Memtest report error...

[John-Y]

I recently used memtest86 after experiencing many serious system
instabilities (random file corruption, system crashes). Basically memtest
would report thousand of errors in a matter of minutes.Turned out the the
problem was caused by a faulty Bios setting (memory was set to 'performance
mode'. Whatever that means) After setting the value to default, the problem
was corrected.

But now I'm still getting an error. A single one this time. Happens on test
7 always at the same address (0000006952c).
The error occured semi-randomly it seems (but always at the same address)
every 2 passes or so on average.

So I wonder if this is normal. Is this a sign a defective/"incompatible"
Ram?

 

[Alceryes]

So I wonder if this is normal. Is this a sign a defective/"incompatible"

Ram?

You should receive absolutely NO errors no matter how many passes it makes.
It sounds like the RAM might be defective. You might also want to change the
BIOS setting from 'default' to custom and manually enter the timings your
RAM is rated at.
--


"I don't cheat to survive. I cheat to LIVE!!"
- Alceryes

 

[w_tom]

Any one error from any semiconductor inside the computer -
and your system has crashed. Not only should memory test just
fine every time, but it should also test just fine even when
heated with a hairdryer on highest setting. If in doubt, then
look at the operating temperature from memory datasheets -
well above what your hand (or hair) will call comfortable.
But a defective memory sometimes may fail Memtst86 (and other
memory tests) only when at elevated (and perfectly normal)
temperatures.

Heat finds weak memory before it causes hard failures
later. Unfortunately some never learned this and install more
fans to cure a defective semiconductor problem. Pig's Heaven
to memory is a temperature at the uppermost and lowermost
temperature numbers on that datasheet. Any memory that fails
at the 'paradise perfect' temperature of a hairdryer is 100%
defective - even if only one memory location fails 1 in 100
tests. Hairdryer tends to find defective (intermittent) memory
faster.

To work, every essential digital IC in a computer must work
correctly every time - which is why we use digital logic and
don't use analog logic circuits. In computers, no tolerance
for digital logic errors - especially from memory.

 

[John-Y]

I recently used memtest86 after experiencing many serious system
instabilities (random file corruption, system crashes). Basically memtest
would report thousand of errors in a matter of minutes.Turned out the the
problem was caused by a faulty Bios setting (memory was set to 'performance
mode'. Whatever that means) After setting the value to default, the problem
was corrected.

But now I'm still getting an error. A single one this time. Happens on
test 7 always at the same address (0000006952c).
The error occured semi-randomly it seems (but always at the same address)
every 2 passes or so on average.

So I wonder if this is normal. Is this a sign a defective/"incompatible"
Ram?

Thanks to Alceryes and w_tom for the informations.


I brought the PC to the place from where I got it so they can have a look a
it. I explained what's wrong it,the error reported by memtest and
everything. Guy said: "all right I'll have a look". Anyway, a week later the
guy tells me he fixed the problem, said it was "related to gfx card". Well
he fixed it my a**... Problem is still present.(at least he didn't charged
me.)


But now,I'm observing some other problems...So far, this is what I've seen:

-The PC does not always boot. That is, it "powers up" fine but the bios
won't execute.So nothing happens.No choice but to turn off and restart. This
seem to happen more frequently when the Pc is "cold" (when it's been off for
more than an hour iow)

-After the above failure to boot, it'll report: "CMOS checksum
error.Returning to default". Regardless, after a while the problem will
reappear.

-Memtest fails the "Memory sizing - Bios All" test
(configuration,option3,option2) Reports nothing but errors.

-And of course the error reported on test 7 at address 0000006952 like I
said above...




Does the above shed new lights on what may cause the problem? And no, I
haven't changed the Ram module yet...(I kinda hoped the "repair guy" would
do so)

 

[John-Y]

-The PC does not always boot. That is, it "powers up" fine but the bios
won't execute.So nothing happens.No choice but to turn off and restart.
This seem to happen more frequently when the Pc is "cold" (when it's been
off for more than an hour iow)

-After the above failure to boot, it'll report: "CMOS checksum
error.Returning to default". Regardless, after a while the problem will
reappear.

-Memtest fails the "Memory sizing - Bios All" test
(configuration,option3,option2) Reports nothing but errors.

-And of course the error reported on test 7 at address 0000006952 like I
said above...

