Path: utzoo!utgpu!news-server.csri.toronto.edu!bonnie.concordia.ca!nnews!uunet!cbmvax!daveh From: daveh@cbmvax.commodore.com (Dave Haynie) Newsgroups: comp.sys.amiga.advocacy Subject: Re: Computer Architecture question -- Daye Haynie Message-ID: <21394@cbmvax.commodore.com> Date: 8 May 91 17:34:48 GMT References: Reply-To: daveh@cbmvax.commodore.com (Dave Haynie) Distribution: comp Organization: Commodore, West Chester, PA Lines: 53 In article melling@cs.psu.edu (Michael D Mellinger) writes: >Can you just drop a 68040 into any computer and expect it to be as >fast as a computer designed around the 68040? Just the '040? Not likely. But I don't think anyone talks about just "dropping" a 68040 chip, alone, into any system and expecting anything to happen. The bus signals for 68000/10, 68020/30, and 68040 are about as different as you can get; there isn't all that much family hardware resembalence between generations of any processor family. In the 680x0 family, the bus interface has been improving considerably about every other generation. Add-ins are, minimally, a 68040 and a small amount of "glue". That alone may not do much, or it may, depending on how much the designers of the system you're plugging these things into considered about the '040, or about such coprocessor devices in general. Now, if you're talking about a small board with the '040, gluse, and maybe a chunk of '040-specific memory, then you'll be real close. Take the accelerator boards for the A2000 for example. The A2000 coprocessor slot provided absolutely no support for any 68020->68040 type evice, just a 68000 interface the basic signals necessary for such an arbitrary device to share or own the A2000's motherboard bus. Later on, I came up with this 68030 thing, the A2630. Other companies have built even faster 68030 things to go in there. An A2000 equipped with an A2630 is only a little bit slower than an A3000, a machine designed around the 68030. Because the A2000 had this slot, you weren't forced to try to wedge some kind of 68030 board into a 68000 socket, which is absolutely NOT the way to upgrade the CPU in your system, though it can of course be done that way. The A3000 has a new 200 pin coprocessor slot, and we knew something about the 68040 when we came up with this system. That doesn't mean that you can simply plug a 68040 into some magic socket on the A3000 motherboard and all of sudden go real fast. It does mean that the A3000 was designed to allow accelerator boards to be added, and has some specific features which makes this easy. And even a few features (for example, burst writes to memory and Zorro III expansion), that aren't used by the 68030. >Now, can you just drop an 88K processor in a computer that is based on the >68K and expect it to work at all? No more than you can drop any arbitrary CPU into any other arbitary system and expect it to work at all. But, assuming you had some software for it, you could build an 88K coprocessor board for the A3000 that works like a 68040 coprocessor board, and if done right, it would be arbitratily close to a native 88K machine. Actually, 88K is cheating a little, because it's quite a bit like the 68040, hardware-wise. >-Mike -- Dave Haynie Commodore-Amiga (Amiga 3000) "The Crew That Never Rests" {uunet|pyramid|rutgers}!cbmvax!daveh PLINK: hazy BIX: hazy "That's me in the corner, that's me in the spotlight" -R.E.M.