X-pop3-spooler: POP3MAIL 2.1.0 b 3 961213 -bs-
Delivered-To: pcg AT goof DOT com
Message-Id: <m0yI9zc-0001zaC@chkw386.ch.pwr.wroc.pl>
Date: Thu, 26 Mar 98 10:34 
From: strasbur AT chkw386 DOT ch DOT pwr DOT wroc DOT pl (Krzysztof Strasburger)
To: Hannes DOT Loeffler AT uibk DOT ac DOT at, hernando AT cnea DOT edu DOT ar
Subject: Re: pg77 speed
Cc: beastium-list AT Desk DOT nl
Sender: Marc Lehmann <pcg AT goof DOT com>
Status: RO
Content-Length: 612
Lines: 11

>> 2) Did you get the 40% improvement on another programs?

>We do MD simulations here with AMBER. I observed speed improvements of only
>some percents (maybe ~5% maximum) compared with g77 0.5.21.
The optimization efficiency is very code-dependent. I have my own program,
which runs with only small speed difference between pgcc and gcc-2.7.2.3
(heavy floating point calculations).
When I compiled GAMESS (quantum chemistry), I found a critical subroutine.
It ran almost 3x faster with pgcc than with gcc-2.7.2.3 on PPro.
The recompilation gave 25-250% speedup of GAMESS (task dependent).
Krzysztof Strasburger