Forgot the most important one:

-PC crashes randomly.And this under almost any software
environments,including during installation of Windows XP,something I've
never experienced before. On average,it'll take 15 to 30 minutes to get a
crash.

 

[kony]

Forgot the most important one:

-PC crashes randomly.And this under almost any software
environments,including during installation of Windows XP,something I've
never experienced before. On average,it'll take 15 to 30 minutes to get a
crash.

Then it's still instable. Inform the seller that you need
this resolved asap. Their idea about it being the video
card seems like nonsense, but you might find a polite way to
suggest that can't be it since the problem still occurs.

You could try a newer motherboard bios but most likely the
resolution will be a different (make) of memory... not just
a swap for another of exact same thing.

 

[John-Y]

Then it's still instable. Inform the seller that you need
this resolved asap. Their idea about it being the video
card seems like nonsense, but you might find a polite way to
suggest that can't be it since the problem still occurs.

You could try a newer motherboard bios but most likely the
resolution will be a different (make) of memory... not just
a swap for another of exact same thing.

Yeah,first thing I'll try is to flash the bios. Then I guess I'll have to
try another Ram module...

Yesterday I brought the comp to another sellers who also do "repairs". The
cost of the estimation was quite low at least..One thing he said when I told
him about the Cmos checksum error message is that,according to him, it was
related to the PC time clock...Does that even make sense?

 

[kony]

Yeah,first thing I'll try is to flash the bios. Then I guess I'll have to
try another Ram module...

Yesterday I brought the comp to another sellers who also do "repairs". The
cost of the estimation was quite low at least..One thing he said when I told
him about the Cmos checksum error message is that,according to him, it was
related to the PC time clock...Does that even make sense?

You never mentioned the hardware, a primary requirement in a
hardware group.

CMOS checksums are not related to the time clock. If you
had a bad battery, or had manually or inadventantly reset
the CMOS, then the clock would be too, but there is no
significance in trying to relate them beyond this. Check
your battery voltage.

 

[John-Y]

You never mentioned the hardware, a primary requirement in a
hardware group.

Yeah sorry. It's a foxconn 661M03 Motherboard with a SiS 661FX+ 963 chipset.
P4 at 2.4ghz (not overclocked). Single Ram module: Samsung 512M PC3200
DDR400

Short recap from last days: Got my computer back, guy said the problem was
caused by the Ram (now, did he knew was he was talking about or did he just
took an educated guess, I don't know) The Ram was not changed. I tested with
Memtest for more than 12hours, no error occured...I thought "hallujah, the
problem is miraculously solved!" (wishful thinking heh)

Now,the problem is more serious than ever...When I try to boot,I just
receive an "Verifying DMI pooll data"...then nothing. It just stays there.
So I can't possibly boot the PC, be it via floppy,harddrive or cd-rom...I
can access the bios configuration but that's all.

I did flashed the bios in the hope that it would cure the problem. I did had
the latest bios allready though (27dec2004) so I just flashed the same
version over it. After the flashed,everything seem to worked fine and then
I did the memtest thing for 12hours like I said.


Anyway, I'm gonna try another ram module and if that doesn't work, I'm all
out of ideas.

CMOS checksums are not related to the time clock. If you
had a bad battery, or had manually or inadventantly reset
the CMOS, then the clock would be too, but there is no
significance in trying to relate them beyond this. Check
your battery voltage.

How do I check the battery voltage?

 

[kony]

Yeah sorry. It's a foxconn 661M03 Motherboard with a SiS 661FX+ 963 chipset.
P4 at 2.4ghz (not overclocked). Single Ram module: Samsung 512M PC3200
DDR400

If it will allow underclocking the FSB and/or memory bus,
try lowering it to see if that regains any stability.

Short recap from last days: Got my computer back, guy said the problem was
caused by the Ram (now, did he knew was he was talking about or did he just
took an educated guess, I don't know) The Ram was not changed. I tested with
Memtest for more than 12hours, no error occured...I thought "hallujah, the
problem is miraculously solved!" (wishful thinking heh)

Why not just ask him what/why/etc?

Now,the problem is more serious than ever...When I try to boot,I just
receive an "Verifying DMI pooll data"...then nothing. It just stays there.
So I can't possibly boot the PC, be it via floppy,harddrive or cd-rom...I
can access the bios configuration but that's all.

