tom@syssoft.com (Rodentia) (02/13/91)
Ever since I was a little hacklet, I had been told software
cannot damage hardware (walking disk drives excluded). Imagine
my surprise when my next door neighbor installed windows on his
2.5Mbyte 16MHz 386, ran notepad, and POOF! He got a crash, and
whenever he boots, the RAM test fails after 512K (the point
where it starts accessing the 2Mbyte expansion slot RAM).
I tried running the setup from ROM hoping it was just a CMOS
problem, but to no avail. Now I'm wondering if there was
some obscure Gate A20 problem that was just waiting for
certain lines to assert in the right combination. Actually,
I'm afraid I'm grasping at straws. Was it just bad luck?
The machine is some nameless clone. If I can't figure this
out, I may have to give up engineering and open a deli ;-{)
--
Thomas Roden | tom@syssoft.com
Systems and Software, Inc. | Voice: (714) 833-1700 x454
"If the Beagle had sailed here, Darwin would have | FAX: (714) 833-1900
come up with a different theory altogether." - me |
--
Thomas Roden | tom@syssoft.com
Systems and Software, Inc. | Voice: (714) 833-1700 x454
"If the Beagle had sailed here, Darwin would have | FAX: (714) 833-1900
come up with a different theory altogether." - me |js@dukee.egr.duke.edu (Jeffrey A. Shorey) (02/15/91)
From article <1991Feb12.220834.1991@syssoft.com>, by tom@syssoft.com (Rodentia): > Ever since I was a little hacklet, I had been told software > cannot damage hardware (walking disk drives excluded). Imagine > my surprise when my next door neighbor installed windows on his > 2.5Mbyte 16MHz 386, ran notepad, and POOF! He got a crash, and > whenever he boots, the RAM test fails after 512K (the point > where it starts accessing the 2Mbyte expansion slot RAM). > > Thomas Roden | tom@syssoft.com I had this happen on my CompuAdd with other programs that crashed. Sometimes, even resetting the CMOS settings didn't help...one time the step rate on my harddrive table was set to something random, and the setup program was unable to change it for some reason. What I had to do was take the batteries out of the computer and let it sit for a day while the CMOS discharges (tech support said an hour or two - that never did it for me). Then I reinstalled the batteries, ran setup, and it was fixed. You would think the designers would put some kind of safeguard on the CMOS so that it wouldn't get randomly trashed like that...it would save them a lot of tech support time. - Jeff Shorey
bwb@sei.cmu.edu (Bruce Benson) (02/16/91)
In article <1991Feb12.220834.1991@syssoft.com> tom@syssoft.com (Rodentia) writes: >Ever since I was a little hacklet, I had been told software >cannot damage hardware (walking disk drives excluded). Imagine >my surprise when my next door neighbor installed windows on his >2.5Mbyte 16MHz 386, ran notepad, and POOF! He got a crash, and >whenever he boots, the RAM test fails after 512K (the point >where it starts accessing the 2Mbyte expansion slot RAM). I installed Windows on a no-name 286/12 that had 150ms ram. Windows would cause a parity error, but no other DOS program I ran would - and the boot-up test always passed it. When I de-turboed down to 8mhz I had no problem with windows. I suspect windows gave my memory a better workout then the built-in ram diagnostics. Bruce -- * Bruce Benson + Internet - bwb@sei.cmu.edu + + * Software Engineering Institute + Compuserv - 76226,3407 + >--|> * Carnegie Mellon University + Voice - 412 268 8469 + + * Pittsburgh PA 15213-3890 + + US Air Force
rogerhef@matt.ksu.ksu.edu (Roger Heflin) (02/16/91)
In <13920@as0c.sei.cmu.edu> bwb@sei.cmu.edu (Bruce Benson) writes: >In article <1991Feb12.220834.1991@syssoft.com> tom@syssoft.com (Rodentia) writes: >>Ever since I was a little hacklet, I had been told software >>cannot damage hardware (walking disk drives excluded). Imagine >>my surprise when my next door neighbor installed windows on his >>2.5Mbyte 16MHz 386, ran notepad, and POOF! He got a crash, and >>whenever he boots, the RAM test fails after 512K (the point >>where it starts accessing the 2Mbyte expansion slot RAM). >I installed Windows on a no-name 286/12 that had 150ms ram. Windows would >cause a parity error, but no other DOS program I ran would - and the >boot-up test always passed it. When I de-turboed down to 8mhz I had no >problem with windows. >I suspect windows gave my memory a better workout then the built-in ram >diagnostics. The memory that you are getting the parity error on is probably above the 1M mark. If you where not using extended and/or expanded memory for anything the memory was unused, and since nothing used it there was no reason to get a parity error. Putting the machine down to 8Mhz is very similar to adding a wait state to the ram, it the fact that it slows down the ram access. The boot-up test passes because on most machines I have seen the boot-up test starts without reading the CMOS setup on the machine, and so the ramtest test with a certain maximum number of wait states, it does not see it the wait states are set properly for the memory, it just checks to see it the memory works at the slowest speed allowed by your computer. rogerhef@matt.ksu.ksu.edu
quimby@madoka.its.rpi.edu (Tom Stewart) (02/17/91)
rogerhef@matt.ksu.ksu.edu (Roger Heflin) writes: >The memory that you are getting the parity error on is probably above >the 1M mark. If you where not using extended and/or expanded memory for >anything the memory was unused, and since nothing used it there was no >reason to get a parity error. Putting the machine down to 8Mhz is very >similar to adding a wait state to the ram, it the fact that it slows down >the ram access. The boot-up test passes because on most machines I have >seen the boot-up test starts without reading the CMOS setup on the machine, >and so the ramtest test with a certain maximum number of wait states, it >does not see it the wait states are set properly for the memory, it just >checks to see it the memory works at the slowest speed allowed by your >computer. Another reason is that the boot check doesn't execute above 1M -- it only does read/write. The timing for an opcode fetch is much tighter than for read/write. DOS, even with EMS, never executes about 1M, so the potential problem doesn't appear until you try to run Windows, or UNIX. (Sometimes it doesn't appear until you install UNIX, ship the machine to a customer, and let him try to run 2 users on the machine at once. :) ) Quimby (mailer disfunctional, replies to: quimby@mts.rpi.edu, quimby@rpitsmts.bitnet)