[Coco] MPI question

[jdaggett at gate.net]

On 7 Jul 2008 at 17:21, Robert Gault wrote:

> jdaggett at gate.net wrote:
> > On 7 Jul 2008 at 13:54, Chuck Youse wrote:
> > 
> > 
> >>Ok, so I know that multi-paks had some sort of problem with the Coco
> >>3, requiring a modification to the PAL - something to do with
> >>conflicts with the GIME.  As far as I can tell, the MPI only uses
> >>address FF7F - did it only partially decode and thus overlap with
> >>the GIME?  Or was there a different problem?
> >>
> >>C.
> > 
> > |--------------------
> > 
> > The later model sof the MPI uses $FF7F. If I remember correctly the
> > original MPI uses $FF9F to do slot switch. This conflicts with the
> > GIME's Horizontal Offset register. There is another conflict that I
> > am not familiar with.
> > 
> > The GIME chip knows when and where the MPU addresses anything within
> > the 64K memory map of the 6809. The internal registers to the GIME
> > chip are at $FF90 to $FFDF. Also the GIME chip mirrors the PIA
> > registers at $FF00 - $FF03 AND $FF20 - $FF23.
> > 
> > james
> > 
> 
> James, I think there is a minor error above. The MPI always used $FF7F
> but ghosted to $FF9F. There is also a ghosting problem with some other
> third party hardware PAKs. The ghosting is removed by the MPI
> modifications.
> 

Just going on memory. I thought there ws a conflict with the GIME and 
maybe what you state is what I am thinking of. 

james