RE: Subing EZ80 microcontrollers for Z80A cpus

Just throwing this out there - will the protection chip handle the speed
increase since it's essentially regenerating the addressing? Or are the
implementations talked about here are complete overhauls with unprotected
EPROM code?

Chris Brooks

From: owner-vectorlist@vectorlist.org
[mailto:owner-vectorlist@vectorlist.org] On Behalf Of Clay Cowgill
Sent: Sunday, December 04, 2011 1:30 PM
To: vectorlist@vectorlist.org
Subject: RE: VECTOR: Subing EZ80 microcontrollers for Z80A cpus

I really don't think you're going to need a *lot* more performance. The
game runs pretty smooth after the enemies get down to a certain point (so
more CPU *after* the "back to full framerate" threshold is just overkill).
To be 'clever', I would try something like maybe just a CPLD that sits on
the Z80 bus and clock and watches the Z80 cycle and address range. Put a
16MHz crystal on the CPLD clock. Then (assuming this board also has all the
ROMs in a fast modern memory like a multigame uses) I'd switch from a
divided 8MHz clock to the 16MHz clock during an instruction fetch, or maybe
just double it *any* time it's hitting the ROM address range.

If a standard instruction cycle is, say, 10 T-states and the first four are
getting the opcode, then when we see /M1 assert we could substitute four
double-speed clock cycles. If the Z80 tolerates that brief over-clocking,
you'd get about a 20% CPU boost for instruction fetches. (saves two 'slow'
T-states worth of time out of the ten in the cycle) That would shorten the
/RFSH as well, but I'd actually be surprised if that was used for anything
in the vector games. (I'm thinking it's probably only used for DRAM in the
frame buffer based games on the G80 system?)

OTOH, I also suspect that just given how things run in the system that you
could stick a Z80B in there and change the cpu crystal to ~10MHz or so and
it'd probably work to speed up 'slow spots' in the game with essentially no
effort. ;-)

Another angle if you really just want to mess with a CPU replacement for fun
would be to see if there were any 5V Z180's. (I think there were.) Like a
Z80180 or Z80182... They run up to like 33MHz or something crazy. (That's
what Namco had in that Pacman/Galaga "Class of 81" machine they sold for
ages to run the original code without needing multiple Z80's and whatnot.)
Might be less work without needing to worry about level translation and the
3.3V rail, etc.

It might not be too much work to turn my old G80 menu system into a "loop
counter". Basically just count how many times a loop can execute in between
display refresh interrupts and print the number. That could essentially be
a "benchmark" for messing with CPU speed tweaks then.

-Clay

  _____

From: owner-vectorlist@vectorlist.org
[mailto:owner-vectorlist@vectorlist.org] On Behalf Of Zitt Zitterkopf
Sent: Sunday, December 04, 2011 1:26 AM
To: vectorlist@vectorlist.org
Subject: Re: VECTOR: Subing EZ80 microcontrollers for Z80A cpus

Thanks Clay... and Bill.

For now; I'll table this plan... Too busy with the Xy Gen and other projects
anyway.

I did spent about two nights thinking and planning; I think it's possible to
do the substitution - I'm just not sure if it'll have the desired impact.

This biggest problem is the EZ80 seems to have removed the !refresh pin...
so that would have to be duplicated somehow. I'm also concerned about
possible uC peripherals conflicting with Sega's memory (or IO) map.

As Bill mentioned; the PCB would have to have a 3.3V regulator... and the
EZ80 doesn't come in a package I really like. BGA or 144pin LQFP... both may
be on the big side for a DIP sized drop in.

I wouldn't really care about "loosing" 90% of the EZ80's functionality; as
long as it "improved" the lag problem.

I know Z80s are widely available as pulls / etc; I just wanted to get a
pipelined version to see what it did to the lag.

A side effect; might be that highscores could be saved in flash; but still a
complex project I'm not motivated to tackle yet.

Too bad I have a day job. ;P

John

---------------------------------------------------------------------------
** Unsubscribe, subscribe, or view the archives at http://www.vectorlist.org
** Please direct other questions, comments, or problems to chris@westnet.com
Received on Sun Dec 4 14:38:37 2011