P4 instablility/heat problems

K

kd

I built a computer for my brother a few months back, and ever since I
put it together, it's been having problems with either heat or
instability.

Whenever we did anything mildly cpu intensive, from encoding MP3s to
playing games, the system would run fine for a few minutes and then
shut off. This sounded like a typical overheating problem, so I tried
buying a better heatsink/fan combo, cleaning and reapplying liquid
silver, etc, and nothing worked. It still got very hot very quickly.

One day I got the bright idea of underclocking it by lowering the
voltage to the CPU slightly, and for some reason, the instablility
immediately ended. The computer never actually halts, but it does go
very slowly, slower than I would think with a tiny underclocking. All
games are basically unplayable, and even desktop usage goes slowly at
times. This is obviously ridiculous for a 3.2ghz P4 with a gig of ram
and a modern video card running Win2k.

I've done some heavy parts swapping, and am very certain that the CPU
is the problem, as it occurs no matter what motherboard, video card, or
ram is attached.

I'm assuming that the CPU is still overheating and that the P4 is
throttling the speed down and down as the temperature rises. I don't
have any idea what other options remain, short of buying a new CPU.

Any ideas? I'm so tired of kicking this thing around I'm ready to buy
prebuilt pc's from now on (heresy!).

Thanks a bunch,
kd
 
B

badgolferman

kd, 11/8/2005, 11:57:22 AM,
Eh...

What do you recommend, then?

Is liquid silver the same thing as heat sink compound? That's all I've
ever used. Some heat sinks have that brown stiff on it that I usually
scrape off.
 
J

John McGaw

kd said:
I built a computer for my brother a few months back, and ever since I
put it together, it's been having problems with either heat or
instability.

Whenever we did anything mildly cpu intensive, from encoding MP3s to
playing games, the system would run fine for a few minutes and then
shut off. This sounded like a typical overheating problem, so I tried
buying a better heatsink/fan combo, cleaning and reapplying liquid
silver, etc, and nothing worked. It still got very hot very quickly.

One day I got the bright idea of underclocking it by lowering the
voltage to the CPU slightly, and for some reason, the instablility
immediately ended. The computer never actually halts, but it does go
very slowly, slower than I would think with a tiny underclocking. All
games are basically unplayable, and even desktop usage goes slowly at
times. This is obviously ridiculous for a 3.2ghz P4 with a gig of ram
and a modern video card running Win2k.

I've done some heavy parts swapping, and am very certain that the CPU
is the problem, as it occurs no matter what motherboard, video card, or
ram is attached.

I'm assuming that the CPU is still overheating and that the P4 is
throttling the speed down and down as the temperature rises. I don't
have any idea what other options remain, short of buying a new CPU.

Any ideas? I'm so tired of kicking this thing around I'm ready to buy
prebuilt pc's from now on (heresy!).

Thanks a bunch,
kd

Question: why do you keep guessing about the CPU temperature? It is easy
to measure it and know for certain. You can download either Everest Home
Edition or MBM are free and they will give you real-time CPU core
temperature readings while the computer is doing CPU intensive tasks.
There are probably other programs to do the same thing but these are
ones I use all the time and have some trust in.
 
K

kony

I built a computer for my brother a few months back, and ever since I
put it together, it's been having problems with either heat or
instability.

Whenever we did anything mildly cpu intensive, from encoding MP3s to
playing games, the system would run fine for a few minutes and then
shut off. This sounded like a typical overheating problem, so I tried
buying a better heatsink/fan combo, cleaning and reapplying liquid
silver, etc, and nothing worked. It still got very hot very quickly.

One day I got the bright idea of underclocking it by lowering the
voltage to the CPU slightly, and for some reason, the instablility
immediately ended. The computer never actually halts, but it does go
very slowly, slower than I would think with a tiny underclocking. All
games are basically unplayable, and even desktop usage goes slowly at
times. This is obviously ridiculous for a 3.2ghz P4 with a gig of ram
and a modern video card running Win2k.

I've done some heavy parts swapping, and am very certain that the CPU
is the problem, as it occurs no matter what motherboard, video card, or
ram is attached.

I'm assuming that the CPU is still overheating and that the P4 is
throttling the speed down and down as the temperature rises. I don't
have any idea what other options remain, short of buying a new CPU.

Any ideas? I'm so tired of kicking this thing around I'm ready to buy
prebuilt pc's from now on (heresy!).

Thanks a bunch,
kd

You didn't even tell us what temps it was hitting. Kind of
hard to know how far off you are from a reasonably
attainable temp.

Are you sure the heatsink mounting bracket is correct and
installed properly, as is the heatsink?

Does the heatsink have a poorly finished or otherwise
irregular bottom surface?

Arctic Silver should do fine, but so should just about any
thermal compound (at least in the short-term, for a dozen
months or more I recommend only synthetic based compounds
for modern CPUs, but it need not be Arctic Silver, Arctic
Alumina or other synthetic based compound should suffice).

Is the fan running at full RPM?
Are you sure the heatsink is adequate? Some heatsink
manufacurers have rather lofty target applications,
particularly the mid-low end 'sinks often claim to support
CPUs that would run fairly hot using some 'sinks.

