Memory puzzle

  • Thread starter Thread starter Russ M.
  • Start date Start date
R

Russ M.

Yesterday, with everyone's help, I solved a big piece of my system's
instability due to bad CPU cooling. But unfortunately, some instability
still exists. It appears to be related to the memory so I've been playing
around with it all day along, taking notes along the way. I'm wondering if
anyone sees my findings unusual or if any thoughts or advice come to mind.

MB=Chaintech 7VIF4 (2 memory banks)
CPU=AMD XP Barton 2600+ (runs at FSB-166, 49C temp at full load, 42C at
idle)
Module#1=PNY(Samsung) 256MB PC2700 DDR333
Module#2=Centon 256MB PC2700 DDR333
2-80GB WD Hard drives (On same cable)
1 DVD R/W on secondary
New case with (supposedly) premium quality power supply

Here's what I've found:

1) PNY in bank 1. Not workable. Frequent instantaneous reboots in Win2K
when copying from 1 HD to the other or when reading from the DVD. Sometimes
brings up the scanner software upon Win boot-up. Random programs can't
initialize/start. Manually re-boot and the software problems are solved.
This is by far the worst configuration. Fails very quickly.
2) PNY in bank 2. Works much better. Only one phantom re-boot during
Sandra 2004 burn-in after 3 hrs.
3) Centon in bank 1. A phantom re-boot after an hour of Sandra 2004
burn-in. After re-boot, ran fine for 3 hrs.
4) Centon in bank 2. Not a single problem after 6 hours of Sandra 2004
burn-in and still running. Before I started taking notes and doing more
testing, it also ran 9 hrs overnight without a failure.
5) Together, either bank. Results similar to #1 above. Fails quick and it
doesn't matter what the bank configuration is.
6) Everything seems to work great at FSB-133. But I haven't done any
rigorous testing as I have with 166. I just know that the system worked
well under constant load for 2 straight days, even with the CPU running at
60C.

For testing, I copied 30GB from 1 drive to the other repeatedly. Copied a
DVD disk to HD. Ran Sandra 2004 CPU/memory burn-in continuously. These are
the things that I know causes the phantom reboots.

So does anyone have any thoughts or ideas? Is it fairly common to have
memory this touchy? Any other diag/testing tools I might try to get more
info? I'm kind of surprised that the Centon is more stable than the PNY
(Samsung). I would have bet it would have been the other way around.
Tells you how much I know :)

Thanks for any input/advice....
 
Russ M. said:
Yesterday, with everyone's help, I solved a big piece of my system's
instability due to bad CPU cooling. But unfortunately, some instability
still exists. It appears to be related to the memory so I've been playing
around with it all day along, taking notes along the way. I'm wondering if
anyone sees my findings unusual or if any thoughts or advice come to mind.

MB=Chaintech 7VIF4 (2 memory banks)
CPU=AMD XP Barton 2600+ (runs at FSB-166, 49C temp at full load, 42C at
idle)
Module#1=PNY(Samsung) 256MB PC2700 DDR333
Module#2=Centon 256MB PC2700 DDR333
2-80GB WD Hard drives (On same cable)
1 DVD R/W on secondary
New case with (supposedly) premium quality power supply

Here's what I've found:

1) PNY in bank 1. Not workable. Frequent instantaneous reboots in Win2K
when copying from 1 HD to the other or when reading from the DVD. Sometimes
brings up the scanner software upon Win boot-up. Random programs can't
initialize/start. Manually re-boot and the software problems are solved.
This is by far the worst configuration. Fails very quickly.
2) PNY in bank 2. Works much better. Only one phantom re-boot during
Sandra 2004 burn-in after 3 hrs.
3) Centon in bank 1. A phantom re-boot after an hour of Sandra 2004
burn-in. After re-boot, ran fine for 3 hrs.
4) Centon in bank 2. Not a single problem after 6 hours of Sandra 2004
burn-in and still running. Before I started taking notes and doing more
testing, it also ran 9 hrs overnight without a failure.
5) Together, either bank. Results similar to #1 above. Fails quick and it
doesn't matter what the bank configuration is.
6) Everything seems to work great at FSB-133. But I haven't done any
rigorous testing as I have with 166. I just know that the system worked
well under constant load for 2 straight days, even with the CPU running at
60C.

