home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!news.claremont.edu!ucivax!ofa123!Aric.Caley
- From: Aric.Caley@ofa123.fidonet.org
- Newsgroups: comp.sys.amiga.hardware
- Subject: Re: A4000 vs 486
- X-Sender: newtout 0.02 Nov 17 1992
- Message-ID: <n1127t@ofa123.fidonet.org>
- Date: 11 Dec 92 12:11:00
- Lines: 36
-
- > From: meek@csu.murdoch.edu.au (Lindsay Meek)
- > Message-ID: <1g85ajINN3n0@newsman.csu.murdoch.edu.au>
- > Newsgroups: comp.sys.amiga.hardware
- >
- > >And finaly, 100MegaBYTES per second? I rather doubt it. With a real 50Mhz
- > >486 (a REAL 50Mhz chip, not an internaly clock doubled one), thats only 2
- > >clock cycles per transfer - a mere 40ns. I don't see how a 486 is going to
- > >execute at least a couple of instructions, and do two memory accesses (one
- > >to read, one to write back) in a mere 2 cycles! Not to mention how fast
- > >the ram would have to be. No, I dont think so. 100 MegaBits, yeah.
- > >100MegaBytes, no.
- >
- > It is possible that they use a burst-style cycle to load a line in the
- > caches.
- > This would give you
- > your 40nS figure if it was done properly, and the memory supported nibble-
- > mode
- > or page-mode accesses. (Although it wouldn't be sustaining 40nS..more likely
- > a
- > peak performance).
- > Another way is to use a very-wide data path..but i don't think you'll find
- > that in a '486 BS (alias DX)
-
- Well, thats possible - but that would then limit that figure only to simple
- block oriented operations like blitting and filling. Even still, that
- would be pushing it. Memory that fast is pretty expensive.
-
- > Lindsay Meek
-
- -Dances With Coyotes - World's slowest programer-Given that God is infinate-
- -AKA Dances V2.0, Coyote, Critter, Magnet, Major-and that the universe is-
- -- Former Net-Lurker - YES, an AMIGA godammit! --also infinate, would you-
- --- Still limping along with AmigaOS 1.3 :( ---like a toasted tea cake?-
-
-
- --- Maximus 2.00
-