Does your motherboard bios have a thermal throttling
setting? Is it a decent board or a cheapie? Some cheap
boards actually overvolt the CPU a little to make up for a
poor (cheap) power regulation design. In the short term it
seems to help stability except for it being hotter running
but in the longer term it wears out the board too.

Does the case have adequate intake and exhaust?
If you leave the side panel off and point a desk fan at it,
does it (PCU) run more than 510C cooler? That would tend to
suggest you need change the case airflow or use another
case.

What's the ambient temp? You haven't mentioned a LOT of the
important details about this system's cooling. What case (a
linked picture of the front, rear, inside might help)?

Is it possible the system was in a dusty environment and the
front intake of the case is partially clogged up?
 
D

DaveW

That CPU you installed is a Prescott series unit and is famous in the
industry for running VERY HOT. T o tame the heat you need to install a
heavy duty CPU fan and heat sink from Zalman, etc. on the CPU. Then make
sure that the case you are using has many intake and exhaust fans to get rid
of the heat that the Zalman will be giving off. A typical standard computer
case will not work.
 
M

Michael C

kd said:
I built a computer for my brother a few months back, and ever since I
put it together, it's been having problems with either heat or
instability.

Whenever we did anything mildly cpu intensive, from encoding MP3s to
playing games, the system would run fine for a few minutes and then
shut off. This sounded like a typical overheating problem, so I tried
buying a better heatsink/fan combo, cleaning and reapplying liquid
silver, etc, and nothing worked. It still got very hot very quickly.

I had similar problems with an AMD once and took out the board to return it.
After I took it out I noticed the fan was on an angle and was only touching
the CPU on one edge! I straightened it up, put it back in the customers
computer and never heard from them again.

With your computer there's 2 things I'm interested in knowing. The first,
what temp does the bios say the cpu is getting to and second, does the
problem go away if the sides are off the case?

Michael
 
J

John McGaw

Michael said:
I had similar problems with an AMD once and took out the board to return it.
After I took it out I noticed the fan was on an angle and was only touching
the CPU on one edge! I straightened it up, put it back in the customers
computer and never heard from them again.

With your computer there's 2 things I'm interested in knowing. The first,
what temp does the bios say the cpu is getting to and second, does the
problem go away if the sides are off the case?

Michael

The temperature the BIOS reports isn't a good indicator for failures
under load. By the time you can get out of the testing mode and get back
to the BIOS to check the temps even a truly feeble cooler can bring the
temperature down to within limits. The key to troubleshooting this sort
of problem is to run under load and test at the same time. It is trivial
to do since there are several free utilities to do just that. Everest
Home is probably the easiest to deal with for most people -- the worst
it seems to do on a few MBs is to get confused about which temperature
sensor is on the CPU but that is pretty easy to figure out. MBM is good
but it can be pretty tricky when presented with an odd MB and can even
cause crashes with a few of them.
 
B

Bob

The temperature the BIOS reports isn't a good indicator for failures
under load. By the time you can get out of the testing mode and get back
to the BIOS to check the temps even a truly feeble cooler can bring the
temperature down to within limits.

You could set a low alarm limit in the BIOS and use that to indicate
when the test temp is reached.
 
J

John McGaw

Bob said:
You could set a low alarm limit in the BIOS and use that to indicate
when the test temp is reached.


Yes, I guess one could do that, assuming that the BIOS in question
supports alarm it -- some don't. But it seems to be a long way around if
one is looking to find out the actual temperature when it can be done in
one easy step which costs nothing except a few minutes of download and
install time. De gustibus non est disputandum...
 
K

kd

John said:
The temperature the BIOS reports isn't a good indicator for failures
under load. By the time you can get out of the testing mode and get back
to the BIOS to check the temps even a truly feeble cooler can bring the
temperature down to within limits. The key to troubleshooting this sort
of problem is to run under load and test at the same time. It is trivial
to do since there are several free utilities to do just that. Everest
Home is probably the easiest to deal with for most people -- the worst
it seems to do on a few MBs is to get confused about which temperature
sensor is on the CPU but that is pretty easy to figure out. MBM is good
but it can be pretty tricky when presented with an odd MB and can even
cause crashes with a few of them.

The interesting thing is that before I underclocked it, the bios would
report temps climbing to 100 deg C. To me, that says the heatsink
isn't mounted properly, but I've mounted and remounted it, cleaned off
the thermal paste and reapplied it, and until I underclocked it, it
kept doing that.
 
K

kony

The temperature the BIOS reports isn't a good indicator for failures
under load. By the time you can get out of the testing mode and get back
to the BIOS to check the temps even a truly feeble cooler can bring the
temperature down to within limits.

While that is true and I am a strong advocate of ignoring
idle temps and focusing on full load temps, in this
particular case if the heatsink wasn't even on good the bios
temp report would probably be sufficient for showing a
problem, providing the user is somewhat familiar with the
expected temp of the CPU.
 

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments. After that, you can post your question and our members will help you out.

Ask a Question

Top