@Seiya

But what is done here?

Different hardware was compared - OK this is nice
But also different tools / version/ tunings / with SIMD /without - were used.

We know that different LAME versions have different performance.
We know that different compile flags give significant different performance .
We know that with/without SIMD gives like 100% difference.

So what value have these results under these conditions?
One LAME binary could be 3 times faster than another used one.

_________________
APOLLO the new 68K : www.apollo-core.com

@BigGun

i know.
the best things to do is the same benchmark for alls.
the problem is that OS4, MOS & AROS software has different optimization.


Maybe the best test should be linux.
Sam460, X1000, X86, Sam440, AmigaOne with the same version of linux (or at least similar version or similar features) with the same version of software to test

Blender is another ambigous test becuase OS4 has old version, MOS has more modern and optimized version and no port for AROS-.

Linux could have a further importance: 3D acceleration via Gallium.
With linux you could compare 3D with OS4/MOS/AROS 3D.
Gallium vs TinyGL / Gallium vs Warp3D-MiniGL


Sam440ep-Flex is 800 Mhz
Sam460ex is 1150 Mhz
X1000 is 1800 Mhz
AmigaOne is 800/933/1000 Mhz
Pegasos 2 is 1000 Mhz
X86 best use old system with 800/900 and 1000 Mhz

in the past i tried to convince to partecipate, but without success..

Last edited by Seiya on 08-May-2015 at 11:08 AM.

_________________