Unresponsive system caused by bad CPU?

[Jef Driesen]

I have an older PC with a 550MHz Pentium III processor. During the last
days, the system became very unresponsive. Booting is very slow and when
the system is finally up, the mouse keeps hanging very often. And
applications load very slow. There are no crashes or errors.

I don't think this is an operating system issue, because it affects both
Windows (2000) and Linux (Ubuntu 2.10). To exclude memory errors, I used
memtest86+ and errors were reported during test 5 (Block move, 64
moves). By removing dimms one by one, I managed to identify the faulty
dimm. But even with the bad dimm removed, the problem remains.

Yesterday I had the opportunity to test with a different processor
(450MHz) and the system is much more responsive with it. The system
seems to run much faster, even if with the 100MHz speed difference.

But before I start looking for a new CPU, I want to be sure there is no
other problem. Because I also tested my CPU in another PC and there
seems to be no problem. Is there a way to verify that my CPU is the real
problem?

 

[GT]

I have an older PC with a 550MHz Pentium III processor. During the last
days, the system became very unresponsive. Booting is very slow and when
the system is finally up, the mouse keeps hanging very often. And
applications load very slow. There are no crashes or errors.

I don't think this is an operating system issue, because it affects both
Windows (2000) and Linux (Ubuntu 2.10). To exclude memory errors, I used
memtest86+ and errors were reported during test 5 (Block move, 64 moves).
By removing dimms one by one, I managed to identify the faulty dimm. But
even with the bad dimm removed, the problem remains.

Yesterday I had the opportunity to test with a different processor
(450MHz) and the system is much more responsive with it. The system seems
to run much faster, even if with the 100MHz speed difference.

But before I start looking for a new CPU, I want to be sure there is no
other problem. Because I also tested my CPU in another PC and there seems
to be no problem. Is there a way to verify that my CPU is the real
problem?

Can you test the processor in a different machine?

How much memory did you remove? If you reduced the RAM significantly, then
windows will struggle by comparisson.

 

[kony]

I have an older PC with a 550MHz Pentium III processor. During the last
days, the system became very unresponsive. Booting is very slow and when
the system is finally up, the mouse keeps hanging very often. And
applications load very slow. There are no crashes or errors.

I don't think this is an operating system issue, because it affects both
Windows (2000) and Linux (Ubuntu 2.10). To exclude memory errors, I used
memtest86+ and errors were reported during test 5 (Block move, 64
moves). By removing dimms one by one, I managed to identify the faulty
dimm. But even with the bad dimm removed, the problem remains.

Yesterday I had the opportunity to test with a different processor
(450MHz) and the system is much more responsive with it. The system
seems to run much faster, even if with the 100MHz speed difference.

But before I start looking for a new CPU, I want to be sure there is no
other problem. Because I also tested my CPU in another PC and there
seems to be no problem. Is there a way to verify that my CPU is the real
problem?

I believe you just did when you swapped it into the other
system.

Put the CPU back in and benchmark it, something like Sisoft
Sandra will allow crude (but good enough for this puprose)
benches of subsystems so if the CPU is acting up it'd be
evident. Also try running Prime 95's torture test, since it
checks it's work you would see errors.

If you had not swapped the old CPU back in yet i'd try that.

 

[philo]

I have an older PC with a 550MHz Pentium III processor. During the last
days, the system became very unresponsive. Booting is very slow and when
the system is finally up, the mouse keeps hanging very often. And
applications load very slow. There are no crashes or errors.

I don't think this is an operating system issue, because it affects both
Windows (2000) and Linux (Ubuntu 2.10). To exclude memory errors, I used
memtest86+ and errors were reported during test 5 (Block move, 64 moves).
By removing dimms one by one, I managed to identify the faulty dimm. But
even with the bad dimm removed, the problem remains.

Yesterday I had the opportunity to test with a different processor
(450MHz) and the system is much more responsive with it. The system seems
to run much faster, even if with the 100MHz speed difference.

not likely to be a bad cpu (but anything is possible)
try the 550mhz cpu again...but clock the system down to 450mhz...
and see what happens...
if it works ok...try 500mhz

 

[Jef Driesen]

Can you test the processor in a different machine?

