compilation architecture

Michael Schwendt mschwendt.tmp0701.nospam at arcor.de
Tue Jan 15 14:19:25 UTC 2008


On Tue, 15 Jan 2008 14:07:30 +0100, Jakub 'Livio' Rusinek wrote:

> So what in your opinion should be made to make Fedora faster?

Have you performed any tests other than perception-based comparison of
Firefox start-times under two different [unspecified] configurations? Any
benchmark results? Is it just Firefox that starts more quickly? Or do you
see faster builds when compiling software, compressing archives, encoding
Ogg files, or FPS values in graphical rendering applications? Can you post
even simple "time command" based results which compare execution times of
several programs? First you say the compiler optimisation flags are
insufficient. Then, with the assistance from list subscribers, it is found
out that more likely the faster start-times are due to a different
preloading strategy.

It's not the first time somebody bumps into this list and claims that the
compiler flags are either "wrong", "should be for i686 not i386" or "don't
optimise enough". It happens once a year or so and is covered by the list
archives. Usually, there is interest in discussing it. Even experts
comment on such topics and give serious answers instead of treating the
original poster like a troll. Nobody would mind if they ignored topics
like this one. But what is missing in all these cases is a scientific
approach to analysing what you believe is a problem. Even the step where
you would collect data, from which to proceed, is missing.




More information about the devel mailing list