Distro Speeds

[Gtrax]

Searching 'benchmarks' on this site shows clearly that replies/views ratio is .. well .. thin, and the actual questions are more about benchmarking particular applications.

It it reasonable to wish for a speed test for Distros that automatically takes into account the processor type, hardware speed, memory size advantage, kernel configuration, etc. ? maybe?

I suppose one should include a probability figure for getting them to work first time.

[techieMoe]

Searching 'benchmarks' on this site shows clearly that replies/views ratio is .. well .. thin, and the actual questions are more about benchmarking particular applications.

It it reasonable to wish for a speed test for Distros that automatically takes into account the processor type, hardware speed, memory size advantage, kernel configuration, etc. ? maybe?

I suppose one should include a probability figure for getting them to work first time.

I'm not sure I understand what you mean. All distributions of Linux use the same kernel and underlying utilities. The only real difference is the version number and the user software they include by default. The reason it's hard to give performance comparisons for an entire distribution is that performance relies on so many variables.

The default set of applications that are installed on a machine, as well as any daemons that are selected to run, will affect performance. As will hardware differences, obviously. So will the versions of libraries, the kernel, and individual applications.

Misconfigured video hardware can give the illusion of poor performance because the graphical interface may be slow or choppy. Any misconfigured device has the potential to impact your overall performance.

What kind of benchmark are you looking for?

[Gtrax]

Indulge me a little while I do the fantasy wish list of the maybe impossible.

I imagine a set of benchmark programs that anyone can try, coupled with a 'policeman' program to apply the handicapping and frequency scaling, depending on the hardware, and the distro. The idea is that even a mis-configured setup could be rated, (and hence the setup problem revealed) . Users could compare the numbers to those achieved by the gurus, from a given distro, with the hardware largely compensated for.

Speed impacts from hardware clocking, memory swapping and the like might be OK to auto-detect (Knoppix-style). So too can the kernel configuration choices.

In this way, the effect of all the running applications, being partly what comes with the standard distro, possibly with some user choices overhead added on, can be compared. I don't mean something silly like "Benchmark #3 goes way crap if I play Tux Racer while I'm waiting". I am thinking more of something like "Benchmark #3 using Debian Etch is worse than when using Mepis 7.0" Maybe this comparison could be made, even from different machines, because the hardware clocking differences are scaled to compensate.

I admit this is coffee-lounge pipe-dreaming. I never thought it through yet.