home *** CD-ROM | disk | FTP | other *** search
- Path: redstone.interpath.net!usenet
- From: jamie@jamie.interpath.net (Jim Cooper)
- Newsgroups: comp.sys.amiga.programmer
- Subject: Re: Ideas
- Date: 4 Jan 1996 14:56:26 GMT
- Organization: Interpath -- Providing Internet access to North Carolina
- Message-ID: <4cgpqq$c2g@redstone.interpath.net>
- References: <769_9512300249@genesplicer.org> <3658.6574T577T2708@summat.demon.co.uk>
- NNTP-Posting-Host: raleigh-17.interpath.net
- Mime-Version: 1.0
- Content-Type: text/plain; charset=iso-8859-1
- Content-Transfer-Encoding: 8bit
- X-NewsReader: Interworks_GRn 3.0b23 December 28, 1995
-
- In article <3658.6574T577T2708@summat.demon.co.uk> mike@summat.demon.co.uk (Mike Dodd) writes:
- > Mike staggered back from the pub and slurred to Joey McDonald:
- > >The Agnus chip controls the amount of ram the custom chips can
- > >access. Isn't there SOME way to "trick" the custom chips into using
- > >a section (a larger section) of fast ram as though it were chip
- > >memory? Kind of "retargeting" all calls for chip ram to fast ram?
- >
- > >I have always been told that it's a "hardware" thing.. but I still
- > >don't see why a clever programmer couldn't do this.
- >
- > It IS a hardware thing, and it isn't something that the cleverest programmer can
- > get around. The fact is that chip mem and fast mem are located on two separate
- > buses, fast mem is connected directly (-ish, via a multiplexer and refresh
- > circuitry) to the CPU, AGNUS is also connected to this bus, but 'gates' the bus
- > onto a separate bus which is subject to arbitrated access from other chips
- > (actually, AGNUS and other internal bits of AGNUS - Copper). There is no way in
- > software to physically re-link the fast memory onto this separate bus.
- >
- > The funny thing is that if Commodore had a bit more fore-sight (easy to say now)
- > then it would have cost very little to give a much larger chip-mem addressing
- > range (say, 16 Megs [convenient for 24bit addressing space]) - only problem
- > would be to provide a fast-memory system you'd have needed a processor with a
- > full 32-bit addressing range from the start (a full 68020, for example). The
- > problem now is one of compatibility. Still, by the end of the year we'll all be
- > PPC'd / CHRP'd, won't we ?
-
- "if Commodore had a bit more fore-sight..." Sorry, Mike. Don't mean to pick
- on you, but it gets tiresome reading the same old half-truths over, and over,
- and over, and over, and over, and over, and...
-
- In Fall 1991, Dave Haynie showed the A3000+ to the developers at the Denver
- DevCon. It had a 68030, AA chipset, DSP, room for 16Megs of CHIP RAM and 16
- Megs of FAST, all on the motherboard. It was supposed to be released in
- Spring 1992, and would be sold both as a whole new computer, and as a
- motherboard swap for current A3000 owners. The A3000+ was supposed to sell
- for about the same as the A3000.
-
- Then, before the end of 1991, Mehdi Ali brought in his new, hand-picked, Head
- of Engineering, Bill Sydnes (of PCjr fame - remember the "chiclet" keyboard?).
-
- Bill promptly put the A3000+ project "on hold," and had the engineers work on
- the wonderful new A600 instead, followed by the A4000...
-
- Hopefully, AT won't hire Mehdi Ali *or* Bill Sydnes, and we really will get a
- PPC machine in '97.
-
- --
- Jim Cooper | World: jamie@interpath.com | "I-Net 225 is my life... or at
- | BIX: jcooper | least it seems like it!" :-)
-
- Definition of Perfection: Dysfunctionally Challenged.
-