lenny@icus.UUCP (Lenny Tropiano) (06/23/88)
About 4 months ago I replaced my *DEAD* UNIX pc fan with a Radio Shack
12VDC 27CFM (cubic feet/minute) 3.5" look-alike. Back then I didn't have
any problems, but just recently I'm experiencing sporatic memory parity
errors. Diagnostics show nothing, although I think I am attributing it
to the extreme heat and humidity my bedroom is experiencing.
A little about my configuration:
67MB 3B1 with 3.0MB RAM (1 dual EIA w/1.0MB on the board)
DOS-73 board
two Radio Shack 3.5" 27CFM 12VDC fans
A little about my problem and a few questions:
...
NMI (parity error) at 0x2FFAC8 (*hpte: 0xE17B) Wed Jun 22 15:00:07 1988
NMI (parity error) at 0x84298 (*hpte: 0x412F) Wed Jun 22 17:52:22 1988
NMI (parity error) at 0x84298 (*hpte: 0x412F) Wed Jun 22 17:52:28 1988
NMI (parity error) at 0x84298 (*hpte: 0x412F) Wed Jun 22 17:53:02 1988
NMI (parity error) at 0x82200 (*hpte: 0x4169) Wed Jun 22 19:05:11 1988
Problem didn't exist in cool weather even with the EIA/RAM expanded memory.
Problem didn't happen [as often] with EIA/RAM board out.
Machine is running 24 hrs/day 7 days/wk.
Programs die with "Memory fault - core dumped" (signal 8) when the
NMI parity error occurs. (as suspected)
All socketed memory is seated tight.
Boards get very hot, does this effect memory?
Addresses vary from low 0x80000 to high 0x2FFFFF (does the lower addresses
correspond to physical location of memory or just where the program is
at that time which can be on either the motherboard memory or expanded
memory?
Was the 3B1 original fan a more output (ie. 34CFM) and maybe it is not
pulling enough cool air?
How does the heat and humidity affect the operation?
Will air conditioning alliviate this?
What does "(*hpte: 0x4169), etc..." mean?
Using the "s4test" expert mode can I run the memory parity test on
a particular segement is a "SOAK" mode where I it will run over and
over on that test?
Sort of related question...
Has anyone tried the 120VAC fan instead of the 12VDC fan and reduce
the strain on the power supply? Does the AC hum interfere with
any operation?
Thanks for any help,
Lenny
--
US MAIL : Lenny Tropiano, ICUS Software Systems IIIII CCC U U SSS
PO Box 1 I C U U S
Islip Terrace, New York 11752 I C U U SS
PHONE : (516) 968-8576 [H] (516) 582-5525 [W] I C U U S
TELEX : 154232428 [ICUS] IIIII CCC UUU SSS
AT&T MAIL: ...attmail!icus!lenny
UUCP : ...{talcott, boulder, pacbell, sbcs, mtune, bc-cis}!icus!lenny ford@elgar.UUCP (Mike "Ford" Ditto) (06/24/88)
In article <401@icus.UUCP> lenny@icus.UUCP (Lenny Tropiano) writes: >About 4 months ago I replaced my *DEAD* UNIX pc fan with a Radio Shack >12VDC 27CFM (cubic feet/minute) 3.5" look-alike. Back then I didn't have >any problems, but just recently I'm experiencing sporatic memory parity >errors. Diagnostics show nothing, although I think I am attributing it >to the extreme heat and humidity my bedroom is experiencing. I have heard many stories of 3b1's dying because of temperature problems. When mine was serviced (it was DOA when I got it), the AT&T technician warned me to make sure the back end of the system was always given access to air flow (so the fan(s) don't have to work too hard). He said he had seen several 3b1's with the top of the case (to the right side of the monitor) melted down into the power supply. I would make sure that your system is getting *at least* the amount of air flow as a normal 3b1. I have an extra 120V fan sitting in back of my 3b1 which I turn on during hot weather. As far as I know, I've never had any temperature problems. >A little about my problem and a few questions: > ... > NMI (parity error) at 0x2FFAC8 (*hpte: 0xE17B) Wed Jun 22 15:00:07 1988 > NMI (parity error) at 0x84298 (*hpte: 0x412F) Wed Jun 22 17:52:22 1988 > NMI (parity error) at 0x84298 (*hpte: 0x412F) Wed Jun 22 17:52:28 1988 > NMI (parity error) at 0x84298 (*hpte: 0x412F) Wed Jun 22 17:53:02 1988 > NMI (parity error) at 0x82200 (*hpte: 0x4169) Wed Jun 22 19:05:11 1988 > > [ ... ] > > Boards get very hot, does this effect memory? It sure can. It definitely can make a flakey memory chip act up. > Addresses vary from low 0x80000 to high 0x2FFFFF (does the lower addresses > correspond to physical location of memory or just where the program is > at that time which can be on either the motherboard memory or expanded > memory? > What does "(*hpte: 0x4169), etc..." mean? Those addresses are the virtual addresses (programs start at 0x80000 on the Unix PC). The *hpte value is the hardware page table entry, described in <sys/pte.h>. You might be interested in the bottom 10 bits (0x169 in this example) which are the physical page number of the location with the error. So in this case, it was physical address 0x169000 (part of the second meg of ram on the motherboard). > Was the 3B1 original fan a more output (ie. 34CFM) and maybe it is not > pulling enough cool air? > How does the heat and humidity affect the operation? > Will air conditioning alliviate this? I don't know, but you should definitely look into the fan capacity and make sure you're not torturing your system. > Using the "s4test" expert mode can I run the memory parity test on > a particular segement is a "SOAK" mode where I it will run over and > over on that test? I have done this, but it was a long time ago and I don't feel like rebooting to try it now. I don't think it even required expert mode. > Sort of related question... > > Has anyone tried the 120VAC fan instead of the 12VDC fan and reduce > the strain on the power supply? Does the AC hum interfere with > any operation? I've used an external 120VAC fan as described above, but primarily just to keep air flowing past the back of the machine. No ill effects so far. -=] Ford [=- "Once there were parking lots, (In Real Life: Mike Ditto) now it's a peaceful oasis. ford@kenobi.cts.com This was a Pizza Hut, ...!sdcsvax!crash!kenobi!ford now it's all covered with daisies." -- Talking Heads
