From owner-freebsd-hardware  Fri Feb 21 19:02:03 1997
Return-Path: <owner-hardware>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.5/8.8.5) id TAA15934
          for hardware-outgoing; Fri, 21 Feb 1997 19:02:03 -0800 (PST)
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.8.5/8.8.5) with ESMTP id TAA15926
          for <freebsd-hardware@freebsd.org>; Fri, 21 Feb 1997 19:02:00 -0800 (PST)
Received: (from bde@localhost) by godzilla.zeta.org.au (8.8.3/8.6.9) id NAA25747; Sat, 22 Feb 1997 13:59:10 +1100
Date: Sat, 22 Feb 1997 13:59:10 +1100
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199702220259.NAA25747@godzilla.zeta.org.au>
To: freebsd-hardware@freebsd.org, robsch@robkaos.ruhr.de
Subject: Re: Performance problem with P6-200
Sender: owner-hardware@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

>The following may be somewhat off topic, but I have some problems with my new
>P6NP5 motherboard from ASUS and a PPro-200 (256k). 
>When I run the C-linpack or other floating point stuff I can't get
>more than 14 MFLOPS (with double precision). With single precision
>I get 69 MFLOPS which is fine. Is it notmal that the P6 has such a lousy
>double precision performance?

Perhaps many memory variables are misaligned.  I forget whether alignment
is important for P5's or P6's or both.  Anyway, the only support for
double alignment is gcc's -malign-double, which is not the default and
not supported by the linker.

Bruce