So memory is a primary suspect. The other two more common
would be overheating or power supply. You might Google
search for errata on your motherboard model though, perhaps
there is some other oddity to it.

I did flashed the bios in the hope that it would cure the problem. I did had
the latest bios allready though (27dec2004) so I just flashed the same
version over it. After the flashed,everything seem to worked fine and then
I did the memtest thing for 12hours like I said.

It's a bit of a catch-22 when it comes to flashing a bios to
resolve memory issues... or stablity in general. It's not
good to do it with the instable system as it's being read
into (potentially corruptable) memory and thus the bios
itself could become corrupt during the process. Perhaps not
totally corrupt, even a bit here or there could be a
nightmare to find later. If you can set the bios to lowest
FSB & memory speed, and possibly the most consertive memory
timings (larger numbers), AND if stability isnt regained and
your temps and power seem OK, I'd try reflashing the bios
but perhaps to the last version... occasionally a new bios
will have as-yet-undiscovered bugs.

Anyway, I'm gonna try another ram module and if that doesn't work, I'm all
out of ideas.


How do I check the battery voltage?

EIther it's in the bios "health monitor" or similar screen
or pull it and use a multimeter to get a reading.

If you had not cleared CMOS after flashing the bios, unplug
the system and do that, first.

 

[John-Y]

Ok...before this thread gets any more chaotic and confusing (I only blame
myself) I'll try to make this as short and clear as possible:



("short" detailed version

-The "Verifying DMI pool data" issue has been solved. Noting to have with
the Ram module,it was caused by a "wrong" bio setting...*hit myself with
palm on the forehead*

-I forgot to mention that immediately after the successful 12hour memtest,
the error did reappeared...So that problem never really went away...

-I finally have access to a new Ram module. I'll test with memtest later
(I'm gonna test it over night) to see if that solve the crashing/memtest
error problem.

-This is the issue I'm currently most concerned with:

"The PC does not always boot. That is, it "powers up" fine but the bios
won't execute.So nothing happens.No choice but to turn off and restart. This
seem to happen more frequently when the Pc is "cold" (when it's been off for
more than an hour iow)
After the above failure to boot, it'll report: "CMOS checksum
error.Returning to default". Regardless, after a while the problem will
reappear."

Extremely annoying and weird...What could cause this?Corrupt bios? Faulty
cmos battery? Needless to say, the new Ram did not solved this problem...





("cut the crap and get to the point" version)

Problem1) Computer crash(ed?). Need to confirm if new ram solve problem.

Problem2) Sometimes when PC is turned on/reset, nothing happens. New ram
didn't solved problem.


That's it. Thanks for the answer so far,an awful thread I concede

If it will allow underclocking the FSB and/or memory bus,
try lowering it to see if that regains any stability.

Afaik, 2.4ghz/ 133FSB is the lowest the system will go so I can't
underclock. Tried the most conservative/highest values for memories too, no
changes noticed.

If you had not cleared CMOS after flashing the bios, unplug
the system and do that, first.

That's what I did after the flash, cleared it with the jumper. Problem no2
still present.

 

[kony]

"The PC does not always boot. That is, it "powers up" fine but the bios
won't execute.So nothing happens.No choice but to turn off and restart. This
seem to happen more frequently when the Pc is "cold" (when it's been off for
more than an hour iow)
After the above failure to boot, it'll report: "CMOS checksum
error.Returning to default". Regardless, after a while the problem will
reappear."

Extremely annoying and weird...What could cause this?Corrupt bios? Faulty
cmos battery? Needless to say, the new Ram did not solved this problem...


("cut the crap and get to the point" version)

Problem1) Computer crash(ed?). Need to confirm if new ram solve problem.

Problem2) Sometimes when PC is turned on/reset, nothing happens. New ram
didn't solved problem.


That's it. Thanks for the answer so far,an awful thread I concede

Tackle each issue one at a time. Memory with random,
infrequent errors will not "usually" prevent a system from
turning on, POSTing at all. I would suspect motherboard
problem or power. You'd mentioned the bios setting- are all
settings now at default values? Have you tried
underclocking the system to see if that improves it's odds
of POSTing?

As for power supply (and battery) take voltage readings with
a multimeter. Get system POSTing 100% of the time and then
see if there are still any (seemingly) memory-related issues
remaining.