home *** CD-ROM | disk | FTP | other *** search
Wrap
Received: from Norway.EU.net (nic.eunet.no [193.71.1.2]) by nacm.com (8.6.10/8.6.9) with SMTP id NAA20341 for <executor@nacm.com>; Thu, 27 Apr 1995 13:49:42 -0700 Received: from troll.no by Norway.EU.net with UUCP id AA19759 (5.65c/IDA-1.4.4/EUnet/NO for nacm.com!executor); Thu, 27 Apr 1995 22:49:36 +0200 Date: Thu, 27 Apr 1995 20:16:09 +0200 (MET DST) From: Arnt Gulbrandsen <agulbra@troll.no> To: executor@nacm.com Subject: Re: 486 optmization In-Reply-To: <Pine.LNX.3.91.950427073305.2351A-100000@zeus.achilles.net> Message-Id: <Pine.LNX.3.91.950427200945.14728B-100000@popcorn.troll.no> Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-executor@nacm.com Precedence: bulk > How about a Pentium-specific Executor? Any idea on how much > speed-up? With the 'Pentium' GCC, you get about 25% speed increase. I immediately thought, shouldn't matter since executor compiles the Mac code itself, and that's the code that takes most of the time. But is that true? How much time does executor spend in its various modules (screen update, instruction emulation, routine compilation, routine execution)? Btw, the assembler output from gcc (and presumably djgcc) looks quite a bit more pentium-friendly if -fforce-mem is specified. OTOH there seems to be a bug in -fforce-mem in the latest gcc snapshot, so it might not work. --Arnt