🚀 go-pugleaf

In <563@ethz.UUCP>, lengge@bernina.UUCP (Thomas Lenggenhager) said:  

>Next month, I hope, I will get some of the very first ** 4 Meg-SIMMs **, but a
>friend of mine told me the other night that there are problems with some piece
>of system software that the new chips won't be recognized.
>...
>1) 4 MB SIMMs won't be recognized at all in Plusses and SEs, right?  Why not,
>   it's really a pitty :-(

True.  The Mac Plus & SE were simply "designed that way".  Perhaps someone
at Apple could spell out the engineering reasons; my personal guess was
that it turned out (probably) to be a fair amount cheaper to do it that
way.  In any case, while there is address space on the 68000 to have two
of those SIMMs filled with 4MB parts, there isn't space for all four to
be filled and still have anything left over for memory-mapped i/o (see my
response to #3 below...).

>2) In the Mac II, the following things happen:
>-  The motherboard looks at the first 4 SIMMs and checks if these are 16 MB
>   parts. If it doesn't find them, the next lower MB (KB) SIMMs are searched
>   for (16MB -> 4MB -> 1MB -> 256KB).
>-  Then, the second 4 SIMMs are checked if there are the same SIMMs as in the
>   first four rows or if there are lower parts installed.

It's my understanding that 4MB parts are at the top of that loop, and not
16MB; making the max you could go directly on the motherboard 32Meg.  But
I haven't poked around the Mac Family Hardware Ref. manual in a few months
and might well be wrong on this one.

>3) Is there somewhere a problem for me that makes it difficult or impossible to
>   access this memory with the Mac OS? (I think, with Unix that should be no
>   problem).

The MacOS in versions 6.0 and below runs in 24-bit addressing mode, making
the maximum amount of RAM addressable 16MB.  Of that 16, 1 goes to each of
the six card slots, and much of another Meg is used for memory-mapped I/O,
leaving something not much over 8 Meg left (again, I haven't been sniffing
around the Hardware Ref. for a while & my numbers may be a tad off...).
With 6.01 coming so quickly on the heels of 6.0 I would doubt that we're
out of 24-bit addressing and into 32-bit.  What this all means is that we
can't talk to more than 8MB of SIMMs on the motherboard currently in the
Mac OS.  Running A/UX this shouldn't be a problem.  I don't have any
direct knowledge of anyone testing SIMMs bigger than 1 Meg in a MacII
under A/UX, though Apple says that it should work (again in that wonderous
Mac Family Hardware Ref. manual).

>Thank you very much for your help. As soon as I'll get them I will send a short
>report on these SIMMs.

I would be very interested in that report...

Peter
--
Peter "Arrgh" Korn
korn@ucbvax.Berkeley.EDU
{decvax,hplabs,sdcsvax,ulysses,usenix}!ucbvax!korn