For testing, I copied 30GB from 1 drive to the other repeatedly. Copied a
DVD disk to HD. Ran Sandra 2004 CPU/memory burn-in continuously. These are
the things that I know causes the phantom reboots.

So does anyone have any thoughts or ideas? Is it fairly common to have
memory this touchy? Any other diag/testing tools I might try to get more
info? I'm kind of surprised that the Centon is more stable than the PNY
(Samsung). I would have bet it would have been the other way around.
Tells you how much I know :)

Thanks for any input/advice....
Click to expand...

Your problems are typically caused by a bad power supply. You ALSO had a
CPU running too hot, until you got that ironed out. I think it's possible
that your power supply has been bad all along. It's either that, or you've
got two bad memory sticks. While somewhat unlikely, that's possible.
Another possibility is that the CPU was damaged by running too hot. Best
guess is poor power, though. -Dave
 
Thanks Dave,

You don't think it could be the motherboard? I initially thought it was
being caused by a poor power supply, so I bought a new case w/350w psu. It
acted the same way. The folks I bought the case from have been in biz for a
long time and they do a great job. But that's not to say you're not
correct. I don't even know what to say to them.

So why would going from 133(x2) to 166(x2) tax the PSU so much? I'm afraid
my lack of knowledge is haunting me.

Thanks again.....
 
Russ said:
Thanks Dave,

You don't think it could be the motherboard? I initially thought it was
being caused by a poor power supply, so I bought a new case w/350w psu. It
acted the same way. The folks I bought the case from have been in biz for a
long time and they do a great job. But that's not to say you're not
correct. I don't even know what to say to them.

So why would going from 133(x2) to 166(x2) tax the PSU so much? I'm afraid
my lack of knowledge is haunting me.
Click to expand...

All else being equal, CPU power consumption increases linearly with
increasing speed.
 
Russ M. said:
Thanks Dave,

You don't think it could be the motherboard? I initially thought it was
being caused by a poor power supply, so I bought a new case w/350w psu. It
acted the same way. The folks I bought the case from have been in biz for a
long time and they do a great job. But that's not to say you're not
correct. I don't even know what to say to them.

So why would going from 133(x2) to 166(x2) tax the PSU so much? I'm afraid
my lack of knowledge is haunting me.

Thanks again.....
Click to expand...


Well it COULD be the motherboard, but I hope it isn't. Instability problems
like you posted before are usually caused by poor power, unless there is
some other obvious cause, such as an overheating CPU. The source of power
is (obviously) the power supply, but the mainboard also helps regulate
power. Any component that runs faster will draw more power. If the voltage
is low or unstable to begin with, speeding things up will not help any. (!)

Generally, power supplies that are included with cases are crap. There are
some exceptions, but they are definitely exceptions, and not the rule.
Without knowing who made your power supply and what model number it is, I
can't comment on your specific power supply. However, the typical generic
350W may be weaker than a good brand name 200W. In fact, the "350W" means
nothing, as that is a theoretical maximum power output. What counts is how
much current can be delivered continuously to each rail. If a computer is
starved for +5V power (for example), it won't matter that the other voltage
rails are strong enough to theoretically add up to 350W total. That's why
there can be such a great difference between power supplies of the same
stated maximum wattage. If you'd like a recommendation for a decent power
supply that won't cost an arm and a leg, I could recommend you try the
following. (see url) That's not a premium brand, but it gets good reviews
and has decent specifications.

You're in a tough situation. My gut is telling me you need a new power
supply. You've already tried that, which makes it somewhat LESS likely that
the power supply is bad. But if it's not the power supply, this one could
be very difficult to nail down. Before you try another power supply, find a
KNOWN GOOD stick of RAM (borrow one from somebody?) and try that, installed
without any other RAM stick in the system. That will tell you whether the
RAM is bad or not. It's still possible that both sticks of RAM are bad. If
you eliminate RAM and power supply as possible suspects, then you are down
to motherboard or CPU, with motherboard being the more likely of those two.

