X-pop3-spooler:	POP3MAIL 2.1.0 b 4 980420 -bs-
Message-Id:	<m0zFuzq-0000FVC@chkw386.ch.pwr.wroc.pl>
Date:	Mon, 7 Sep 98 06:41
From:	strasbur AT chkw386 DOT ch DOT pwr DOT wroc DOT pl (Krzysztof Strasburger)
To:	diep AT xs4all DOT nl, hanke AT nada DOT kth DOT se
Subject:	Re: pgcc and your remarks
Cc:	beastium-list AT Desk DOT nl
Sender:	Marc Lehmann <pcg AT goof DOT com>
Status:	RO
Lines:	25

Michael Hanke <hanke AT nada DOT kth DOT se> wrote
>On Sat, 5 Sep 1998, Vincent Diepeveen wrote:
>> Wow, that's weird. Why need it? Another problem in (p)gcc?
Yes, but don't forget - this is "bazaar" model. We all are finding
bugs and report them, and the compiler can be improved.
>> 
>> I've tried all gcc compilers and almost all options available.
>> NOT A SINGLE OPTION gives better code than -O2
Hmm... my program is better with -O5. About 5% in comparison with -O2.
The remarks concerning pattern optimizations should be anyway valuable
for developers.
>an improvement by 200-300% is possible. Since I am using my PC (AMD
>K5 based) for number crunching, the heart of all efforts is careful
>optimization of the whole system for carefully selected routines
>(kernels of routines).

>Krzystof, maybe, you can speed up your computations by using
>hand-optimized blas etc for Pentium. Or have a look at atlas or
>phipac.
I'm using already blas routines written in asembler, but matrix and vector
operations are not the time-critical code. The computation of matrix elements
requires approximately n^3*n! operations, where n is the number of electrons.
This part can be, fortunately, parallelized.

Krzysztof

- Raw text -