it, it would have been compiled.

[smitty3268]

I remember a similar bug a long time ago with Ubuntu's xvid package, caused by them miscompiling it and leaving out all the assembly optimizations. Probably the same thing again.

[sylware]

have you recompiled natively with gcc the code for each architecture? Or have you compared generic machine code?