Stability Problems--CPU or MOBO? How To Fix? (Intel D915, ECS Mobo)

[James G]

CPU: Intel Pentium D 915 [2.8GHz]
MOBO: ECS P4M800PRO-M
RAM: Kingston 2GB PC4200 533MHz

Problem: When using After Effects, I've been experiencing restarts,
with Microsoft reporting a "Hardware failure: replace failed component"
message and further stating that the problem could lie with the RAM,
MOBO, CPU or PS.

I allowed MemTest to run overnight for zero errors. I then tried
Prime95, first with the "In-place large FFTs" test and I got an error
in just a few minutes of testing. It reported a "FATAL ERROR: Rounding"
message. I then tried Prime95's "Small FFTs" test and again got an
error after just a couple of minutes (this time reporting a "Resulting
sum was XXX, expected ZZZ" error).

So it seems that my RAM is good and the problem is most likely with the
CPU, yes? Is there anything I can do at this point besides replacing
the CPU? I'm not particularly BIOS savvy, but I do take direction well
and I'm willing to give tweaking a try if anyone thinks that this might
help.

All I want is a stable, non-crashing machine. I don't care if I don't
squeeze every last ounce of oomph out of the CPU.

Thanks for any and all help!

 

[Rod Speed]

CPU: Intel Pentium D 915 [2.8GHz]
MOBO: ECS P4M800PRO-M
RAM: Kingston 2GB PC4200 533MHz

Problem: When using After Effects, I've been experiencing
restarts, with Microsoft reporting a "Hardware failure:
replace failed component" message and further stating that
the problem could lie with the RAM, MOBO, CPU or PS.

I allowed MemTest to run overnight for zero errors. I then tried
Prime95, first with the "In-place large FFTs" test and I got an error
in just a few minutes of testing. It reported a "FATAL ERROR:
Rounding" message. I then tried Prime95's "Small FFTs" test and
again got an error after just a couple of minutes (this time reporting
a "Resulting sum was XXX, expected ZZZ" error).

So it seems that my RAM is good and the
problem is most likely with the CPU, yes?

Nope, more likely to be a ram problem. CPUs dont normally die like
that unless you have been stupid with the voltage when overclocking etc.

Is there anything I can do at this point besides replacing the CPU?

Monitor the cpu temp while running Prime95, it may have a fan
clogged with fur or may not have had the heatsink installed properly.

I'm not particularly BIOS savvy, but I do take direction well and I'm
willing to give tweaking a try if anyone thinks that this might help.

 

[James G]

CPU: Intel Pentium D 915 [2.8GHz]
MOBO: ECS P4M800PRO-M
RAM: Kingston 2GB PC4200 533MHz

Problem: When using After Effects, I've been experiencing
restarts, with Microsoft reporting a "Hardware failure:
replace failed component" message and further stating that
the problem could lie with the RAM, MOBO, CPU or PS.

I allowed MemTest to run overnight for zero errors. I then tried
Prime95, first with the "In-place large FFTs" test and I got an error
in just a few minutes of testing. It reported a "FATAL ERROR:
Rounding" message. I then tried Prime95's "Small FFTs" test and
again got an error after just a couple of minutes (this time reporting
a "Resulting sum was XXX, expected ZZZ" error).

So it seems that my RAM is good and the
problem is most likely with the CPU, yes?

Nope, more likely to be a ram problem. CPUs dont normally die like
that unless you have been stupid with the voltage when overclocking etc.

But I did allow MemTest to run for something close to 20 hours and it
reported no errors when checking my RAM.

These components (CPU, MOBO, RAM) are all new and this machine has had
this problem from the get-go. I'm not an overclocker so I haven't
messed with anything like the voltage, etc.

Monitor the cpu temp while running Prime95, it may have a fan
clogged with fur or may not have had the heatsink installed properly.

I'll definitely check the integrity of the components, but (as I said),
the stuff is less than 30 days old and is being run in a non-carpted,
clean environment so I'm sure it's not a fan issue.

It might be a CPU temperature thing--what's the best way to monitor the
CPU temp?

Thanks!

 

[Rod Speed]

Rod Speed wrote

CPU: Intel Pentium D 915 [2.8GHz]
MOBO: ECS P4M800PRO-M
RAM: Kingston 2GB PC4200 533MHz
Problem: When using After Effects, I've been experiencing
restarts, with Microsoft reporting a "Hardware failure:
replace failed component" message and further stating that
the problem could lie with the RAM, MOBO, CPU or PS.
I allowed MemTest to run overnight for zero errors. I then tried
Prime95, first with the "In-place large FFTs" test and I got an
error in just a few minutes of testing. It reported a "FATAL ERROR:
Rounding" message. I then tried Prime95's "Small FFTs" test and
again got an error after just a couple of minutes (this time
reporting a "Resulting sum was XXX, expected ZZZ" error).
So it seems that my RAM is good and the
problem is most likely with the CPU, yes?

Nope, more likely to be a ram problem. CPUs dont normally die like that
unless you have been stupid with the voltage when overclocking etc.

But I did allow MemTest to run for something close to 20
hours and it reported no errors when checking my RAM.

