Soon: Windows 7 vs. Ubuntu 10.04 vs. Mac OS X 10.6, Oh My!

[mo0n_sniper]

yay \m/

Having windows as a benchmark option would be great because windows proprietary drivers receive the most attention from vendors.

Can't wait for the Windows vs Linux tests

[disi]

How do benchmark the slow window management, slow application startup and shitty multitasking? Please include some tests, that are not pure hardware/driver related

[alpha_one_x86]

I have particular query for this test:
Read/write raw hdd
I/O over file system
AND the more important for me:
Copy of 10x1G file on the same hdd and same partition. (no on ssd for keep be in common hardware)
I thinks you will see great difference between all the Os.

[alpha_one_x86]

How do benchmark the slow window management, slow application startup and shitty multitasking? Please include some tests, that are not pure hardware/driver related

+1, for exemple you can track application start up of Qt/KDE application with as special soft, open and close lot of windows for track windows manager performance.

[kraftman]

$ CONFIG_DEBUG_INFO is even used in released kernels. They aren't toggled with during the development cycle.

Ok, thanks Michael, so it probably doesn't matter too much if release is final or not, because it seems debugging is turned on even on final releases.

[loonyphoenix]

Ok, thanks Michael, so it probably doesn't matter too much if release is final or not, because it seems debugging is turned on even on final releases.

I think he meant released KERNELS, not released distributions. I think all these symbols will be stripped by release.

[Michael]

I think he meant released KERNELS, not released distributions. I think all these symbols will be stripped by release.

No, that output is from the released Ubuntu 9.10 with stable updates on my main system.

[Apopas]

I'd love to see Ubuntu, windows, osx and Fedora to compare. With two Linux distributions at the same time, we can see where Linux relly lacks and where lacks because of bad configuration.

[deanjo]

I have particular query for this test:
Read/write raw hdd

What is the point of a raw read/write test? People don't use drives in raw.

[wizard69]

This should be very interesting! Suggestions follow.

Here are some suggestions:

1. Wait for 10.6.3 on the Mac if at all possible. The thought here is that this rev is heavily focused on drivers and is very near release.
2. Consider at least one other Linux distro. I've never really gotten into Ubuntu due to some poor design choices so consider adding Fedora to the test.
3. Drop the low resolution tests. This may be controversial as many will try to justify the tests for any number of reasons. I'd like to think that Phoronix is a little more forward look and realizes that the native resolutions on LCD monitors is rather high these days. I guess I'm one that sees data as being useful only when I'm likely to use it.
4. Oh for the Mac you really need more than one platform. I know the stuff doesn't grow on trees but at the very least an iMac would be nice.
5. This is likley not possilbe anytime soon but it wold be nice to be able to tailor a test or two to threading. The idea beign to exercise the native approaches to threading on each platform. In otherwords is GCD all it is cracked up to be. Obviously this is a huge programming hurdle but it would be nice to see how some of the approaches stack up when applied against a common task. The more I think about this the more it begins to look like a test suite of its own.
6. The behaviour of each OS while runing on a SSD would be intersting to see.

A lot of stuff but maybe a few can be considered. I'm primarily a Mac user and to a lesser extent right now a Linux user. Apples Aqua is certainly interesting but one can't deny it is demanding performance wise. At least it seems that way to me. It would be interesting to see a framework comparison further down the road with a focus on how framework over head impacts performance.

In any event keep up the good work.

Dave