CPU maxed out in BIOS setup, fire and damnation awaits ye

I received a PC i5-2400 on Asus P8P67 mainboard. In the UEFI gooey BIOS,
I noticed the CPU fan hits top speed and temperature keeps rising, rising
to 80 Celcius if you sit and watch long enough, whereupon I chicken out
and switched it off.
Now the CPU is going at maximum frequency, and also the voltage is 30 mV
more than needed to operate. I presume all cores are being thrashed.
Now a single-core 500 MHz CPU would be plenty to run the BIOS menus.
I am curious why it is going pedal-to-the-metal.
The case has 3 fans, and the graphics card is only slot powered, so
there should not be an overheating problem.

BTW I have also seen this with MSI mainboards.

wrote:
I received a PC i5-2400 on Asus P8P67 mainboard. In the UEFI gooey BIOS,
I noticed the CPU fan hits top speed and temperature keeps rising, rising
to 80 Celcius if you sit and watch long enough, whereupon I chicken out
and switched it off.
Now the CPU is going at maximum frequency, and also the voltage is 30 mV
more than needed to operate. I presume all cores are being thrashed.
Now a single-core 500 MHz CPU would be plenty to run the BIOS menus.
I am curious why it is going pedal-to-the-metal.
The case has 3 fans, and the graphics card is only slot powered, so
there should not be an overheating problem.

BTW I have also seen this with MSI mainboards.

All it takes is software written with a polling
loop in the code (i.e. not event driven code),
to rail one core of a CPU. That should not draw
max power (TDP) from the CPU, but somewhat less
power on a multicore processor. As only 1 of N cores
is running at max.

I would be checking that the CPU cooler is sitting flat
on the CPU, and the user has not removed all the
thermal paste for fun. Some people take the heatsink
off the CPU, clean off the paste, then realize...

"Hey, I don't got no paste. I'm just going to slap
this cooler back on without any, and nobody is
going to notice."

So disassemble and re-paste the thing. Or, check
at least, that the clips haven't been rotated so
they're released or whatever. Maybe the heatsink
is not pressing down on the CPU.

There has to be an explanation for why the cooler
is not removing the heat, such as a lack of a working
thermal path.

On heatsinks that use heatpipes, if the heatpipes
lose the working fluid in each pipe, that can cause
cooling failure too. Back when heatpipe coolers first
came out, some coolers shipped without the two drops of
fluid installed in each pipe. Apparently the seals on
the pipes, didn't work worth a damn. On modern coolers,
this no longer seems to be a problem. They've figured it
out. But one of the reasons for a cooler to have four
pipes on it, is if one pipe leaks and the working
fluid escapes, the other three continue to work.
Only the most grotesque manufacturing error, results
in "all pipes empty" and then the heatsink no longer
conducts heat properly.

A working heatpipe is more conductive than a solid
copper tube of equal diameter. The phase change
fluid inside the pipe, is much more effective
at thermal transport, than a piece of solid metal
of equal size. The thing works like a distillation
column.

Paul

On Tue, 15 Mar 2016 01:32:25 -0700 (PDT), wrote:

BTW I have also seen this with MSI mainboards.

A few years ago, especially now with AMDs and die/micron technology,
there's consequence to CPU coolers for struggling with "the wall of
limits." Even with a proper CPU cooler install (and I'm a generation
prior for my CPU), I've watched temperatures rise from updating the
CPU from single, dual, and then a quad. Updating to the latest in
some highend decoders the other day, I'm hitting limits. Aside from
stuttering during the decoding, a temperature to core efficiency
coefficient has the MB struggling to keep up with processes at 15deg.
F higher than anything I'd experienced prior. I just have to trust
the MB manufacture quality for preventing a meltdown at its upper
limits. An aspect I'd suspect exacerbated by less leeway not to
ignore an overall tighter precision to MB and CPU environments at
their latest limits.

On Tuesday, March 15, 2016 at 9:05:49 PM UTC+8, Paul wrote:

I would be checking that the CPU cooler is sitting flat
on the CPU, and the user has not removed all the
thermal paste for fun. Some people take the heatsink
off the CPU, clean off the paste, then realize...


I had checked that already. There was a little dust on the heatsink,
I blew that out. Then I noticed the (Intel) cooler only had 0.17 A fan.
Seems too low for a TDP of 95 W. The only spare cooler I have to
swap is 0.2 A fan.
Back in the socket 775 days, I used to bung Pentium D coolers on
everything - they had 0.6 A fans.
Also checked all the BIOS settings. Everything was normal, except the
memory timings had been cut to 6-6-6-20. I doubt that would affect
CPU.

wrote:
On Tuesday, March 15, 2016 at 9:05:49 PM UTC+8, Paul wrote:
I would be checking that the CPU cooler is sitting flat
on the CPU, and the user has not removed all the
thermal paste for fun. Some people take the heatsink
off the CPU, clean off the paste, then realize...


I had checked that already. There was a little dust on the heatsink,
I blew that out. Then I noticed the (Intel) cooler only had 0.17 A fan.
Seems too low for a TDP of 95 W. The only spare cooler I have to
swap is 0.2 A fan.
Back in the socket 775 days, I used to bung Pentium D coolers on
everything - they had 0.6 A fans.
Also checked all the BIOS settings. Everything was normal, except the
memory timings had been cut to 6-6-6-20. I doubt that would affect
CPU.

How about CPU voltage setting.

Is it boosted ?

Is the voltage wrong for that processor ? Occasionally,
a BIOS will use the wrong value (in cases where a boost
option is available in hardware). That accounted for excessive
thermal in at least one case.

Paul

On Thursday, March 17, 2016 at 4:26:48 AM UTC+8, Paul wrote:


How about CPU voltage setting.

Is it boosted ?

Is the voltage wrong for that processor ? Occasionally,
a BIOS will use the wrong value (in cases where a boost
option is available in hardware). That accounted for excessive
thermal in at least one case.

Paul

Yes it is slightly high. They have probably added a margin to keep it
stable.
Anyway, I decided that either the CPU or mainboard is suspect, so did
a few more tests before scrapping it. If I reduced the number of cores
in BIOS settings, no effect on temperature. Only if I reduce the frequency
to less than 2 GHz would the temperature stop rising. So it seems that just
one thread is in a polling loop, as you suggested.