g77 in egcs: suboptimal results ?

[Toon Moene <toon@moene.indiv.nluug.nl>]

>  Note that mdbnch (at least the version I know) is in
>  double precision.  Thus other performance considerations
>  are typically overshadowed on ppro by the double alignment
>  problems.  See the g77 manual.

Agree, but if he used the same g77 options as documented on the  
mdbnch Web page, this is not the cause:  -malign-double is listed  
there.

At first I thought that the absence of my performance patches would  
go a long way to explaining the results, but on my machine they  
make only a 1 % difference for this code.

[ Note that I can't say anything about egcs yet:  It doesn't build  
on my machine - I just compared the results of mdbnch.f when  
compiled with:

g77 -O2 -fforce-addr
(494 seconds)

and

g77 -O2 -fforce-addr -fno-rerun-loop-opt -fno-reduce-all-givs  
-fno-move-all-movables
(500 seconds)
]

using g77-0.5.21-970811 + gcc-2.7.2.2

HTH,
Toon.