Yes, but it doesnt always detect a ram compat problem.

There have been systems that fail with Prime95
and which work fine with different ram.

These components (CPU, MOBO, RAM) are all new
and this machine has had this problem from the get-go.

OK, in that case its most likely that the bios isnt using
the correct timing detail for the ram being used.

I'm not an overclocker so I haven't messed
with anything like the voltage, etc.

OK, then the cpu is fine and you have a problem
with the speed detail that the bios is running the
ram at if it isnt the temperature due to the heatsink
not being mounted on the cpu properly.

I'll definitely check the integrity of the components,

Its not a good idea to remove the heatsink to check that without
evidence that the cpu is getting hotter than it should do.

but (as I said), the stuff is less than 30 days old and is being run
in a non-carpted, clean environment so I'm sure it's not a fan issue.

Sure, but you've only just provided that detail now. We cant read minds.

It might be a CPU temperature thing--what's
the best way to monitor the CPU temp?

There are a number of utes that can do that, speedfan is easy
to use and everest will show the various temps available too.
http://www.majorgeeks.com/download.php?det=4181

 

[Dave]

CPU: Intel Pentium D 915 [2.8GHz]
MOBO: ECS P4M800PRO-M
RAM: Kingston 2GB PC4200 533MHz

Problem: When using After Effects, I've been experiencing restarts,
with Microsoft reporting a "Hardware failure: replace failed component"
message and further stating that the problem could lie with the RAM,
MOBO, CPU or PS.

I allowed MemTest to run overnight for zero errors. I then tried
Prime95, first with the "In-place large FFTs" test and I got an error
in just a few minutes of testing. It reported a "FATAL ERROR: Rounding"
message. I then tried Prime95's "Small FFTs" test and again got an
error after just a couple of minutes (this time reporting a "Resulting
sum was XXX, expected ZZZ" error).

So it seems that my RAM is good and the problem is most likely with the
CPU, yes? Is there anything I can do at this point besides replacing
the CPU? I'm not particularly BIOS savvy, but I do take direction well
and I'm willing to give tweaking a try if anyone thinks that this might
help.

All I want is a stable, non-crashing machine. I don't care if I don't
squeeze every last ounce of oomph out of the CPU.

Thanks for any and all help!

The vast majority of problems like this are caused by a bad power supply.
However, you've got an ECS mainboard, which is known to be crap. Your RAM
is a good brand, should be no problems there. CPUs do go bad, but so rarely
that it's not worth considering. Try a GOOD brand of power supply (fortron
or enermax or seasonic are three good suggestions). If that doesn't work,
get yourself an epox or biostar brand mainboard. I suggest those two brands
as they are both very stable and both usually inexpensive. -Dave

 

[James G]

The vast majority of problems like this are caused by a bad power supply.
However, you've got an ECS mainboard, which is known to be crap. Your RAM
is a good brand, should be no problems there. CPUs do go bad, but so rarely
that it's not worth considering. Try a GOOD brand of power supply (fortron
or enermax or seasonic are three good suggestions). If that doesn't work,
get yourself an epox or biostar brand mainboard. I suggest those two brands
as they are both very stable and both usually inexpensive. -Dave

I should have mentioned this in my first post, but the case (Antec) and
the power supply are not new components. So unless the power supply
decided to die just as I did this upgrade, I think we can rule it out.

I fully understand your position on ECS. I definitely know their
reputation, but this particular board was one of the few that I could
find for this CPU that would allow me to use my AGP video card (a
dual-DVI Matrox P750). However, if this indeed is the issue, I will
definitely consider replacing it (but I really hate losing this video
card--it's perfectly suited for my needs and any replacement on par
with it will cost me double the motherboard at least).

In your (or anyone else's) opinion, should I consider replacing the RAM
with DDR PC3200? I am using PC4200 because I mistakenly thought that
533MHz was the proper FSB for this CPU and it is instead 400MHz. The
RAM I have now is DDR2 PC4200 (533MHz). Could this be the problem?

 

[James G]

OK, in that case its most likely that the bios isnt using
the correct timing detail for the ram being used.


OK, then the cpu is fine and you have a problem
with the speed detail that the bios is running the
ram at if it isnt the temperature due to the heatsink
not being mounted on the cpu properly.

Ah, I see--so it definitely could be the RAM running at a FSB of 533MHz
while the CPU is running at 400MHz? Can I throttle the RAM down to run
at the same speed or should I replace the RAM?

Its not a good idea to remove the heatsink to check that without
evidence that the cpu is getting hotter than it should do.


Sure, but you've only just provided that detail now. We cant read minds.


There are a number of utes that can do that, speedfan is easy
to use and everest will show the various temps available too.
http://www.majorgeeks.com/download.php?det=4181

I'm definitely going to run that ASAP. Thanks!

 

[Rod Speed]

Rod Speed wrote

Ah, I see--so it definitely could be the RAM running at
a FSB of 533MHz while the CPU is running at 400MHz?

No, I didnt mean that. There is a lot finer control over the detail
of how the ram is driven than that, configurable in the bios.
Modern systems are supposed to work that out for themselves,
using the spd on the ram sticks, but particularly with DDR2 ram,
that is still a real problem, doing that reliably. Thats almost certainly
your problem. Best to check you are using the latest bios first.

