Re: rpmbuild speed P3 versus XP?

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Jesse Keating Wrote

{
You've got to ask 
yourself, why are you recompiling all the packages?  The ones that
really do 
benifit from athlon/i686 like opts. are already provided by Red Hat
compiled 
for this.

See the many many discussions we've already had on this list about this
issue.
}

I know about the many many discussions we've already had on this list
about the issue.  I think I started most of them (or perhaps it was the
Phoebe list).  ;-)  And again, a simple question, not intended to cause
a divide, gets tons of unrelated replies.  Ah well.  As I recall from
the Phoebe list, someone asked to see some cold, hard numbers.

I agree that really the low level stuff (kernel, GLIBC) should, at least
in theory, make the biggest difference.  My own benchmarks seem to prove
that.  As it turns out, the reason I'm recompiling everything is I am
suspicious as to whether the AMD chips are crippled in some way.

I'm performing some crude benchmarks.  The only performance benchmarks I
can think of are frames per second for screen savers and games.  Of
course, the main thing I'm trying to improve is program startup time. 
OO.o takes *drumroll* one minute and ten seconds to start on my laptop
in Red Hat.  It takes *drumroll* one and a half seconds to start in
Windows 98.  It's like that on all the computers here.  (Don't get me
started on prelinking, haha.)

The laptop runs acceptably well.  In fact, it runs quite nicely,
considering the awful speed of the hard drive.  On the other hand, my
brother's K6-2-550 had 512MB video RAM and a 7200 RPM 40GB hard drive. 
Using Red Hat was torturous.  Sometimes he'd click the GNOME menu button
and he'd be waiting so long (usually around 20 seconds) before anything
happened that he'd think he didn't actually click the button, so he'd
click it again.  Then, thinking he'd locked up his computer (I've since
shown him to check the hard drive light and system monitor applet.) he
would try opening programs from icons on his panels.  The system would
go into shock.

Sorcerer, although way way too torturous to install since it took two
weeks, would operate faster than Windows 3.11 on the same computer! 
Programs, even big ones like Netscape, Abiword (I couldn't figure out
how to get OO.o to install), and Quake 3 would actually start faster
than I could click on them.

Now, I understand that actual performance in Red Hat is very high.  In
fact, I would be willing to bet that Red Hat Linux has one of the
fastest servers.  Once a program started on my brother's computer, it
ran acceptably fast.  Programs compiled on Sorcerer ended up being much
smaller, and so they would start faster.  Everything ran acceptably
fast.  I didn't perform any benchmarks to see if performance were really
improved.

We're Linux desktop users here with some relatively minor server
applications.  We don't even play games in Windows.  So for us, being
able to serve files for a thousand simultaneous logins while performing
remote compiles is not necessary, but starting software quickly would be
nice.

(BTW, I'm not recommending Sorcerer.  It's quite a mess.)

Here are my results so far.  Rebuilding everything on the Laptop, only
one thing was improved.  Of all the weird, bizarre twists, it was
KDiskFree.  System startup averages perhaps a second faster.  Memory
consumption is staying exactly the same as it was before.  Mozilla,
OO.o, Abiword, XScreenSaver, Galeon, Konqueror, Gedit, and a slew of
slow starting programs are still exactly as slow as before.

On one install before compiling, I tried prelinking the way the release
notes recommended.  A funny thing happened.  My startup time went UP by
10 seconds.  Then I -u the prelink and my startup time went UP by
another 10 seconds.  The program startup times also did about the same
thing, as far as taking longer and longer.  I scrapped that installation
so that I had a more scientific test with compiling.

Last time I rebuilt everything for my desktop system, I was a more
carefree OS junkie without enough space, so I didn't stick with one
distro for more than a week, usually more like two days.  When I rebuilt
everything, my memory consumption (at startup) dropped from 35%ish to
25%ish.  Programs felt much faster.  I'm getting ready to start
benchmarking my system to see if my hunches were true.

My laptop definitely does not benefit from recompiling.  My desktop
might.  I think the difference has to do with the Laptop having an Intel
chip and the desktop having an AMD chip.

Here's a strange thing.  I'm using RH 8.0 all around.  I read that RH 9
and especially GNOME 2.2 (which I use) have much tighter code, but RH 9
is clearly much slower on the laptop than 8.0.  Everyone else seems to
be experiencing only better performance from Shrike.

Good thing my mail is going thru again...I've got my soap-box back.  ;-)




[Index of Archives]     [Fedora Users]     [Centos Users]     [Kernel Development]     [Red Hat Install]     [Red Hat Watch]     [Red Hat Development]     [Red Hat Phoebe Beta]     [Yosemite Forum]     [Fedora Discussion]     [Gimp]     [Stuff]     [Yosemite News]

  Powered by Linux