blender benchmarks

[itix]

What is it with your obsession for benchmark on X1000? All I needed to know is that its the fastest machine you can get AmigaOS4 on and if that is what i'm interested in and want to use then that is what I'm going to have to get. Plus hopefully drivers and support for the hardware should only get better over time.

Plus I'm really look forward to getting it, I haven't been this excited getting a computer since...well my last Amiga. i dont think any PC or Mac no matter how good they are and for the amount the x1000 cost me, I could of got some serious kit, could of given me the same feelings.

I think it is interesting to OS4 users. Pegasos II G4 is still good choice if you want performance. Pegasos II is not the fastest OS4 hardware anymore but close.

[itix]

Seems to me like people are trying to point out that their PowerMacs using MorphOS are faster than the fastest AmigaOS 4 machine out there.

Which is probably true. It's also true that any modern x86 out there will blow either system out of the water. To outsiders we're all crazy for arguing over whose platform is faster, when they're both far slower than what everybody else uses.

I think those benchmarks are perfectly demonstrating where PPC hardware is going today.

It's also true that for myself I don't care, and I value my AOS 4 machines much more than any PC. Power is not important to me, it's how it uses it - and if the X1000 offers me a way to use it in a way that no other machine does, then that's reason enough to get excited.

I am sure you wouldn't mind if X1K was winning by wide margin :-)

[itix]

I disagree. These benchmarks are a small part of the full story. They're maths-heavy (which is arguably the X1000's weak point against the G4 Macs), whereas in bandwidth-heavy apps it's a completely different story.

Or in other words -- when data no longer fits in L2 cache in G4 then 1682M is winner?

Then you have to account for the fact we're only use one core in the X1000 at the moment.

Using 2nd core wouldn't improve Lame benchmarks much.

PLUS it's unoptimised.

The OS or benchmarks? I could buy that it was using conservative memory settings but according to memory benchmarks it is not case.

[itix]

X1000: BENCHMARKs: VC: 82.849s (+a strangely high 6.697s for I/O, bringing it to 89.552s in total)
MacMini: BENCHMARKs: VC: 84.626s

This to decode a 69 seconds long clip.

Are my following conclusions about this test correct?

Conclusion 1: The X1000 is about on par(!) with a Mac Mini G4 when decoding x264 HD video.
Conclusion 2: The X1000 seems *unable*(!) to decode and play 1080p H264 clips. (This I didn't expect!)

Apparently he wasn't using AltiVec version. According to results posted by stevieu AltiVec enabled MPlayer benchmark it is 69 seconds.

[itix]

Sure. And what happens when they get tired of ARM? Or is that going to be as indisputably eternal as x86 seemed five years ago?

ARM is going forward on every fronts. Even DSPs and media processors include ARM core these days.