***** charles wrote:
> Hi,
>
> I have an ECS L7VMM3 v1.0c with an XP2000+ running
> at 1.667GHz and 256M of ram in one stick. I have run
> memtest for a day and a half with no lockups. Sometimes
> the cpu runs at around 40C degrees and other times I have
> seen as high as 61C degrees. Is this higher number still
> within spec? I have tried different new power supplies and
> different ram but it still locks up once in a while. Any
> other ideas on whats wrong? I run Kubuntu 7.04 with all
> the updates. Sometimes the cursor just freezes and I have
> to pull the plug to get it restarted.
>
> later,
> charles.......
>
If you have a second computer, and both computers are networked,
the next time it freezes, try the "ping" command. I've had a
Windows machine "freeze" and yet it still returned a "ping"
without a problem. Which means that, perhaps a user interface
element was frozen, but not the whole machine.
If you have set up the ability to log into your L7VMMS
remotely, you could also try connecting to it from a
second computer.
The processor gets hotter at full (100%) CPU load. 61C is not
impossible, especially if the heatsink is on the small side.
Regular heatsinks are about the size of the socket area. Some
aftermarket heatsinks, are much larger, and may use a larger
fan. That can help a bit with heat.
Note that the heatsink cannot do its job, if all the hot air
sits around the heatsink. There has to be good case circulation,
so the air the CPU fan blows onto the heatsink, is cool.
If the CPU has a particularly small die, another problem is
power density. The small die, combined with a 60W heat output,
means a lot of heat has to flow through a small area. Perhaps
the best way to solve this problem, is with multiple heatpipes
in close proximity to the silicon die. Now, recently, a company
made a heatsink, with heatpipes but no copper heat spreader
plate. But it still didn't seem to function quite as well as
I expected. Heatpipes are still a good solution, for moving
heat from a small area, into a larger collection of fins. But
at least for now, the heatpipes should contine to use a supporting
piece of copper, to help spread the heat from the silicon die more
evenly (without hot spots on the die).
Another test you can use, is Prime95. It is possible to
run a copy of Prime95, even under Kubuntu. Mersenne.org
has versions of Prime95 for both Windows and for Linux.
I've even run multiple copies of Prime95 under
Knoppix. (Download both Linux versions, because I can
never remember whether the statically compiled or the
dynamic one, worked for me.)
Prime95 and the "torture test" option, runs the CPU at 100%
and gives a good heat load. If memtest isn't turning up a problem,
see what the symptoms are like with the Torture Test. When
overclocking, I've had Prime95 error out in about 10 seconds
or less. Once everything is tweaked properly, I can run for
4 hours without seeing an error. The longest I've waited for
Prime95, is about 16 hours on an ancient 440BX motherboard.
For an example of a heatsink with some good properties,
have a look at the design of the Big Typhoon. It is
rather expensive, at $45, but it is a good size. Nice things
about it:
http://59.124.46.91/manual/SoftManual/CL-P0114-01_en_060830.zip
1) Heatpipes, to move heat to the fins.
2) Aluminum fins to reduce overall weight.
3) Threaded fasteners, so you decide how tight to make it.
4) Open construction, with room for your fingers.
It does require removing the motherboard, but if you've worked
on a home built system for very long, you've already had the
motherboard out of the case a half dozen times anyway
HTH,
Paul
FAQ
Profile
Private Messages
Log in
