Path: utzoo!mnetor!uunet!cbmvax!daveh From: daveh@cbmvax.UUCP (Dave Haynie) Newsgroups: comp.sys.amiga Subject: Re: A2000 and CPU slot Message-ID: <3521@cbmvax.UUCP> Date: 24 Mar 88 21:53:56 GMT References: <8803222026.AA06299@jade.berkeley.edu> Organization: Commodore Technology, West Chester, PA Lines: 73 in article <8803222026.AA06299@jade.berkeley.edu>, CRONEJP@UREGINA1.BITNET (Jonathan Crone) says: > What happens to the A2000's 68000 when the 68020 goes in? Good question! > I know that the logic on the 86pin connector provides for the coprocessor > board to steal the system from the 68000 (I think thats right isn't it??) It's not really stealing. The Coprocessor card asks the 68000 for the bus, via a signal called Bus Request. A bit later, the 68000 responds with a signal called Bus Grant, which is basically saying, "I'm cool, just wait for the end of this cycle". This is really the same way all 68000 DMA works. At the end of this cycle, the 68000 protocol says that you assert a signal called Bus Grant Acknowledge, and remove your Bus Request. This is something that the Coprocessor card could do, and is in fact what all the expansion bus DMA devices do. Such DMA device will be in charge of the system until it drops Bus Grant Acknowledge, at which point the 68000 or another DMA device may get the bus. The trick is that, while Bus Grant Acknowledge is active, no other card can get the bus. So if your CPU slot card took the bus that way, no other DMA could take place. On the CPU bus there's another acknowledge signal, called BOSS. If a CPU card activates that signal, it shuts only the 68000 off. The Bus Grant from the expansion bus now goes to the CPU card's Bus Grant line, and the CPU card's Bus Grant line is now an input to the expansion bus arbitation logic. So hard disks and things can request the bus from the 68020 or whatever you've got installed in there. > but does the 68020 run everything in the system or is the > 68000 run as a I/O processor like in some maxi-desktop systems that i've > seen running un*x??? With the A2620 card, we let you kick the 68000 back in on system reset, for testing software and things, but you can't dynamically switch it. The motherboard logic supports dynamic switching between the two, so it would be possible to design a card that let both CPUs run at the same time. That's more complicated, and we didn't really have room for it in the current 68020 board design. > does it not make sense to have the 68000 operating in this fashion???? > after all if you're to be running AMIUNX <------- My nomination for the > name of the new unix... It might make sense, if you could keep the 68020 on it's board most of the time. If it had to go offboard to talk to any motherboard resources very often, it could actually be a loss, since arbitrating between the two busses does take a finite amount of time. > it doesn't make sense to have the 68020 waiting on I/O > so why not have the 68000 as a I/O hound or something.... No one's waiting on I/O around here, bud! We're interrupt and DMA driven, so you only go service the I/O when there's something there to service, at least for the most part. While one tasks "waits" for I/O, the CPU's off doing something else, if there's anything else useful to do. > (I know that with the current Amiga O/S the 68000 isn't stuck waiting on > I/O but with two 68XXX 's under the hood the A2500 could sure make the > MacII look like a slug. That's OK, we intend to anyway. Though I will admit I think the Mac II has a neater case. > This of course is just my opinion.... > jpc > Jonathan P. Crone -- Dave Haynie "The B2000 Guy" Commodore-Amiga "The Crew That Never Rests" {ihnp4|uunet|rutgers}!cbmvax!daveh PLINK: D-DAVE H BIX: hazy "I can't relax, 'cause I'm a Boinger!"