SUPER standard benchmark

From openSUSE

SUPER standard benchmark

This benchmarks are done on an IBM T42p with 1GByte of RAM and Pentium M 2000 openSUSE has been installed on hda1. All tests performed on the same hardware.

Test types

Each test line is based on the standard install (first testline) + one additional change described. The last testline is the cumulative change of running a system with all changes enabled.

NOTE: Timing is done with with "date" and 2 systems. One timing system and the T42p as the test system. I repeated the tests several times and apart from the harddisc value, which seems to change every time you run it, the other values are repeatable. It also is not exact science since they are handmeasured (250ms delay due to human error possible) and are probably only indicative of a trend and not a proof. But for the purpose of testing "real-life" applications under various circumstances the tests should be acceptable and certainly show the trend that certain things do improve more than others and that all changes together improve it the most apart from boot since more things need to happen for the overall system speed to increase. I hope that initng will shave off a bit here.

3d: glxgears
apps: startuptime of Openoffice writer (app1), firefox (app2) measured from activation to full display
and gimp filter Mosaic applied to KDE34.png background (app3) 
boot:  * Boottime measured from hitting enter in Grub to kdm logon screen appearing
       * Logintime, time measured from htting enter at the password prompt in kdm to the taskbar appearing properly.
harddisc: measured with hdparm -t /dev/hda and hdparm -T /dev/hda. This test is really not very useful as it 
changes again when you run it again under the same conditions.
There is as such a margin of error and the tests are not repeatable.
kernel: Compile the standard openSUSE kernel, time in minutes, Done in runlevel 1 to exclude any other influences.