Gentoo Linux faster app-load than (Mandrake|Fedora)

[Owen Taylor]

On Thu, 2003-10-09 at 11:41, Mark Mielke wrote:
> On Thu, Oct 09, 2003 at 10:55:39AM -0400, Owen Taylor wrote:
> > On Thu, 2003-10-09 at 10:41, Mark Mielke wrote:
> > > On Thu, Oct 09, 2003 at 10:08:30AM -0400, Owen Taylor wrote:
> > > > What is somewhat amusing about the test is that the one major 
> > > > advantage of the Gentoo approach is that you *could* create two
> > > > otherwise identical systems except for compiler flags quite easily,
> > > > something that would be fairly hard with more conventional 
> > > > systems.
> > > > But the tester hasn't (as far as I can see) done this, and just
> > > > jumps to the conclusion that compiler flags are responsible for the
> > > > major speed differences.
> > > This perspective is a little cynical. The tester was quite clear that
> > > the results *can* be disputed. It goes further to point out that Mandrake
> > > may be running more services (i.e. the free RAM is not equivalent), and
> > > it also points to the kernel patch set (i.e. not just the compiler flags).
> > ...
> > I think you completely missed the point of my email; what my mail
> > was saying was:
> 
> I apologize if I read you wrong. I will point out, though, that your
> concluding paragraph made an incorrect conclusion. You seem to believe
> that the tester believes that compiler flags made all the difference.
> I do not read the article referred to the same way that you do.

The initial sentence of the "What can be gleaned from these results" is

 "For one, the default optimizations of Gentoo Linux 1.4 for Pentium III
appear to make a significant difference in "real world" application  
  load-time performance with Mozilla."

A claim that I don't see any substantiation of in the rest of the
article. What is documented is that in the test, Mozilla loaded a lot
faster, but no tests were done to separate out different possibilities
for what accounted for the speed difference. 

(It also should be noted that this page has undergone at least some
changes since I read it a few days ago... unlike an article, you 
can't really debate what a web page says.)

I don't really want to get into a big debate here. I think it's great
that people are doing desktop performance testing. I'd just like
to encourage the attitude that as soon as you do a test that
shows A is faster than B, the next step should be to do a test
to figure out *why* A is faster than B.

> >  - Gentoo would be a great platform for verifying that one way or 
> >    the other, so it would be great if those tests were extended
> >    to do an apples-versus-apples comparison with only compiler
> >    flags changed.
> 
> The purpose of the test wasn't to compare GCC optimizer settings. It was
> to compare Gentoo to Mandrake. Unless I am wrong, the reason Mandrake was
> chosen, is because Mandrake purports to be GCC optimized for the
> Pentium instruction set as well... I'm not sure what you would gain from
> the test that you couldn't gain from recompiling a few RedHat packages with
> GCC optimizer settings finely tuned.

Well, doing the tests on Gentoo would be more likely to convince Gentoo
users one way or the other :-). Also, the commonly heard Gentoo claim 
is that you get significant advantages by compiling the *entire system*
with architecture-specific optimization flags.

Regards,
						Owen