Re: A ? for all you old time linux users
lameth wrote:
I remember reading that most experienced linux users believe that
programs run best if you download the source code and compile the code
on your own machine. Is this true, do programs you compile on your own
machine run best? Or is it just a matter of knowing the program and your
computer a little better for having gone through the process of
compilation?
Back when my machines didn't matter I did this mostly as an ego/geek
thing. IE: I'm geekier than thou since I compile everything myself!
However, now that I run production machines my values have changed
slightly. :)
FWIW, I used to go straight to the source tarballs. With all the
interdependancies out there now this is truly a nightmare. The need to
do a security update on Apache would have a nasty trickle down effect on
many of the mods I used, and compiled in statically since it was
1/1000th of 1% "faster".
Uh. No, now I like "apt-get update" and "apt-get upgrade". Stable
branch only, thank you very much.
That said, I'm considering doing some benchmarks with OpenSSL. I keep
hearing that _maybe_ compiling OpenSSL for you specific platform will
have Big Performance Yeilds(tm). Though actually putting that idea into
production will have to wait until my systems are so loaded that might
make a difference.
The only component that *will* be hugely impacted is the kernel.
However, the good Debian developers alread provide a nice assortment of
kernels for vitually every x86 compatible chip out there (and several
non-x86 archs as well).
So, after all that, I think compiling stuff yourself can be very
educational, and depending on what you're using your system for, perhaps
fun as well. But better? Well, I'm not so sure. The difference
between a 486/66 and a Pentium 100 was pretty huge. The difference
between an Pentium III 1Ghz and an Athlon XP 1800 is pretty small.
Just my $0.042. (Inflation, ya know.)
Chris
Reply to: