From owner-freebsd-hackers Sun Jan 18 05:37:23 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id FAA03521 for hackers-outgoing; Sun, 18 Jan 1998 05:37:23 -0800 (PST) (envelope-from owner-freebsd-hackers@FreeBSD.ORG) Received: from baloon.mimi.com (sjx-ca126-27.ix.netcom.com [207.92.177.219]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id FAA03517 for ; Sun, 18 Jan 1998 05:37:09 -0800 (PST) (envelope-from asami@vader.cs.berkeley.edu) Received: (from asami@localhost) by baloon.mimi.com (8.8.8/8.8.8) id FAA13330; Sun, 18 Jan 1998 05:37:01 -0800 (PST) (envelope-from asami) Date: Sun, 18 Jan 1998 05:37:01 -0800 (PST) Message-Id: <199801181337.FAA13330@baloon.mimi.com> To: ksmm@cybercom.net CC: freebsd-hackers@FreeBSD.ORG In-reply-to: (message from The Classiest Man Alive on Fri, 16 Jan 1998 22:38:02 -0500 (EST)) Subject: Re: what does MMX mean? From: asami@cs.berkeley.edu (Satoshi Asami) Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk * : The MMX registers are aliases of regular FP registers. (They did not * : want to add any new registers or flags -- so the OS doesn't need to be * : aware of new MMX chips and the applications can still use them.) So * : it won't help if you want to do a lot of FP operations intermixed with * : MMX operations. * * Is this true for the Pentium II as well? I believe so. The Intel designer who gave a talk here awhile ago said they didn't want old operating systems to break because of new chips and new applications. What this means is that since OS's only save/restore the registers it knows about upon a context switch, you can't add a new register without having the OS support it explicitly. Satoshi