Re: /etc/disktab on vax

[David Laight <david%l8s.co.uk@localhost>]

On Tue, Jul 03, 2012 at 08:11:44PM +0200, Johnny Billquist wrote:
> 
> Assuming that the 8080 did one instruction per clock cycle... Which it 
> don't. If I remember right, it needs about the same number of cycles as 
> the Z80.

Hmmm... my brain must have bit-rotted on that one :-(

> It's not that the Z80 "needs" as 4x clock. It's just that the fastest 
> instructions takes four clock cycles, meaning the actual number of 
> instructions executed is (at most) 1/4 of the clock frequency.
> 
> The 6502, which was the popular contemporary CPU to compare against, did 
> do about one instruction per clock cycle, which is why there was 
> somewhat of an equivalence between a 4 MHz Z80 and a 1 MHz 6502.

You probably mean one bus cycle per clock cycle - be in an instruction
fetch or memory access.

IIRC the Z80 had one unused memory cycle per instruction - when it
did the dram refresh cycle.

The 6809 could do a memory access per clock [1] - and the access was always
in the second half of the cycle. Boards like the Dragon 32 interleaved
the video memory reads by doing them in the first half.
Running code from ROM it was possible to run an address-dependant clock
rate and fetch code at double speed.
[1] I can't remember if there was a dead cycle between fetch and exectute.

The 'fun' z80 instructions are the bit-page ones with IX/IY offset.
Byte order <bit-page-prefix><IX-prefix><IX-offset><opcode>,
some assemblers would get that wrong.

        David

-- 
David Laight: david%l8s.co.uk@localhost