Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!wuarchive!usc!apple!oliveb!amiga!cbmvax!daveh
From: daveh@cbmvax.commodore.com (Dave Haynie)
Newsgroups: comp.sys.amiga.tech
Subject: Re: speeding up the amiga and hardware hacking
Message-ID: <9233@cbmvax.commodore.com>
Date: 7 Jan 90 06:56:01 GMT
References: <90005.150000NETOPRBH@NCSUVM.BITNET>
Organization: Commodore, West Chester, PA
Lines: 40

in article <90005.150000NETOPRBH@NCSUVM.BITNET>, NETOPRBH@ncsuvm.ncsu.edu (Brandon Hill) says:

> 1)  Would it be possible to run the 68000 at a speed other than 7 mhz?
>     It seems like it would be easy to get, say, a 14 mhz clock from the
>     28 mhz master clock (dividing it in half) and a 16 mhz 68000 and
>     more or less double the processing speed.

This is possible.  The CMI PA (Processor Accelerator) uses a derived 14.3MHz clock
to give you a faster 68000; some 32 bit accelerators such as the A2620 do this as
well.  The best way to generate the fast clock depends on what you want to do.  A
free running crystal lets you pick the CPU speed, but requires extra care in youre
design, since everything must be synchronized to the motherboard clocks when the
CPU accesses the motherboard resources.  A 14MHz clock divided down from the 28MHz
clock will act similarly, since there's no guaranteed relationship between the
28MHz crystal and the Amiga chip clocks (especially when you Genlock!).  Building
a 14.3MHz clock from the C7M and CDAC clocks on the motherboard is a safe 
approach, yielding a clock that's reliably synched to the motherboard system.

>     There are many implications in doing this...i.e. the E clock might need
>     to be slowed down, as the 8520 timers should stay at the old
>     frequency.

You must drive everything on the motherboard at the 7.16MHz rates.  The E
clock can't change, AS* and DS* must still fall during S2 of the 7.16MHz 
68000 cycle, DTACK* must be sampled only on the falling edge of S4->S5,
data must be sampled on the S6->S7 edge.  In other words, when the fast CPU
is talking to any motherboard resource, it must be indistinguishable from a
7.16 MHz 68000.

>     I'm just mainly wondering if the rest of system would still communicate
>     with a faster processor (it must be possible, as the 020 and 030
>     accelerators run [asynchronously] faster)

Asynchronous processors add wait states and synch up to the 7.16MHz clock to
look like a 68000.

-- 
Dave Haynie Commodore-Amiga (Systems Engineering) "The Crew That Never Rests"
   {uunet|pyramid|rutgers}!cbmvax!daveh      PLINK: hazy     BIX: hazy
                    Too much of everything is just enough