Old Dual-Pentium3 Woes...

[Eric]

Hi,

I'm throwing this out in hopes that someone may suggest what I could try
next. One of my old Dual-P3's that I wish to keep in service has been
giving me problems lately. I know it has to be hardware because it gives
problems in Windows and Linux. Under Windows, during bootup it often gives
blue screens such as IRQL_DRIVER_NOT_LESS_OR_EQUAL, PAGING_IN_NONPAGED_AREA,
etc. These aren't verbatim, but as close as I remember. Under Linux, it
gets kernel panics during bootups.

Anyway, I tested just about everything. All the hardware cards are fine,
the memory sims are fine, the HDDs are fine, the SCSI cables are fine, the
DVD drives are fine.

These are clean installations that I know have worked in the past.

I am left thinking that it must be one or both of the CPU's or the
motherboard? These are Slot-1, 850Mhz, 100Mhz FSB, 256K L2, CPU's. They
both have perfectly working and clean genuine Intel heat sinks and fans.

Strangley, I'm not able to get any sort of consistancy with errors. I've
tried every combination of CPU and slots without any consistency of errors.
I have, however, found one consistency: if I disable (or set as WriteThru)
the cache in BIOS then the system boots up fine with perfect consistancy.
However, it is dog slow. Having the cache set as WriteBack is the only
practical way to run the system.

I tested the (ECC) memory overnight with a memory tester and they were fine.
I even tried some other sticks laying around, but got the same errors.

Does this sound like both of the CPU's L2 have gone bad somehow? Or maybe
it is the motherboard? How can I rule out the CPU's from the motherboard
now? Unfortunetly, I don't have any other old boards to test them on.

Thanks!

 

[Kent_Diego]

I'm throwing this out in hopes that someone may suggest what I could try

next. One of my old Dual-P3's that I wish to keep in service has been
giving me problems lately. I know it has to be hardware because it gives
problems in Windows and Linux. Under Windows, during bootup it often
gives blue screens such as IRQL_DRIVER_NOT_LESS_OR_EQUAL,
PAGING_IN_NONPAGED_AREA, etc. These aren't verbatim, but as close as I
remember. Under Linux, it gets kernel panics during bootups.

Anyway, I tested just about everything. All the hardware cards are fine,
the memory sims are fine, the HDDs are fine, the SCSI cables are fine, the
DVD drives are fine.

These are clean installations that I know have worked in the past.

I am left thinking that it must be one or both of the CPU's or the
motherboard? These are Slot-1, 850Mhz, 100Mhz FSB, 256K L2, CPU's. They
both have perfectly working and clean genuine Intel heat sinks and fans.

Strangley, I'm not able to get any sort of consistancy with errors. I've
tried every combination of CPU and slots without any consistency of
errors. I have, however, found one consistency: if I disable (or set as
WriteThru) the cache in BIOS then the system boots up fine with perfect
consistancy. However, it is dog slow. Having the cache set as WriteBack
is the only practical way to run the system.

I tested the (ECC) memory overnight with a memory tester and they were
fine. I even tried some other sticks laying around, but got the same
errors.

Does this sound like both of the CPU's L2 have gone bad somehow? Or maybe
it is the motherboard? How can I rule out the CPU's from the motherboard
now? Unfortunetly, I don't have any other old boards to test them on.

There is an off chance you power supply has problems but most likely it is
the motherboard. Boards manufactured in that era useally suffeer from teh
great Chinese capacitor scandal. Look for bulging or leaking electrolytic
capacitors.
http://en.wikipedia.org/wiki/Capacitor_plague

 

[Paul]

Hi,

I'm throwing this out in hopes that someone may suggest what I could try
next. One of my old Dual-P3's that I wish to keep in service has been
giving me problems lately. I know it has to be hardware because it gives
problems in Windows and Linux. Under Windows, during bootup it often gives
blue screens such as IRQL_DRIVER_NOT_LESS_OR_EQUAL, PAGING_IN_NONPAGED_AREA,
etc. These aren't verbatim, but as close as I remember. Under Linux, it
gets kernel panics during bootups.

Anyway, I tested just about everything. All the hardware cards are fine,
the memory sims are fine, the HDDs are fine, the SCSI cables are fine, the
DVD drives are fine.

These are clean installations that I know have worked in the past.

I am left thinking that it must be one or both of the CPU's or the
motherboard? These are Slot-1, 850Mhz, 100Mhz FSB, 256K L2, CPU's. They
both have perfectly working and clean genuine Intel heat sinks and fans.

Strangley, I'm not able to get any sort of consistancy with errors. I've
tried every combination of CPU and slots without any consistency of errors.
I have, however, found one consistency: if I disable (or set as WriteThru)
the cache in BIOS then the system boots up fine with perfect consistancy.
However, it is dog slow. Having the cache set as WriteBack is the only
practical way to run the system.

I tested the (ECC) memory overnight with a memory tester and they were fine.
I even tried some other sticks laying around, but got the same errors.

Does this sound like both of the CPU's L2 have gone bad somehow? Or maybe
it is the motherboard? How can I rule out the CPU's from the motherboard
now? Unfortunetly, I don't have any other old boards to test them on.

Thanks!

To help separate hardware from software issues, I like to boot
a suspect system, with a second OS. For example, I had a P3 based
system, with 440BX chipset, and thought I had a Win98 problem.
About a year later, I happened to have a Knoppix Linux LiveCD (knopper.net),
and tried booting the machine with that. The symptoms were reproduced in
the second OS, which means Win98 was off the hook, and the hardware
was at fault. (I had AGP video related freezing, whenever more than
2x256MB of memory was installed. 3x256MB or 4x256MB would fail.)

While booted in either Windows or Linux, you can get versions of
Prime95 from mersenne.org for testing. That is one way to stress
the CPU and memory, and find problems faster. The "Torture Test"
should run for hours, without reporting an error. With just 2x256MB
of memory present in the computer, I could run Prime95 for 16 hours
without detecting an error. In Linux, you can even run multiple
copies, assigning a smaller than normal chunk of memory to each.

Paul