schwartz@dinl.uucp (Michael Schwartz) (06/03/89)
Thanks to all those who responded with help on the problem of expanded memory with Windows/286. A number of the responses were not applicable to my system, but may interest others. With some pointers from those of you who helped, I'm now up and running. The only question remaining is "what is the difference between an AboveBoard 286 and an AboveBoard Plus", the latter being all that is marketed by Intel today. From: waltw@pogo.wv.tek.com (Walt Wilson) Try looking for memory conflicts, especially if you are using a Paradise VGA. Walt informs me that CONFIG.SYS can be changed to take care of this. From: ames!auvax.AthabascaU.CA!kevinc If you don't have emm set up exactly to how much mem you have then the whole damn system barfs. Kevin's statement is a critical area if the board is being used for extended memory. For expanded memory, EMM.SYS can be told to CHECK to see if one has EXACTLY the amount of expanded memory expected. From: Scott Thurlow <gatech!cs.utexas.edu!uunet.UU.NET!rlgvax!scott> I had a similar problem with my AST Premium 286. I solved it by using the expanded memory driver that came with Windows rather than the expanded memory driver that came with the hardware. Scott's statement is also critical if you have one of the older EMM drivers. Curiously, the AboveBoard came with a driver dated 6 months earlier than the Windows driver, though they could be byte-compared with no differences! One of the MicroSoft engineers (Ralph Lipe) posted a news article (5798@microsoft.UUCP) providing technical detail on the meaning of an Internal Stack overflow, and a mechanism for controlling it in the CONFIG.SYS file. This was disclaimed as a personal opinion. I may never really understand what caused the problem, though the most substantial change I made was replacing the HIMEM.SYS that came with windows/286 with the HIMEM.SYS from my SDK. I believe the problem was caused by conflicts over a software interrupt; SOMETHING caused the IRQ3 handler to go off in the weeds. Additionally, I installed all the windows drivers AFTER the shell command in my CONFIG.SYS (I'm too stubborn to patch COMMAND.COM to give me a reasonably sized environment), though I have not (and probably will not) test to see if this is a problem. For debugging purposes, I tested this with the slow boot" windows installation. I decided I like it this way, so "slow boot" will stay, also. There is the slightest possibility that the problem was tied to the contents of a strange root directory entry that I wiped out in tracking through one of my drivers. If it recurs, I will post. An interesting side effect is that Windows grabs about 128K of the expanded memory at start-up, but gives half of it back at the first opportunity to move segments. I STRONGLY suggest that anyone with a similar problem write a small program to dump interrupt vectors and trace what's going on in the more suspicious ones. Of course, prepare a fresh boot diskette to help control this, and add drivers one at a time, saving the results for comparison. Also, as a side issue, a logitech mouse must be powered down if anything wrote to its serial port before trying to use it again. BTW, I believe the newer HIMEM.SYS is the one available on SIMTEL20 (pardon me, WSMR-SIMETEL20) as PD1:<MSDOS.SYSUTL>HIMEM.ARC, but am not positive (Thanks Ralf Brown and Keith Petersen!). Sorry for the length, but I hope this helps others. -- Michael Schwartz ncar!dinl!schwartz MSchwartz@Dockmaster.ARPA
patrickd@chinet.chi.il.us (Patrick Deupree) (06/05/89)
In article <1036@dinl.mmc.UUCP> schwartz@dinl.uucp (Michael Schwartz) writes: > >The only question remaining is "what is the difference between an >AboveBoard 286 and an AboveBoard Plus", the latter being all that is >marketed by Intel today. > There's a difference that is really kind of small, but actually very big. In the world of EMS there are large frame and small frame EMS. Large frame EMS is usually called 4.0 and the standard was set by Intel. Unfortunatly they made a little boo boo and, even though the AboveBoard 286 uses LIM4.0, it doesn't support large frame EMS. So, they fixed this little error with the AboveBoard Plus (luckily all it involved was replacing one chip, so they were able to use all the old AboveBoard 286's). Since there is no reason to get a small frame board if a large frame one is available, the 286 board was phased out. We're getting one of the Plus's in here soon to give it a shot with our product. Should be interesting. -- "I place my faith in fools. Self confidence, my friends call it." -Edgar Allen Poe Patrick Deupree -> patrickd@chinet.chi.il.us