Phoronix: Major OS Showdown w/ Windows Tests Are Imminent
While this year we have already delivered a number of notable articles and benchmarks, right now we are on the heels of delivering our most extensive set of benchmarks yet. We are in the process of comparing the performance of Windows 7 x64, Ubuntu 10.04 LTS, and Mac OS X 10.6.3 (likely FreeBSD 8.0 and OpenSolaris 2010.03 too) across multiple systems. By early May the first of these numbers will be published...
I'm getting really excited because of this,...I have a baaaad,feeling that Linus T. was right with saying Linux is bloated and i can guess that Ubuntu will exploit that just fine,...and people will draw very bad conclusions about Linux in whole,...I just hope that they know that Linux is actually on their freaking PHONES,..and MSN and Yahoo and rest of the crowd runs it in cluding Apple !!!!!
Both nvidia and ATI Linux drivers have near identical 3D performance in Windows and Linux, bugs aside. WINE performance doesn't really count, since WINE is, well, WINE . Provided the most recent drivers for both nvidia and ATI are used using apps. that run natively on both operating systems (and no major bugs are present, which is a bigger concern for fglrx), I expect 3D performance in Linux vs. Windows to be similar (give or take a few frames).
My guess is that windows will win most 3d performance tests and will positively slaughter linux on 2d tests, provided (a) the proper ICDs are installed and (b) the tests utilize full hardware acceleration on both platforms (and not something half-assed such as the win32 port of GTK+).
Linux will likely win disk and threading benchmarks, although single-threaded performance should be close enough. Video performance should be identical or slightly better on Linux/nvidia but significantly worse on Linux/ati&intel. Browser performance will also be lower on Linux (both firefox and chromium perform worse there. Opera 10.52 is amazingly fast but it's still an alpha on Linux, so...)
Phoronix tests primarily on Ubuntu because that is what actually matters. (And they also test on Fedora and openSUSE when it makes sense).
Much as I like Arch, it is much less relevant in the grand scheme of things - and I'm saying that as a loyal Arch user. Besides, its rolling nature makes it very difficult to obtain meaningful, reproducible benchmarks (one day you get those results, next day they are no longer valid).
Linux will likely win disk and threading benchmarks,
I wouldn't be so confident about that considering that the latest versions of windows are stomping on linux in webserving according to some sites and especially the performance hit that is incurred with the ext4 filesystem default that has bee adopted by so many distros.