Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!zaphod.mps.ohio-state.edu!wuarchive!udel!oscar.ccm.udel.edu!johnston
From: johnston@oscar.ccm.udel.edu
Newsgroups: comp.sys.mac.hardware
Subject: Re: 2meg SIMMs in an SE/30?
Message-ID: <55784@nigel.ee.udel.edu>
Date: 8 Jun 91 07:17:48 GMT
Sender: usenet@ee.udel.edu
Lines: 29
Nntp-Posting-Host: minnie.me.udel.edu

In article <2860@kielo.uta.fi>, ccjapu@uta.fi (Jarmo Puntanen) writes...
>In article <55485@nigel.ee.udel.edu> johnston@minnie.me.udel.edu writes:
>>Will the SE/30 or IIcx recognize 2 MB SIMMs?  From 5 meg, it seems
>>that one has the option of going to 20 with 4 x 4 MB, or 8 with
>>4 x 1 MB.  
>I thought the 24-bit ROMs in SE/30 limit you at 17Mb when using virtual
>memory, 8 without the virtual memory manager. IICx's 32-bit ROMs allow
>larger virtual memory pool, but the 8 Mb physical limit persists.

Some of the details may be inaccurate (SE/30 is limited to 14 virtual, 
and the IIcx has a 24-bit rom) but an interesting point was made about
the "physical limit" of 8-meg.

My understanding is that there IS a good reason for using >8 meg physical
in a 24-bit-ROM Mac.  With VM turned on, paging (swapping physical
for disk-mapped RAM) should occur only when the limits of physical RAM
are exceeded.  Thus VM can be thought of as a way to take advantage
of more physical ram, and not only as a trick to get extra address
space by mapping the RAM to the hard-disk.  True or false?
Does the SE/30 have to actually move the data into the first 8 megs
before it can be addressed by the CPU?

What would be nice is a virtual memory implementation that ropes off
the extra disk space only if available physical RAM is less than
the requested virtual RAM.  That would make 16meg SE/30's reasonable,
and I am sure that somebody could think of a way to use the 16-14=2megs
leftover as a RAM disk.  The DOS world is full of hacks like that.

-- Bill (johnston@minnie.me.udel.edu)