From owner-freebsd-emulation  Wed Nov  6 22:27:08 1996
Return-Path: owner-emulation
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id WAA03036
          for emulation-outgoing; Wed, 6 Nov 1996 22:27:08 -0800 (PST)
Received: from hq.icb.chel.su (hq.icb.chel.su [193.125.10.33])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id WAA03030
          for <emulation@freebsd.org>; Wed, 6 Nov 1996 22:27:04 -0800 (PST)
Received: (babkin@localhost) by hq.icb.chel.su (8.7.5/8.6.5) id LAA05679; Thu, 7 Nov 1996 11:25:39 +0500 (ESK)
From: "Serge A. Babkin" <babkin@hq.icb.chel.su>
Message-Id: <199611070625.LAA05679@hq.icb.chel.su>
Subject: Re: New PC-Emu
To: luigi@labinfo.iet.unipi.it (Luigi Rizzo)
Date: Thu, 7 Nov 1996 11:25:39 +0500 (ESK)
Cc: msmith@atrad.adelaide.edu.au, luigi@iet.unipi.it, emulation@freebsd.org,
        nox@jelal.hb.north.de
In-Reply-To: <199611061512.QAA08343@labinfo.iet.unipi.it> from "Luigi Rizzo" at Nov 6, 96 04:12:07 pm
X-Mailer: ELM [version 2.4 PL23]
Content-Type: text
Sender: owner-emulation@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

It appears that the reason of problem is in the -DALIGNED_ACCESS. It
must be undefined for any machine that supports unaligned access
(i386 shurely does that). If it is defined any 2-byte memory references
are implemented as 2 1-byte memory references.

-SB
> 
> A few speed tests on pcemu1.91
> 
> COMPILATION FLAGS:
> 
> Config. A:
> 
> OPTIONS = -DBOOT1_44 -DALIGNED_ACCESS -DBIGCASE -DINLINE_FUNCTIONS
> CFLAGS  = -I$(XROOT)/include -O2 -pipe #-fomit-frame-pointer
> 
> Config B:
> OPTIONS = -DBOOT1_44 
> CFLAGS  = -I$(XROOT)/include -O
> 
> Runtime test #1:
> 
> 	cl -c progmain.c (559 lines C source resulting in 7KB .obj file)
> 
> 			Config. A	Config. B
> 
> Compile Time		1:02		0:27
> Run time for test #1	0:40		0:37
> stripped pcemu size	176128		114888
> 
> 	Config B wins in all cases!
> 
> It appears that either the compile options or the flags do not do much
> good! Actually, using INLINE_FUNCTIONS is probably harmful since some
> functions are very large, and they get included several times.
> Perhaps some benefit can be achieved by deciding which functions to
> inline case-by-case.
> 
> For the debugging releases, perhaps it is better to limit
> optimizations :)
> 
> 	Luigi
> ====================================================================
> Luigi Rizzo                     Dip. di Ingegneria dell'Informazione
> email: luigi@iet.unipi.it       Universita' di Pisa
> tel: +39-50-568533              via Diotisalvi 2, 56126 PISA (Italy)
> fax: +39-50-568522              http://www.iet.unipi.it/~luigi/
> ====================================================================
>