OSX - Windows = 0 - 5 in OSX's ground...
WOW! Just WOW!
The results are hardly surprising:
- Nvidia provides the whole OpenGL stack on Windows, which is very well optimized. On Mac OS X, half of the stack is provided by Apple and it is safe to say that this causes the performance deficit.
- Intel provides the whole OpenGL stack on Windows, which is pretty awful (especially on older parts like the GMA950). It is safe to say that Apple's stack outperforms Intel's stack here.
- Intel's OSS stack is less optimized than the Intel/Windows and Apple/Mac OS X stacks. (Nvidia's OSS stack is even less optimized).
AMD/ATI results would be an interesting addition to the mix.
I still question these cross platform benchmarks. How do they help anything? It's not like you can compare the code of the Windows Intel driver with the OSS one. And Windows itself, as well as large portions of the iDictate-OS are closed. What's the point of this?
And most of the replies here show it too: they're apologetic, trying to suppress the fact Linux (yes, yes, Linux != Ubuntu, but Ubuntu is representative of Linux) got beat in all of those benchmarks. But that was a known fact long before those pointless benchmarks.
So, I've yet to see the point in these benchmarks. They're merely a show off into your face, losers - kind of thing.
With Vista you get about the same FPS as Windows 7. Anyway, who cares anymore about Vista? Move on dude.
Originally Posted by Adarion
Maybe the people who have paid few hundred bucks to buy them?
Originally Posted by Enrox
So you don't like those benchmarks because you don't like their results? Or should Michael only benchmark stuff that shows Linux in a positive light?
Originally Posted by susikala
I really don't get all the protest about the intel open drivers having to go against closed drivers in other OS's. Ya they take a pounding but isn't part of the whole idea about open development of drivers is to allow others to contribute and fix such issues? Why does intel even provide a open model if everybody still relies on them to fix and improve them?
Originally Posted by deanjo
Yeah, it's a shame that Gallium drivers for Intel are developed by VMWare guys in their spare time. LLVM-optimized vertex shaders would really help here.
Originally Posted by deanjo
Hm, I can't do anything about it if you try really hard to understand my opinions in a light you personally find fitting.
Originally Posted by BlackStar
I'm saying those benchmarks are a waste of time and money (-my- money, if I decide to donate). You don't need to be a genius to know games work smoothlier on Windows than on Linux, you can install two games on both. I can assure you that in any conceivable setup, with every conceivable driver, Windows will beat the crap out of Linux. Not just Ubuntu, but even your tuned-to-insanity -Ox Gentoo.
Rubbing it in people's faces is simply an assholish thing to do. Plus, you haven't answered my question. How can these benchmarks be used to help Linux become better? What wonderful conclusions or insights can you draw from them that would help make Linux suck less? Nada, I say, and I don't think you can come up with something better.
Let me quote Wikipedia, which is, if not the most reliable site around, still a site that tries to keep a natural point of view:
"Phoronix is a technology website that offers product reviews, Linux distribution screenshots, interviews, and news while maintaining a pure Linux orientation."
With emphasis on the last three words -- pure Linux orientation. It is one of the few sites that does that, that's why I visit it. If it's now going to start doing cross platform benchmarks with the majority of the tested platforms being proprietary, then I lose my reason to visit it.