gws@n8emr.UUCP (Gary Sanders ) (06/24/88)
In article <401@icus.UUCP> lenny@icus.UUCP (Lenny Tropiano) writes: > >A little about my problem and a few questions: > ... > NMI (parity error) at 0x2FFAC8 (*hpte: 0xE17B) Wed Jun 22 15:00:07 1988 > NMI (parity error) at 0x84298 (*hpte: 0x412F) Wed Jun 22 17:52:22 1988 > I get NMI errors about every two months. What I do to get rid of the errors is give the 7300 mother board a good dusting..It seems that "dust bunnies" jump around the mother board and if some of the bigger bunnies get across a couple of bits in memory you get an error. Installing the new fans most likley is stiring up more dust than normal. If you have 2 fans on your PC you might do bettery by putting a filter in place of one fan and the remaining fan do the work. 3B1's and upgraded 7300 only have 1 fan on the (power supply) side of the machine and a sponge filter fills the other slot.. Either way, you can get yourself a compressed air bottle and blow the the dust bunnies away....(waste'em...) Seem to help my machine.. -- Gary W. Sanders HAM/SWL BBS 614-457-4227 (uucp) gws@n8emr (uucp) osu-cis!n8emr!gws (packet) N8EMR @ W8CQK (cis) 72277,1325
erict@flatline.UUCP (j eric townsend) (06/30/88)
In article <198@elgar.UUCP>, ford@elgar.UUCP writes: > I have heard many stories of 3b1's dying because of temperature > problems. When mine was serviced (it was DOA when I got it), the AT&T > technician warned me to make sure the back end of the system was > always given access to air flow (so the fan(s) don't have to work too > hard). He said he had seen several 3b1's with the top of the case (to > the right side of the monitor) melted down into the power supply. Well, right now I'm running a 3b1, 3.5Mb RAM, 67Mb HD in an unairconditioned apartment in a two floor duplex. There are trees and such around the south and north sides, an apartment on the north side, so sunlight only seeps in after about 1800hrs. I have a window fan that sucks air through the house at a pretty nice clip. Avg interior temp is probably around 90F or so. About once every hour or so, the fan kicks into what sounds like high gear for a few minutes. I've had no overheating problems of any sort at all. I keep 2 in open on the right and left sides, 8 in in the front and back. Airflow seems to be good, but there's a lot of dust in this old house, so I have to open it up and clean it out once every now and again... :-( Hope this info was useful to someone. -- Skate UNIX or go home, boogie boy... J. Eric Townsend ->uunet!nuchat!flatline!erict smail:511Parker#2,Hstn,Tx,77007 ..!bellcore!tness1!/
greg@csanta.UUCP (Greg Comeau) (07/09/88)
>... He said he had seen several 3b1's with the top of the case (to >the right side of the monitor) melted down into the power supply. Mine did this once: after replacing a hard disk, the AT&T tech forgot to plug the fan back in. Luckily, I caught it before it was too late. >> NMI (parity error) at 0x2FFAC8 (*hpte: 0xE17B) Wed Jun 22 15:00:07 1988 Lenny, one more question about this: You say a program gets a memory dump and then does the system crash? Is it always the same program (or a version of it) that does this? I know I was able to crash my UNIX PC on demand back at the 3.0 level with some sick pointer stuff. I doubt it, but perhaps this is related.
lenny@icus.UUCP (Lenny Tropiano) (07/11/88)
In article <136@csanta.UUCP> greg@csanta.UUCP (Root) writes: ... |> |>>> NMI (parity error) at 0x2FFAC8 (*hpte: 0xE17B) Wed Jun 22 15:00:07 1988 |> |>Lenny, one more question about this: |>You say a program gets a memory dump and then does the system crash? |>Is it always the same program (or a version of it) that does this? |>I know I was able to crash my UNIX PC on demand back at the 3.0 level |>with some sick pointer stuff. I doubt it, but perhaps this is related. No I said it get's a "memory fault, signal 8". Usually the program dumps core. (ie. Memory fault - core dumped). I see core files laying around all over. The system hasn't crashed because of this, unless the NMI (parity error) was in the kernel, then probably the system would panic and die. The pointer stuff and this is unrelated, you were right about that. My problem is either heat, bad memory chip, or a combonation of both. -Lenny -- Paper-net: Lenny Tropiano | @-net: lenny@icus.UUCP ICUS Software Systems | !-net: ...att \ PO Box 1 | boulder \ Islip Terrace, NY 11752 | talcott !icus!lenny Vocal-net: (516) 582-5525 [work] | pacbell / (516) 968-8576 [home] | sbcs / Telex-net: 154232428 ICUS | Another-net: attmail!icus!lenny