I tried that and there seems to be no problem. I'll do some more testing
today.

How much memory did you remove? If you reduced the RAM significantly, then
windows will struggle by comparisson.

Originally there were 3 dimms (2x128MB + 64MB). One of the 128MB dimms
showed errors and was removed. After removing the memory module, there
was no change in the symptoms (maybe a little performance is lost but is
certainly negligible compared to the other problem).

I have also tried the dimms from the other PC (also 128MB + 64MB), but
no difference.

 

[Jef Driesen]

I believe you just did when you swapped it into the other
system.

Put the CPU back in and benchmark it, something like Sisoft
Sandra will allow crude (but good enough for this puprose)
benches of subsystems so if the CPU is acting up it'd be
evident. Also try running Prime 95's torture test, since it
checks it's work you would see errors.

If you had not swapped the old CPU back in yet i'd try that.

I'll try benchmarking today. (If that's even possible, because the
system is so slow I don't know it is possible to install something. But
I'll try.)

 

[Jef Driesen]

not likely to be a bad cpu (but anything is possible)
try the 550mhz cpu again...but clock the system down to 450mhz...
and see what happens...
if it works ok...try 500mhz

Never changed the clockspeed of the cpu before, but I'll try. I suppose
this is changed somewhere in the bios?

The motherboard is an ASKA SST-5887 with VIA Apollo Pro133 chipset. More
details at http://www.silverstar.com.hk/product/Specif/spec5887.htm.

 

[philo]

Never changed the clockspeed of the cpu before, but I'll try. I suppose
this is changed somewhere in the bios?

The motherboard is an ASKA SST-5887 with VIA Apollo Pro133 chipset. More
details at http://www.silverstar.com.hk/product/Specif/spec5887.htm.

yes you can change it in the bios...
on a few rare occasions i've been able to get troublesome machines
to work fine with a slight amount of underclocking...

who knows it may work for you

 

[kony]

Never changed the clockspeed of the cpu before, but I'll try. I suppose
this is changed somewhere in the bios?

The motherboard is an ASKA SST-5887 with VIA Apollo Pro133 chipset. More
details at http://www.silverstar.com.hk/product/Specif/spec5887.htm.

Be very careful about changing clock speeds. Those CPUs are
multiplier locked so the option is of changing the FSB
speed, which changes the PCI speed- and that can cause
random glitches with devices like USB, network adapters, and
more importantly, IDE controller errors that can corrupt
data.

I'd wonder if you had a failing motherboard, maybe the caps
are popped and every little bit of extra power puts it
further into instable operation.

 

[Jef Driesen]

I have an older PC with a 550MHz Pentium III processor. During the last
days, the system became very unresponsive. Booting is very slow and when
the system is finally up, the mouse keeps hanging very often. And
applications load very slow. There are no crashes or errors.

I don't think this is an operating system issue, because it affects both
Windows (2000) and Linux (Ubuntu 2.10). To exclude memory errors, I used
memtest86+ and errors were reported during test 5 (Block move, 64
moves). By removing dimms one by one, I managed to identify the faulty
dimm. But even with the bad dimm removed, the problem remains.

Yesterday I had the opportunity to test with a different processor
(450MHz) and the system is much more responsive with it. The system
seems to run much faster, even if with the 100MHz speed difference.

But before I start looking for a new CPU, I want to be sure there is no
other problem. Because I also tested my CPU in another PC and there
seems to be no problem. Is there a way to verify that my CPU is the real
problem?

I did some more testing yesterday with different combinations of
components from both PCs. And suddenly the problem disappeared again
(with the original configuration). And I have absolutely no idea what
happened. The only thing I can think of is some bad contact somewhere
(caused by dust?) that was fixed moving around the components.

I did the Prime 95 torture test and no errors were reported after a few
hours of operation. I didn't tried underclocking my system, because i
couldn't find out how to do that. (There were only 3 settings i could
change, but i don't know their meaning. I suppose one of them was this
multiplier lock thing but I'm no sure.)

 

[GT]

I did some more testing yesterday with different combinations of
components from both PCs. And suddenly the problem disappeared again (with
the original configuration). And I have absolutely no idea what happened.
The only thing I can think of is some bad contact somewhere (caused by
dust?) that was fixed moving around the components.

