Integrating Peltier cooling directly into chips

[Guest]

Archived from groups: alt.comp.hardware.amd.x86-64,comp.arch,comp.sys.ibm.pc.hardware.chips (More info?)

AMD has filed a patent for integrating the Peltier cooling effect directly
into its CPUs. Currently most Peltier coolers work on the heatsink of the
chip, rather than directly in the chip. This might be something they are
getting ready for when they transition to the 65nm process. So far they've
been able to get away without too many problems at the 90nm node, but
judging by what we're seeing from Intel, it's only a matter of time before
AMD also experiences the same thing; and it might experience it by the 65nm
node.

http://www.neowin.net/comments.php?id=25053&category=main

Yousuf Khan

--
Humans: contact me at ykhan at rogers dot com
Spambots: just reply to this email address ;-)

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

I have checked out the external type Peltier coolers. While they might work
really well, they pull about 70 watts ! This has got to produce some heat.

I wonder if they have special high temp fans for use with a Peltier, it seems
to me that the heatsink is going to be quite hot. While that'll increase the
dissipation rate, I have some difficulty accepting more heat as a good thing.

JURB

 

[Guest]

Archived from groups: comp.sys.ibm.pc.hardware.chips (More info?)

Bitstring <20041031131355.17534.00001980@mb-m20.aol.com>, from the
wonderful person JURB6006 <jurb6006@aol.com> said
>I have checked out the external type Peltier coolers. While they might work
>really well, they pull about 70 watts ! This has got to produce some heat.
>
>I wonder if they have special high temp fans for use with a Peltier, it seems
>to me that the heatsink is going to be quite hot. While that'll increase the
>dissipation rate, I have some difficulty accepting more heat as a good thing.

It has some problems, yes. Basically all a Peltier does is (by burning a
lot of power) put a big kink (discontinuity even) in the temperature vs
'distance from the CPU' graph .. instead of a CMOS transistor junction
at 110c dumping heat into bulk silicon at 90c dumping heat into a HS at
60c dumping heat into air inside the case at 35c dumping heat into a
room at 20c, you can now insert a Peltier and have the bulk silicon
sitting against something that's at 0c or lower, while the other side of
the Peltier device is at some ludicrously high temperature feeding into
the HS. All you have to do it source the extra watts for the Peltier,
and dispose of afterwards, since it all comes out as heat.

AFAIK you don't need a special fan .. it's not like the HS is going to
be glowing red hot, and the working bits of the fan are not touching the
HS anyway.

Your 70w estimate is probably low, ISTR Peltier devices for modern
processors run more like 200-300w (some rule of thumb like you need to
feed in about 2x the power that you expect the Peltier to be shifting ..
and processors don't run at 35w these days).

--
GSV Three Minds in a Can
Outgoing Msgs are Turing Tested,and indistinguishable from human typing.