home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!mcsun!uknet!pyrltd!mwuk!tony
- From: tony@microware.co.uk (Tony Mountifield)
- Newsgroups: comp.sys.atari.st
- Subject: Re: More Falcon news...
- Message-ID: <1295@mwuk.UUCP>
- Date: 23 Nov 92 13:01:39 GMT
- References: <2349658@overmind.citadel> <1992Nov19.213205.86@uoft02.utoledo.edu>
- Organization: Microware Systems (UK) Ltd., Winchester, UK.
- Lines: 54
-
- In article <1992Nov19.213205.86@uoft02.utoledo.edu> jsteiner@anwsun.phya.utoledo.edu (jason 'Think!' steiner) writes:
- > undermind!Steve_Johnson@overmind.mind.org writes:
- > > The Falcon030's 68030 has a 24-bit address/16-bit data bus. It
- > > get's a little confusing, though, because the COMBEL (the system
- > > controller) has a 32-bit bus to RAM and to the VIDEL (the video
- > > controller). The machine on the whole, though, isn't a 'TRUE' 32-
- > > bit machine (kinda like the Mac LC II/Performa 400). The PDS
- > > (processor-direct slot) is also 24-bit address/16-bit data. There
- > > is some hope, though, that a third party will be able to 'fix' this
- > > (there will, most definitely, be a 32-bit FAST RAM expansion for
- > > the Falcon030 that will allow more RAM, at least, from what I've
- > > heard).
- >
- > this is -really- strange. i was under the impression that all 68k
- > processors after the '020 were 32 bit all the way 'round. can someone
- > post a 'family tree' of the 68k series & what is what? going to
- > 24/16 for a 32 bit processor sounds like something Intel would do,
- > not Motorola.
- >
- > i realize that Motorola doesn't have a lot to do with the bus design,
- > but it seems to me that trying to fit a 32 bit external processor
- > on a 24/16 bus would be a lot more hairy than just making it 32 bits
- > in the first place.
-
- The 68020 and 68030 both have a feature called Dynamic Bus Sizing. This
- means that instead of a single DTACK to acknowledge the bus cycle, there
- is a pair of DSACK lines - DSACK0 and DSACK1. When the CPU wants to do a
- longword (32-bit) access, it places the data on all 32 data lines. The
- address decoding then determines whether the addressed device or memory
- can listen to all 32 or just 16 or 8. It then responds to the bus cycle
- with some combination of the DSACK lines to indicate whether 32, 16 or 8
- bits of data have been accepted. In the case of 32 the access is
- complete. Otherwise the data gets shifted over so the unaccepted part is
- on the part of the data bus that the accepted part was on, and another
- bus cycle is run. This keeps happening until all the data has been
- accepted (i.e. twice for a 16-bit port, and four times for an 8-bit
- port). Since the address decoding decides which DSACK lines will be used
- to acknowledge the cycle, it can have different addresses with different
- bus widths, e.g. 8-bit peripherals, 16-bit ROM, 32-bit RAM, etc.
-
- The 68008, 68000, 68010 and 68070 don't have this facility, and neither
- does the 68040 (don't ask me why!).
-
-
- Tony.
-
- --
- Tony Mountifield (G4CJO) | Microware Systems (UK) Ltd.
- -----------------------------------| Leylands Farm, Nobs Crook,
- Email: tony@microware.co.uk | Colden Common, WINCHESTER, SO21 1TH.
- (or: ...!uknet!mwuk!tony) | Tel: 0703 601990 Fax: 0703 601991
- ------------------------------------------------------------------------
- ** Any opinions are mine, not Microware's - but you knew that anyway. **
- ------------------------------------------------------------------------
-