I did the Prime 95 torture test and no errors were reported after a few
hours of operation. I didn't tried underclocking my system, because i
couldn't find out how to do that. (There were only 3 settings i could
change, but i don't know their meaning. I suppose one of them was this
multiplier lock thing but I'm no sure.)

Intel CPUs throttle their own speed when they get too hot -
self-preservation. Try cleaning out the fan that cools the CPU and air will
flow more easily and cool more efficiently. By taking the CPU out and moving
it around, you have probably inadvertantly cleaned it a little.

 

[Jef Driesen]

Intel CPUs throttle their own speed when they get too hot -
self-preservation. Try cleaning out the fan that cools the CPU and air will
flow more easily and cool more efficiently. By taking the CPU out and moving
it around, you have probably inadvertantly cleaned it a little.

I don't think it was caused by heat, because it happened immediately
after booting. And as far as i know, those old CPUs are not capable of
throttling their speed, but i could be wrong about that.

 

[GT]

I don't think it was caused by heat, because it happened immediately after
booting. And as far as i know, those old CPUs are not capable of
throttling their speed, but i could be wrong about that.

Fair point - you might be right. I dont know when Intel introduced the
frequency throttle. Still a clean can't do any harm! At least your problem
has gone away as quickly and misteriously as it arrived!

 

[philo]

I don't think it was caused by heat, because it happened immediately after
booting. And as far as i know, those old CPUs are not capable of
throttling their speed, but i could be wrong about that.

that's a good presumption.
I once had a cpu fan quit on a similar machine...
and it was hours before I even noticed a bit of sluggishness.

Maybe just one pin was not making good contact...who knows?

 

[kony]

Intel CPUs throttle their own speed when they get too hot -
self-preservation. Try cleaning out the fan that cools the CPU and air will
flow more easily and cool more efficiently. By taking the CPU out and moving
it around, you have probably inadvertantly cleaned it a little.

His does not throttle, that is only a feature of Pentium 4
or later generation CPU. His might have overheat shutdown
mechanism, but that would literally shut it down, not run
slowly. It does not rule out the possibility that it was
overheating though, at a certain point overheating induced
instability could cause a type of sluggish failure, more
easily masked on 2K/XP which would merely bluescreen
win9x... at least it seemed more likely with one vs other
OS.

There are a lot of potential things that can cause
sluggishness, like bad IDE cables, even intermittent or
failed network connection in some cases will effect
non-networked activities.

 

[kony]

that's a good presumption.
I once had a cpu fan quit on a similar machine...
and it was hours before I even noticed a bit of sluggishness.

Maybe just one pin was not making good contact...who knows?

With a system that age it is fairly easy for contacts to be
dirty, maybe even corroded in severe cases... a clue is
often provided by the overall condition of the system, if
the inside looks dirty or a film of residue has accumulated
(like from smokers or workplace/industrial dirt), or if th
edges of the metal chassis have rusted unduely as an
indication of higher humidity levels (varies per quality of
case finishing though).

If it's the CPU socket or slot sometimes merely inserting
and removing CPU a few times can help... but whether it's
short or long-term fix can depend on the extent of the
problem as well as whether system is still operating in the
problem environment.

 

[Jef Driesen]

I did some more testing yesterday with different combinations of
components from both PCs. And suddenly the problem disappeared again
(with the original configuration). And I have absolutely no idea what
happened. The only thing I can think of is some bad contact somewhere
(caused by dust?) that was fixed moving around the components.

I did the Prime 95 torture test and no errors were reported after a few
hours of operation. I didn't tried underclocking my system, because i
couldn't find out how to do that. (There were only 3 settings i could
change, but i don't know their meaning. I suppose one of them was this
multiplier lock thing but I'm no sure.)

I cried victory too early. The problem appeared again an hour ago during
booting and when the system is up and running (which takes at least 5
minutes on a clean Windows 2000 install), it becomes a little better
after some time. Very strange.

 

[Jef Driesen]

<snip>

There are a lot of potential things that can cause
sluggishness, like bad IDE cables, even intermittent or
failed network connection in some cases will effect
non-networked activities.

I replaced the IDE cable of the harddrive and the system is running fine
now for several days. And I hope it stays like that!