[rescue] Advice on Octanes

[Kevin]

GCC may be a culprit for some of these speed
differences.  Most of my software is GNU/GPL stuff and
while i do attempt to compile some of it with MIPS Pro
(v7.2) it doesn't work out most of the time and i do
not yet have the requisite knowledge to fix it.

Anyone know what SGI uses to compile their freeware
stuff?

For my larger file operations, the HD subsystem is
going to give the linux box some advantage, SCSI160
adapter and drives.  I've got U2W drives in the Octane
but i'm betting the controller is only UW.

The linux box in question runs a very cut down version
of Slackware 9 and pretty tight 2.4.21 monolithic
kernel (all unused crap removed and only XFS 1.3 and
crypto API are added.)  

I very frequently run 4+ more tasks while encrypting
files with no problems, unless it's the craptacular IDE
based linux box at work, and then doing even two things
at a time that require moderate disk access makes the
machine unusable.

I will say this for the Octane though, other than the
xmame stuff (which i don't count), it may take a while
to complete some tasks but it never really hiccups or
causes perceivable slow downs in the UI or other apps
while it is doing so.  The Octane/IRIX is over all,
much more stable and polished than the Linux/x86
box and it's those qualities that makes it my
favorite, not sheer speed.  

/KRM

On Tue, 23 Sep 2003 23:00:33 -0400
"Patrick Giagnocavo +1.717.201.3366" <patrick at zill.net>
wrote:
> 
> I would not be surprised, when you consider that gcc
> on x86 is probably a lot better than gcc/mips.  And
> if you are using an old o32 build or something...
> 
> For me, I find myself liking IRIX a lot, but x86
> Linux and *BSDs are very lightweight.  
> 
> Think how much more work can be performed per cycle
> when your kernel does not have what, 4000 entry
> points for SMP safety.
> 
> How much faster X can be when you do not have overlay
> planes and lots of other fancy stuff - just 2D X.
> 
> Single task comparisons will always favor Linux or
> x86-BSDs.
> 
> Now, run that crypt and then do 3 other tasks and see
> what happens.