lkml.org 
[lkml]   [1997]   [May]   [15]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Alpha 500 vs Bi Ppro.
On Thu, 15 May 1997, Chris Arguin wrote:

> > Ok, maybe the code generation on alpha is longer ?
>
> Compiling RISC code does take a lot longer, and GCC isn't exactly
> optomized for the Alpha anyway. Also, when you say "Bi pentium pro", do

gcc does still have a ways to go to optimal Alpha code. I'm sure
volunteers are needed - go for it <g>

> you mean you actually have two pentium pro processors? I ask simply
> because I've never seen it written that way. That would make a big
> difference as well. Get two alpha chips and the alpha would probably win.
> Or try something besides compiling code.

That doesn't make any sense. At some point you MUST execute compiled code,
whether you are executing the program itself or the interpreter... An
interpreter will almost always be slower, independant of platform.

> > We didn't test floating point yet, it's not a priority for us. Maybe
> > this is a point where Alpha can blast the Ppro ?
>
> If I remember correctly, at equal clock speeds the Alpha will beat a PPro
> at floating-point, and a PPro will beat the Alpha at integer opperations.
>
> Of course, on the floating point side the Alpha will also give correct
> answers and not crash ;)

The Linux math library (libm) is not optimized for the Alpha, and
therefore probably won't perform as fast as a DEC UNIX one. Correct me if
I'm wrong...

-- Elliot http://www.redhat.com/
May all your PUSHes be POPped.


\
 
 \ /
  Last update: 2005-03-22 13:39    [W:0.037 / U:7.320 seconds]
©2003-2018 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site