> I am unsure. One would hope that powering down the system if a hard-limit 
> temperature is reached *is* in hardware rather than software. OTOH, here 
> we have a laptop which got sufficiently hot for the CPU to malfunction 
> fairly seriously, yet still the power remains on.

http://www.tomshardware.com/reviews/hot-spot,365.html

It's quite old, but the basic outcome seems to be that the Intel CPUs are 
impossible to damage by overheating.  With the AMD chip it relies on the 
motherboard, but their demo shows that a motherboard temperature sensor 
cannot react quickly enough to avoid frying the CPU when the HS comes off.

Post a reply to this message

scott wrote:

> http://www.tomshardware.com/reviews/hot-spot,365.html
> 
> It's quite old, but the basic outcome seems to be that the Intel CPUs 
> are impossible to damage by overheating.  With the AMD chip it relies on 
> the motherboard, but their demo shows that a motherboard temperature 
> sensor cannot react quickly enough to avoid frying the CPU when the HS 
> comes off.

Oh wow... 370°C after 1 second? That's pretty special!

Looking at this, it seems that in "the old days" some CPUs didn't have 
any temperature sensors at all. I would imagine given the *huge* amounts 
of heat that newer CPUs generate, this must have changed by now.

I still don't know whether the fan speed is hardware or software 
controlled, but I would think by now the system will at least turn 
itself off in a thermal emergency without software intervention.

As an aside... How much heat does the human brain generate? (By every 
estimate I've seen, it has vastly superior theoretical computational 
power compared to any supercomputer yet built by man.) How come human 
brains don't ignite and burn during normal operation? I don't see any 
really large heat sinks on a human...

Post a reply to this message

On Fri, 09 Jan 2009 10:04:54 +0000, Invisible <voi### [at] devnull> wrote:

> How come human 
>brains don't ignite and burn during normal operation? I don't see any 
>really large heat sinks on a human...

It is fluid cooled and what about those heatsinks on the side of your head? 
-- 

Regards
     Stephen

Post a reply to this message

>> How come human 
>> brains don't ignite and burn during normal operation? I don't see any 
>> really large heat sinks on a human...
> 
> It is fluid cooled and what about those heatsinks on the side of your head? 

Water cooled? Elite!

But I mean, *damn*, you'd have to have a high-volume fluid pumping 
system, and a very large surface area to dissapate the heat over.



...oh, wait...

Post a reply to this message

> As an aside... How much heat does the human brain generate?

You can do a quick estimation by seeing how much heat is radiated, the skin 
temperature is approximately 32 degrees, room temperature is 20 degrees, and 
the surface area of your head is, umm about 0.08m^2.  Using the 
Stefan-Boltzmann law I make that about 6W of radiated heat.  Given that the 
human body generates about 100W, I think the 6W figure is too low, as some 
of the heat from the brain will be taken into the body by the blood stream 
(liquid cooling!).  I guess around 20W would be a more sensible figure.


> (By every estimate I've seen, it has vastly superior theoretical 
> computational power compared to any supercomputer yet built by man.)

If my computer were as slow as me doing sums it would probably run on solar 
power in the dark!  And if I tried to do 3 billion sums in one second my 
head would surely explode.

Post a reply to this message

scott wrote:
>> As an aside... How much heat does the human brain generate?
> 
> Given 
> that the human body generates about 100W, I think the 6W figure is too 
> low, as some of the heat from the brain will be taken into the body by 
> the blood stream (liquid cooling!).  I guess around 20W would be a more 
> sensible figure.

http://vadim.oversigma.com/MAS862/Project.html

Suggests that the entire body consumes about 100W, and the brain itself 
uses 20 - 40 W. (So... a pretty significant fraction, considering that 
the human body is *not* a data processing device, primarily.) That's how 
much energy it *uses*, I wonder how much of that ends up as heat?

>> (By every estimate I've seen, it has vastly superior theoretical 
>> computational power compared to any supercomputer yet built by man.)
> 
> If my computer were as slow as me doing sums it would probably run on 
> solar power in the dark!  And if I tried to do 3 billion sums in one 
> second my head would surely explode.

Well, you say that, but how many "sums" does it take to compare a 
snippet of sound to many millions of other recorded examples like it and 
determine that, yes, this is The Beetles singing "All you need is love"?

The human brain doesn't work the same way a computer does. This isn't 
exactly news. Indeed, this is why computers are useful in the first 
place! But based on the number of switching elements in a computer vs 
those in the brain, AFAIK the brain comes out rather favourably.

(The same website above gives the number of transisters in a Pentium IV 
as K * 10^7, while the number of neurons in a brain is K * 10^11. Kind 
of a big difference...)

Post a reply to this message

Invisible wrote:

> http://vadim.oversigma.com/MAS862/Project.html

"The power consumption of the Pentium 4 may exceed that of the human 
brain. That explains why Pentium cooling systems are getting close in 
size to the human head."

I LOLd! :-D

Post a reply to this message

> Suggests that the entire body consumes about 100W, and the brain itself 
> uses 20 - 40 W. (So... a pretty significant fraction, considering that the 
> human body is *not* a data processing device, primarily.) That's how much 
> energy it *uses*, I wonder how much of that ends up as heat?

Well there aren't too many options for "used energy" to be converted to. 
Maybe some is used to increase the kinetic energy of some blood, but I would 
say that almost all is converted to heat.

> The human brain doesn't work the same way a computer does. This isn't 
> exactly news. Indeed, this is why computers are useful in the first place! 
> But based on the number of switching elements in a computer vs those in 
> the brain, AFAIK the brain comes out rather favourably.

I guess the brain also makes even Windows ME look good for uptime records 
:-)

Post a reply to this message

scott wrote:
>> Suggests that the entire body consumes about 100W, and the brain 
>> itself uses 20 - 40 W. (So... a pretty significant fraction, 
>> considering that the human body is *not* a data processing device, 
>> primarily.) That's how much energy it *uses*, I wonder how much of 
>> that ends up as heat?
> 
> Well there aren't too many options for "used energy" to be converted to. 
> Maybe some is used to increase the kinetic energy of some blood, but I 
> would say that almost all is converted to heat.

Well, I guess the brain does use *some* of that energy for chemical 
transformations. (It turns raw ingredients into hormones and releases 
them, for example.) But yeah, *most* of it is porbably heat...

FWIW, how much heat does an Intel Core 2 Quad generate?

> I guess the brain also makes even Windows ME look good for uptime 
> records :-)

HAHAHAH! PWN3D! :-D

Post a reply to this message

> FWIW, how much heat does an Intel Core 2 Quad generate?

About 120W under full load.

Post a reply to this message