blender benchmarks

[HenryCase]

As I think someone already said (Karlos?), the really interesting benchmarks won't roll in until we see Linux performance compared on X1000 vs. PPC Mac. That will be a fairer comparison, as we don't know how optimised current AmigaOS builds are for the X1000.

[HenryCase]

Linux is completely irrelevant

Of course Linux benchmarks are relevant. What they will show is the potential waiting to be unlocked in the X1000 hardware, or at least they will more accurately show this potential as the Linux builds have been maturing for longer.

Anyone with half a brain understands that early software builds for new hardware platforms are not the most optimised. Early adopters buy for the hardware, with the promise that the software that runs on it will improve. There is no reason to suspect this pattern won't be repeated with the X1000 too, as OS4 is still in active development, and X1000 is the flagship system of this OS.

So, bring on the Linux benchmarks, let's get a better picture of what the X1000 can do.

[HenryCase]

Well, Linux benchmarks are making only little sense. They will show the theoretical potential of the X1000.

Actually, they don't show the 'theoretical' potential at all, they show the potential as it exists now... if you run Linux.

Anyway, with regards to benchmarking, it's fairly simple. When it comes to measuring anything in a scientific way, the best bet is to remove as many extraneous variables as possible, to isolate that which you are trying to measure. The question then is, what are you trying to measure? If you want to compare the performance of X1000 vs PPC Mac hardware, then your best bet is to make the software you use for testing these machines as similar as possible, so that the relative strengths of the hardware being tested shines through. This is why Linux benchmarks are a fairer test.