schow@bcarh61.bnr.ca (Stanley T.H. Chow) (11/13/89)
In article <371@ucsvc.ucs.unimelb.edu.au> U5533129@ucsvc.ucs.unimelb.edu.au (CARDIOLOGY, R.M.H.) writes: >There was a rumour some months back that a program could be written which >fiddled the NEAT chip set registers to enable the block of RAM between 640k >and 704k (for those without EGA). Does anyone know if this has been done, or >alternatively how it could be done? > >Peter S. Yes, it is possible to fiddle the NEAT chip set to make it hand the memory back. DOS would still have to be told to use it. If you can't find the program (or if no one has written it), I can whip it up real quick. This is useful only if you have on a 286 or a 386 *not* running QEMM. Now a question of my own: Is QEMM truly brain-damaged or have I just not found the right magic spell? Specifically, I am trying to recover the "hidden" 384K of RAM (you know, the stuff above 640K that NEAT dedicates to ROM shadowing). One my system: NEAT chipset with AMI bios, the ROM bios turns off all the hidden 384K (except for shadowing), so QEMM thinks there is no memory there. Oh, that's easy, I said. So I play around with the NEAT registers and turned the memory back on. Now QEMM recognizes the memory. *However*, QEMM now also marks the LOGICAL address range as "Unavailable"! It seems I can have either the physical memory or the logical address space. I want *both*. A friend (hi Mark) tells me this has been discussed before in the group and the conclusion was that QEMM is not up to the job. Please tell me this ain't so. [I did not follow the earlier discussion since I only got my 386 a couple of months ago :-)]. Stanley Chow BitNet: schow@BNR.CA BNR UUCP: ..!psuvax1!BNR.CA.bitnet!schow (613) 763-2831 ..!utgpu!bnr-vpa!bnr-rsc!schow%bcarh61 Me? Represent other people? Don't make them laugh so hard.
michaelk@copper.WR.TEK.COM (Michael D. Kersenbrock) (11/14/89)
>Yes, it is possible to fiddle the NEAT chip set to make it hand the >memory back. DOS would still have to be told to use it. If you can't find >the program (or if no one has written it), I can whip it up real quick. >This is useful only if you have on a 286 or a 386 *not* running QEMM. There's a program "EEMRAM" available archived as "eemram21.arc" that will use LIM 4.0 expanded memory to add memory past 640K. It's not a TSR, it just fills in memory past 640K, tells DOS about it, and exits. My XT-clone has 704K (720,896 bytes), and works just fine. Seems that if the NEAT chip set were set up for expanded memory (and QEMM supplying the LIM driver ???), that this program might work on other machines as well (w/o EGA/VGA). -- Mike Kersenbrock Tektronix Microprocessor Development Products michaelk@copper.WR.TEK.COM Aloha, Oregon
burkett@csd4.csd.uwm.edu (Edward W Burkett) (11/14/89)
In article <700@wrgate.WR.TEK.COM> michaelk@copper.WR.TEK.COM (Michael D. Kersenbrock) writes: > > >>Yes, it is possible to fiddle the NEAT chip set to make it hand the >>memory back. DOS would still have to be told to use it. If you can't find >>the program (or if no one has written it), I can whip it up real quick. >>This is useful only if you have on a 286 or a 386 *not* running QEMM. > >There's a program "EEMRAM" available archived as "eemram21.arc" that >will use LIM 4.0 expanded memory to add memory past 640K. It's not >a TSR, it just fills in memory past 640K, tells DOS about it, and exits. > The only problem most people have with expanded memory is that it must be "True" LIM 4.0. Most expanded memory cards are not hardware LIM 4.0 compatable but rather LIM 3.2 cards that have a LIM 4.0 driver. I have one of these cards, tried EEMRAM and the card is not capable of addressing the correct BIOS address. I have talked with many EMS card owners and this is a common problem. True hardware compatable LIM 4.0 cards tend to be much more expensive than LIM 3.2 cards with a LIM 4.0 driver. Just a note for those of you about to run out and by an inexpensive EMS card. Ed Burkett University of Wisconsin -- Milwaukee Department of Biological Sciences
schow@bcarh61.bnr.ca (Stanley T.H. Chow) (11/15/89)
In article <700@wrgate.WR.TEK.COM> michaelk@copper.WR.TEK.COM (Michael D. Kersenbrock) writes: > >>Yes, it is possible to fiddle the NEAT chip set to make it hand the >>memory back. DOS would still have to be told to use it. If you can't find >>the program (or if no one has written it), I can whip it up real quick. >>This is useful only if you have on a 286 or a 386 *not* running QEMM. > >There's a program "EEMRAM" available archived as "eemram21.arc" that >will use LIM 4.0 expanded memory to add memory past 640K. It's not >a TSR, it just fills in memory past 640K, tells DOS about it, and exits. > >My XT-clone has 704K (720,896 bytes), and works just fine. Seems that >if the NEAT chip set were set up for expanded memory (and QEMM supplying >the LIM driver ???), that this program might work on other machines as >well (w/o EGA/VGA). > >-- >Mike Kersenbrock >Tektronix Microprocessor Development Products >michaelk@copper.WR.TEK.COM >Aloha, Oregon The problem is not quite what you think it is. On systems with NEAT & AMI bios (like my 386), the memory is *missing*. There are 2 Meg worth of memory chips in my machine, the POST memory test confirms there is 2048K. By the time DOS gets control, there is only 640K of conventional memory and 1024K of extended memory. The region between 640K and 1 Meg is gone. Some of the missing 384K is used for ROM shadowing. The rest is just plain "turned-off". It is not hiding in extended memory. The CPU really cannot talk to that memory. This is a quirk of how AMI handles NEAT. As a consequence, QEMM cannot use that memory. What I said in the original article is that it is possible to set the NEAT registers so that the missing memory shows up again. You can then use it anyway you see fit. (I also complained that QEMM does the wrong thing with it, so that it is *better* to leave it missing). Also, the NEAT chip set knows nothing about expanded memory. QEMM (or some other 386 memory manager) is using the 386 MMU to implement EMS. BTW, all these things are true for *my* 386 system. I only expect other 386 and 286 systems to be like that. -- Stanley Chow BitNet: schow@BNR.CA BNR UUCP: ..!psuvax1!BNR.CA.bitnet!schow (613) 763-2831 ..!utgpu!bnr-vpa!bnr-rsc!schow%bcarh61 Me? Represent other people? Don't make them laugh so hard.