Short bio: Computer Scientist, FOSS supporter (read more)
Tux Machines (TM)-specific
SUPER is a project to optimize SuSE for speed and performance. In looking for an idea for my next article, I thought this project's lastest effort might make an interesting review. It's based on OpenSuSE's latest release, which is 10.0 RC1. Being concerned with speed and performance, this review could not help but compare SUPER's times with that of OpenSuSE's. However, there was another kink in the armor. I'd already compared OpenSuSE with Mandriva. I got to thinking, was the test fair? Did they use the same boot options? Were the same services and deamons being started at boot? Did they both use parallel=yes? So, what started out as a review of SUPER has morphed into a speed test. Is Super really faster than other two contenders?
SUPER stands for SUSE Performance Enhanced Release. To quote the site: "SUPER is a testbed for anything that makes your SUSE Linux OS perform in a different manner than intended in the more stable and enterprise oriented SUSE Linux tree. This can be a speed improvement geared towards the specialized desktop user or an additional feature that is considered too experimental or dangerous by the main SUSE Linux tree." So, in other words they are trying to make a good thing even better by compiling base packages in a certain order, optimized per architecture (for dvd version), utilizing RUN_PARALLEL=yes, limiting boot time file operations, pre-linking and caching files for boot-up with readahead.
SUPER is quite scaled down in comparison to the OpenSuSE SuSE Linux distribution. Applications are limited in number, and in fact the entire distro comes on one cd and takes about 1.3 gig of hard drive space. As a result the menus are minimized as is the selection of applications. The full rpm list as tested is located here. One could set up an SuSE ftp mirror in yast and install whatever else they'd like I speculate. But otherwise it is almost identical on the surface to SuSE Linux 10.0 rc1.
For the sake of this rough and unscientific experiment, I installed on Reiserfs, used the same boot options, matched the start up services as closely as possible, booted to run level 3 with vga=normal. The clock is started when the <enter> key is depressed.
Discrepancies between this and previous tests may be contributed to booted run level, frame buffer, and having opened applications previously that boot. Represented below are the averages of 3 time tests for each area, first open of each application.
|Mdv 2006rc1||OSS 10.0rc1||Super 10.0rc1||Boot||17||24||21.6|
|Mdv 2006rc1||OSS 10.0rc1||Super 10.0rc1||X&KDE||6.9.0,3.4.2||6.8.2, 3.4.2||6.8.2, 3.4.2|
As you can see, the optimizations in compiling methods of SUPER are having a significant performance increase once the system is booted, whether that was the compile order of the base system or the flags used in the individual applications. However, Mandriva is still smokin' 'em on the boot and shutdown times. There is no clear winner here, as it will depend on personal preference. If you are the type that has 3 month uptimes, you aren't going to care about boot times as much. On the other hand if you are the type that leaves your applications open just about from boot to boot, yet has to reboot to another os often, then application start times might seem less important. Is SUPER superior? It's up to you to decide.