Re: /etc/disktab on vax

[Dave McGuire <mcguire%neurotica.com@localhost>]

On 07/03/2012 07:52 PM, ..I'd rather be coding ASM! wrote:
>>  It's not that the Z80 "requires a 4x clock", but four T-states (clock
>> cycles) make up one machine cycle.  Instructions take anywhere from one
>> to several machine cycles to execute, depending on the instruction.
> 
> This is pretty much my recollection. My Zeta system I knocked up
> recently as part of the N8VEM project can have usually 6-20Mhz
> oscillators assuming you have correct divider crystal for the disk
> controller (FCD266). At 20Mhz, the thing goes like stink compared to my
> spectrum. Decompresing 128kb via UNARC.COM is perhaps 6-7 seconds.
> Spectrum+3 doing the same deal is around 40 seconds and the C128Dcr
> rumbles in at around 2min 10s because the Z80 is running effectively at
> 2Mhz becuase it shares the bus with the MOS6502 derivative. Old
> communist UB880-D (Z80 clone) at 2mhz takes ~4 minutes (I wandered off
> to make a cuppa).

  Jeeze!  Neat stuff though.

> I was talking more "higher level". Have you guys checked out the Zilog
> ez80's? They don't support the full z80 ISA but that can be
> micro-coded/patched pretty simply. I've seen one project that ran a port
> of CP/M 2.2C and ZSYS. I don't care how much emulation of a couple of
> missing instructions it has to do, but they run at 50Mhz and have a full
> 4x4x4 pipeline. That'd be 'quick' in my books.

  I've done a few commercial embedded projects with the eZ80F91.  A
50MHz Z80 is a BEAST.  I enjoyed working with it.

  I abandoned it a couple of years ago in favor of ARM7 due to far
better tools and chip availability.  Zilog's management is nearly as
clueless as SGI's!

               -Dave

-- 
Dave McGuire, AK4HZ
New Kensington, PA