One of the problems with the bleeding edge, quite a bit of bleeding.

Can I throttle the RAM down to run at the same speed

Shouldnt be necessary.

or should I replace the RAM?

I'd try the latest bios first. If that doesnt help, its likely possible
to use more conservative timing detail to eliminate the problem
until the motherboard manufacturer gets its act into gear.

I'm definitely going to run that ASAP. Thanks!

Its almost certainly the DDR2 problem if that is what sort of memory you are using.

 

[Rod Speed]

I should have mentioned this in my first post, but the case (Antec)
and the power supply are not new components. So unless the power
supply decided to die just as I did this upgrade, I think we can rule
it out.

I fully understand your position on ECS. I definitely know their
reputation, but this particular board was one of the few that I could
find for this CPU that would allow me to use my AGP video card (a
dual-DVI Matrox P750). However, if this indeed is the issue, I will
definitely consider replacing it (but I really hate losing this video
card--it's perfectly suited for my needs and any replacement on par
with it will cost me double the motherboard at least).

In your (or anyone else's) opinion, should I consider replacing the
RAM with DDR PC3200? I am using PC4200 because I mistakenly thought
that 533MHz was the proper FSB for this CPU and it is instead 400MHz.
The RAM I have now is DDR2 PC4200 (533MHz). Could this be the problem?

Yes, its almost certainly the problem. I'd check for the latest bios available
and if that doesnt fix it, certainly using DDR ram should fix the problem.

 

[Dave]

I should have mentioned this in my first post, but the case (Antec) and
the power supply are not new components. So unless the power supply
decided to die just as I did this upgrade, I think we can rule it out.

Not so fast. A good power supply can be over-taxed by new components. Keep
in mind that power supplies are more than wattage suppliers. It's
possible that one or more rails of your "old" power supply is running near
its limit with the new hardware it is trying to push. That's quite common,
actually. People love to upgrade, but often fail to consider the power
supply can't handle the planned upgrade.

I fully understand your position on ECS. I definitely know their
reputation, but this particular board was one of the few that I could
find for this CPU that would allow me to use my AGP video card (a
dual-DVI Matrox P750). However, if this indeed is the issue, I will
definitely consider replacing it (but I really hate losing this video
card--it's perfectly suited for my needs and any replacement on par
with it will cost me double the motherboard at least).

In your (or anyone else's) opinion, should I consider replacing the RAM
with DDR PC3200? I am using PC4200 because I mistakenly thought that
533MHz was the proper FSB for this CPU and it is instead 400MHz. The
RAM I have now is DDR2 PC4200 (533MHz). Could this be the problem?

If it's really DDR2, then YES, that is the problem. If you put DDR533 RAM
in a DDR400 mainboard, it should work OK. No guarantee that it will, but it
SHOULD work. But DDR2 is an entirely different animal. -Dave

 

[James G]

Not so fast. A good power supply can be over-taxed by new components. Keep
in mind that power supplies are more than wattage suppliers. It's
possible that one or more rails of your "old" power supply is running near
its limit with the new hardware it is trying to push. That's quite common,
actually. People love to upgrade, but often fail to consider the power
supply can't handle the planned upgrade.


If it's really DDR2, then YES, that is the problem. If you put DDR533 RAM
in a DDR400 mainboard, it should work OK. No guarantee that it will, but it
SHOULD work. But DDR2 is an entirely different animal. -Dave

I just upgraded the BIOS and I have the same problems.

I think I'm going to try to hunt down some of the prescribed RAM
(DDR400) and see what happens.

Hopefully I can do it this weekend.

 

[James G]

Not so fast. A good power supply can be over-taxed by new components. Keep
in mind that power supplies are more than wattage suppliers. It's
possible that one or more rails of your "old" power supply is running near
its limit with the new hardware it is trying to push. That's quite common,
actually. People love to upgrade, but often fail to consider the power
supply can't handle the planned upgrade.

That makes sense, too. I think I'm going to start with the RAM first
and see where that gets me. After that, I'll take a look at the power
supply.

Thanks a lot for the advice and help!

 

[James G]

Not so fast. A good power supply can be over-taxed by new components. Keep
in mind that power supplies are more than wattage suppliers. It's
possible that one or more rails of your "old" power supply is running near
its limit with the new hardware it is trying to push. That's quite common,
actually. People love to upgrade, but often fail to consider the power
supply can't handle the planned upgrade.


If it's really DDR2, then YES, that is the problem. If you put DDR533 RAM
in a DDR400 mainboard, it should work OK. No guarantee that it will, but it
SHOULD work. But DDR2 is an entirely different animal. -Dave

Argh. I bought some PC3200 and my machine still failed the Prime95
test. The "in-place large FFTs" test brought a failure on Test 3 (I
guess this is an improvement--the old memory brought a failure on Test
1.

Okay, so on to the next most likely culprit. It's down to the power
supply, the mobo or the cpu. Based on Dave's suggestion, I will next
try the power supply and then the mobo.

Thanks for the help, guys.