RM5I%DFVLROP1.BITNET@cunyvm.cuny.edu (01/10/89)
Date: 10 January 1989, 12:32:46 MEZ From: RM5I at DFVLROP1 To: info-min at udel.edu Subj: Extended/Expanded Memory on PC's I read a lot of things about using EXTENDED MEMORY with MINIX. What is about EXPANDED MEMORY on PC's with MINIX ? Maybe it is a little bit confusing (EXTENDED/EXPANDED MEMORY). What i mean is the extra memory above the 640k on a PC which is not a 286 or 386. Could please someone drop a line on this...... Regards Roland (rm5i@dfvlrop1) Bitnet/Earn
pcm@iwarpj.intel.com (Phil C. Miller) (01/11/89)
In article <6470@louie.udel.EDU> RM5I%DFVLROP1.BITNET@cunyvm.cuny.edu writes: >Date: 10 January 1989, 12:32:46 MEZ >From: RM5I at DFVLROP1 >To: info-min at udel.edu >Subj: Extended/Expanded Memory on PC's >I read a lot of things about using EXTENDED MEMORY with MINIX. >What is about EXPANDED MEMORY on PC's with MINIX ? Maybe it is >a little bit confusing (EXTENDED/EXPANDED MEMORY). >What i mean is the extra memory above the 640k on a PC which is >not a 286 or 386. >Could please someone drop a line on this...... A brief correction and note of explanation... First of all, expanded memory can be used on a 286 just fine; I assume this is also true of a 386. Extended memory is memory which is mapped to the address space above 1 Meg (0x100000). For that reason, an operating system which runs in REAL mode cannot access extended memory; operating systems like Microport Unix run in PROTECTED mode, a mode unavailable on 8088's. This is one of the more important differences between an XT and an AT. Expanded memory uses hardware tricks to make more memory available, but always with a fixed amount available at a given time, 640k. When more memory is needed, a 64k-chunk of memory is "mapped out" and another is "mapped in". The new "chunk" is accessed using the same addresses as the old "chunk" and the old chunk is unavailable. In both cases, the user has the 16Mb of memory available for use; with expanded memory, operating systems which run in REAL mode (most notably DOS and MINIX) can have access to more than 640k. >Regards Roland (rm5i@dfvlrop1) Bitnet/Earn Phil Miller Intel iWarp program Intel Development Operations Hillsboro, Oregon, USA
ralph@cc.brunel.ac.uk (Ralph Mitchell) (01/16/89)
In article <4054@omepd.UUCP> pcm@iwarpj.UUCP (Phil C. Miller) writes: [discussion deleted...] > >In both cases, the user has the 16Mb of memory available for use; with >expanded memory, operating systems which run in REAL mode (most notably >DOS and MINIX) can have access to more than 640k. > >Phil Miller What about the case where a PC has 3x256K ram chips. Surely the whole 768K is contiguous memory, at least in the chips. Are you saying that the address selection hardware is deliberately stopping access above 640K and that there's some more hardware to map the remaining 128K elsewhere ?? Sounds like the designer had a bad day... Ralph Mitchell -- From: Ralph Mitchell at Brunel University, Uxbridge, UB8, 3PH, UK JANET: ralph@uk.ac.brunel.cc ARPA: ralph%cc.brunel.ac.uk@cwi.nl UUCP: ...ukc!cc.brunel!ralph PHONE: +44 895 74000 x2561 "There's so many different worlds, so many different Suns" -- Dire Straits
wbeebe@bilver.UUCP (bill beebe) (01/18/89)
In article <533@Terra.cc.brunel.ac.uk> ralph@ccs.brunel.ac.uk (Ralph Mitchell) writes: > >What about the case where a PC has 3x256K ram chips. Surely the whole >768K is contiguous memory, at least in the chips. Are you saying that >the address selection hardware is deliberately stopping access above >640K and that there's some more hardware to map the remaining 128K >elsewhere ?? Sounds like the designer had a bad day... > In the original PC spec, IBM said in effect that there would be no more than 640K of contiguous, program memory, starting from 0000:00 to 9FFF:00. Memory starting from A000:00 to BFFF:00 belongs to video display(s). Some systems can get away with an additional 64K starting at A000:00 because MDA and Herc video cards actually start at B000:00. There is indeed hardware in the new PCs (via chipsets and some newer discrete designs) that map memory above 640K into extended memory (ATs) and/or as shadow RAM for display BIOS, ROM BIOS, etc. And no, the designer did not have a bad day, considering the design is vintage 1980, when the majority of personal systems (CPM and Apple II) had 64K or less. I don't want to get into this, as this properly belongs in one of the architecture conferences, so read it, yell at it, even delete it, but don't clutter comp.os.minix with replies to it. I just get a little tired of potshots being taken by armchair engineers at a PC design approaching it's tenth birthday...