Heat Tech Question

[Tuff]

Can anyone tell me if any research has been done about how much heat is generated at the sides of the CPU die of an Athlon cpu? And if so...would it be a good idea to try to remove it. As far as I can tell the heatsink does not come in any kind of contact with the sides..unless you are sloppy with the thermal grease. I will be visiting a machine shop tomorrow to see if anything can be made for the purpose of removing the heat from the sides. If anyone has any idea's or knows of any results...please post them here.

Tuff

 

[JC]

I think there's such small surface area on the sides that it wouldn't be worth it.

m2c

JC

 

[Tuff]

I am still curious though....I do not have my cpu in front of me atm...so I do not know how much surface there is there...I would assume all added together it would be roughly 10% of the total die size maybe even 15%. Intel seems to think its a good idea with their heat spreader as it touches the sides.

Tuff

 

[JC]

Hmm....I guess there are heat spreaders available for Athlons, aren't there?

 

[mechBgon]

Have you cut open an Intel CPU to confirm that the heat spreader really does touch the sides of the die? Seems to me like getting the manufacturing tolerances tight enough to ensure that side contact didn't prevent top contact would be very, very difficult, not to mention increasing the risk of core breakage from the side pressure.

Interestingly enough, some AMD laptops do have a setup where the bottom of the CPU package is in contact with a heatsink that drains heat to the bottom of the case itself. Can you say "sandwich?" Maybe in the coming years we'll see CPUs go back to a slot format, but with a heatsink on each side of the core and a ducted downdraft/crossdraft fan...

 

[ChampionAtTufshop]

Originally posted by: Tuff
I am still curious though....I do not have my cpu in front of me atm...so I do not know how much surface there is there...I would assume all added together it would be roughly 10% of the total die size maybe even 15%. Intel seems to think its a good idea with their heat spreader as it touches the sides.

Tuff

actually it doesnt touch the sides impretty sure

 

[mechBgon]

I have some old PPGA Celerons at work. Maybe sometime I'll take a hacksaw to one and see what's in there

 

[Jeff7]

I agree that it would take considerable accuracy in machining the heatsink to get it to fit on the core. There's another problem with that - the sealant that is used around the core is not always applied the same on each processor - if the heatsink would hit that before it contacts the core surface, the CPU could easily overheat. Plus, the forces used when securing the heatsink to the motherboard can be a bit harsh - assuming a heatsink did exist that fit snugly on the core, the slightest rotation would cause severe core damage.

 

[RalfHutter]

Originally posted by: mechBgon
Have you cut open an Intel CPU to confirm that the heat spreader really does touch the sides of the die?

The Intel PIII and PIV heat spreaders don't even come close to touching the sides of the die. The die is similar in size to an AMD die and you know how big the Intel IHS is, right? There must be at least 3/8" of space between the die sides and the IHS. Look at the drawing on page33 of this Intel P4 datasheet pdf to see a cut-away illustration of the package.

 

[Tuff]

Ok...yes it would appear that Intel's heatspreader does not touch the sides. Now would a heat spreader of this nature benefit the AMD Cpu? I am thinking it would give a larger area for the heat to be pulled off the cpu? As the spreader would be a thin Copper metal it would disperse the heat quickly away from the die. Later to be scooped up by the heatsink.

Does anyone know if a thinner piece of copper will disperse the heat quicker than a thicker piece?

 

[Nohr]

It sounds like a good idea in theory, but in reality it'd probably be very difficult to make an exact fit.