I still think it's power supply, though. Instable voltage can cause the
symptoms you were describing. Disk drives spinning up can exaggerate
existing instable voltage problems. I think you were writing that the
problem happens more often when using your various disk drives, right? Good
luck on this one. -Dave

http://www.newegg.com/app/ViewProductDesc.asp?description=17-103-604&depa=0
 
Russ M. said:
MB=Chaintech 7VIF4 (2 memory banks)
CPU=AMD XP Barton 2600+ (runs at FSB-166, 49C temp at full load, 42C at
idle)
Module#1=PNY(Samsung) 256MB PC2700 DDR333
Module#2=Centon 256MB PC2700 DDR333
2-80GB WD Hard drives (On same cable)
1 DVD R/W on secondary
New case with (supposedly) premium quality power supply
Click to expand...
<snip>

hi. i was just searching in hopes of solving a very similar problem
with a very similar setup. heres what i've got:

MB: Chaintech 7VIF4
CPU: AMD XP 2600+ Barton core
512MB Ultra PC2700 DDR333
512MB Crucial PC2700 DDR333
425W Power supply (Just 4 PC)
Thermaltake TR 2 M4 CPU fan

like you, i just put this system together, and immediately started
running into problems that seem like memory issues. if either stick
of ram is in either slot i will get badram errors in minutes when i
run memtest-86 (tests 6 and 10 using the modulo-x algorithm). it is
completely unstable. i can't hardly get through a linux kernel
compile without it giving me mysterious segfaults or worse. i have
tried everything i can think of to make it work... ...at 166 fsb.

here's the kicker. everything seems solid at 133 fsb. i haven't been
able to replicate any of the errors. my cooling setup and power
supply is more than what i should need, especially when just running
memtest-86 and no hard drives. i haven't completely eliminated the
other possibilities, but i am beginning to think that the 7VIF4 can't
really handle 166 fsb.

i think that we should combine all of our results from our various
experiments. please e-mail me directly.

Steve Asher
 
<snip>

hi. i was just searching in hopes of solving a very similar problem
with a very similar setup. heres what i've got:

MB: Chaintech 7VIF4
CPU: AMD XP 2600+ Barton core
512MB Ultra PC2700 DDR333
512MB Crucial PC2700 DDR333
425W Power supply (Just 4 PC)
Thermaltake TR 2 M4 CPU fan

like you, i just put this system together, and immediately started
running into problems that seem like memory issues. if either stick
of ram is in either slot i will get badram errors in minutes when i
run memtest-86 (tests 6 and 10 using the modulo-x algorithm). it is
completely unstable. i can't hardly get through a linux kernel
compile without it giving me mysterious segfaults or worse. i have
tried everything i can think of to make it work... ...at 166 fsb.

here's the kicker. everything seems solid at 133 fsb. i haven't been
able to replicate any of the errors. my cooling setup and power
supply is more than what i should need, especially when just running
memtest-86 and no hard drives. i haven't completely eliminated the
other possibilities, but i am beginning to think that the 7VIF4 can't
really handle 166 fsb.

i think that we should combine all of our results from our various
experiments. please e-mail me directly.

Steve Asher
Click to expand...
I just built 4 of these systems for a customer and they do not work at
166 fsb. They work perfect with the 133 fsb. I have used a program
called goldmemory to test the memory at 166 and get hundreds of
errors. At 133 I do not get any errors. I have tried 4 different power
supplies, 12 different sticks of memory including Crucial, and
Kingston. I have tried 333 and even 400 speed memory. I have tried
playing with all the memory settings in the BIOS and still no good
unless I run my Barton 2600 at 133FSB which then gives me a Athlon
2000. I wished I knew about this problem before I made my purchase. I
got them from Newegg and the customer reviews were all pretty good.
Does anybody know of a VIA KM400 chipset motherboard that works well
at 166 FSB? I might go back to the Biostar with the Nvidia chipset
with onboard geforce video, but they cost a lot more.
BEWARE of the CHAINTECH motherboard 166FSB problem before you spend
your money!!!
Rick
 
Back
Top