A Big Operating System Benchmark Comparison

[AdrenalineJunky]

i really don't think having half a billion linux distro's is a good idea, for one, it will take micheal forever for two it will just get in the way.

personally i think max 1 current community distro, and one stable/enterprise distro. as enterprise distro's tend to be a bit behind the curve, i think that would make for an interesting look at how things have progressed.

current community would most likely be ubuntu as thats whats usually benchmarked... though fedora/arch/mandriva/etc would work fine. personally i'd like to see arch as it represents very up to date developments.

[The WallAR]

Gentoo?
I know that you have to configure gentoo and etc. But from LiveCD i Think configuring is less dificult (and time consuming)


And Mandriva, DreamLinux and stock Debian

Cheers!

[Dragoran]

Waiting on someone to send me some new Apple hardware then...

You could also use a Linux distro with a non broken Intel driver,

[thefirstm]

Just make sure you also give 64-bit LInux a chance, otherwise M@c0$ will win again...

[kraftman]

Archlinux - it is a performance beast and always one of if not the most up to date linux distro with it's rolling release architecture. Not to mention that AUR already has phoronix text suite.

Oh, yeah. Arch please :> It will be 'slower' in some SQL tests, but just because of some change:

http://www.phoronix.com/forums/showpost.php?p=86992&postcount=3

Just make sure you also give 64-bit LInux a chance

Naturally 64-bit should be tested. Isn't OS X using 64-bit capabilities like someone mentioned in another thread?

[Apopas]

Naturally 64-bit should be tested. Isn't OS X using 64-bit capabilities like someone mentioned in another thread?

Yup, also the new OSX will be pure 64 bit as they say.
Also, if only Ubuntu is used, then the comparison will be between a new product (OSX) and a distro that will be 4 months old. Thus, to stay in the real world, two distros should be used.
i) Ubuntu 9.04 or preferably Fedora 11 which is newer both 64 bit (only if there is time and 32 bit as well)
ii) Arch or if possible Sabayon or gentoo so to be able to test our modern features and software in this modern hardware.

(Truth be told I don't really think it will be a big trouble to have a test of Arch as well along with Ubuntu. It worths the time if we want the benchmarks to be fair).

[L33F3R]

(Truth be told I don't really think it will be a big trouble to have a test of Arch as well along with Ubuntu. It worths the time if we want the benchmarks to be fair).

Just because you use arch doesnt make it an ideal testing platform. I use linux mint but you dont see me asking for that to be tested, i can already assure you its bloated. If the man had the time i would say go for it but he and hopefully others here have lives to attend .

[lordmozilla]

Could we get the latest Arch linux? Since its a nice clean linux it would show how current linux development is doing compared to other OSs.

[Kano]

32 bit is faster when the code has been hand optimized with asm. But usually the speed difference is not that important. More import is that you can playback fullscreen flash videos with opengl accelleration with nvidia or fglrx binary driver and on 64 bit you can not. On 64 bit that does not look good at all, only when you dl h264 flash movies and play em with mplayer using xv/vdpau (or opengl for fglrx).

[deanjo]

yes but you miss an point
'
Benchmarking is only for new hartware becourse you canot "not" buy your old hartware becourse of bad benchmark results!

you only can handel your buy in hartware befor you buy it and read benchmarkes as a shopping guid!

32bit linux is vor old hartware only!

so to make a big benchmark on 32 bit is useless!

No, you completely miss the point. 32-bit machines are out there in mass. Ranging from personal use, computer labs, servers, corporate workstations, etc and many of those places update machines when they absolutely need too. In such environments it is not uncommon to see those machines go even 7 years before replacement. Intel made 32-bit only mainstream cores well into 2007. The tests would be totally relevant.

Of course your SSE2 argument doesn't hold crap either as most of PTS is compiled from scratch and the compilers pick up on the capabilities of the chip meaning if the CPU supports SSE2 it will be compiled with such support. It's called runtime detection. Also SSE2 will mean diddley shit squat to a lot of the benchmarks as many don't even use SSE2 and are not coded to utilize it either.