Aucbvax.1945
fa.works
utzoo!duke!decvax!ucbvax!DP@MIT-ML
Fri Jun 26 05:28:30 1981
Address space vs. physical memory.
DP@MIT-ML 06/26/81 01:26:23 Re: Address space vs. physical memory.
To: WorkS at MIT-AI
The two are not equal. It is perfectly reasonable to have far
more of one than you have the other (and there are examples of both
in common use). There are problems that cannot be made to fit in the
small address space of a 16 bit machine. But wait there is overlaying,
certainly that can solve things. Unfortunately there comes a time when
the largest branch of the overlay tree just wont fit in the address
space. This happend at DEC with the Bliss compiler for the pdp-11. It
to this day remains a cross compiler, because no matter how clever they
were with overlays something or other just wouldn't fit.
The only thing that mapping and large physical memory give you
with a small address space, is speed it saves you reading overlay
segments (or users in a multi user system) in and out of memory from the
disk.
BTW, even large systems are not immune. There are people complaining that
their problems wont fit into the 16mb of the 370. (even with the crockish
region scheme that ibm built (memory mapping, but done painfully I am
told)). All these new machines with huge (4gb) address spaces are claimed
to satisfy medium range requirements. It is apparently expected that in
8 - 10 years or so, 32 bits wont be enough.
Jeff
-----------------------------------------------------------------
gopher://quux.org/ conversion by John Goerzen <
[email protected]>
of
http://communication.ucsd.edu/A-News/
This Usenet Oldnews Archive
article may be copied and distributed freely, provided:
1. There is no money collected for the text(s) of the articles.
2. The following notice remains appended to each copy:
The Usenet Oldnews Archive: Compilation Copyright (C) 1981, 1996
Bruce Jones, Henry Spencer, David Wiseman.
