Riskfree to underclock a T-breed XP2400

[Ralf63]

The new upcomming T-breed XP2400 (version B) seems to be ar least as hot as an Palomino XP1800, but with a much smaller core. Not so good seen from a quiet cooling perspective.
Is it problemfree/riskfree to deliberately underclock such a XP2400 processor to, let say XP2000 speedlevels. The big idea here is to have at least Xp2000 prestanda, but perhaps with only XP1600 heat-development. Am I on the right track here?

[MikeC]

Yup, you're on the right track! While you're on this track, you might think about a copper heat spreader like the P4's. I came across an article somewhere recently where someone pried one off the P4, discovered it was just a piece of copper with a flange. The copper was NOT bonded to the core; simply had some thermal paste on it.

Anyway, since reading the piece, I've been thinking about a very thin piece of copper (so that it won't appreciably change the height of the core to ensure minimal extra pressure from HS clips), but with a wide flange about as "tall" as the core. In otherwords, the distance the core sticks up from the --um... the AMD CPU top. Let's see if I can do something graphic with the keys...

[[[[[[[[[[[[[[[[[[[[[[]]]]]]]]]]]]]]]]]]]]]]]]]]]
[[[[[[[[[[[[[[ xxxxxxxxxxxx ]]]]]]]]]]]]]]]]]
[[[[[[[[[[[[[[ xxxxxxxxxxxx ]]]]]]]]]]]]]]]]]

Does that make sense? The xxxxx represents the core. The whole thing would not be much more than 1" square & could be made from two pieces of copper, one very thin, the other just a hair under the height of the core, and bonded together in some fashion. The extra copper would help dissipate the heat.

Anyone here who has copper metal working skills, please speak up!!

[ChiefWeasel]

What about using a copper shim for the bit that goes around the core? They're only about £3 and would be absolutely perfect for this.

Hopefully this link will work...
http://www.theoverclockingstore.co.uk/p ... eid=800234

Then i think it would be quite easy to get a thin piece of copper to use for the top bit. Not sure how you would fix them together thou - Arctic Silver Thermal Epoxy?

Also wondering if this plan would actually benefit if you have a copper HS already? Also with a Au HS, im thinking maybe the extra layer of thermal paste would negate any benefit gained?

[MikeC]

A shim, of course! Thermal epoxy should work fine. Whether there'd be any gain? No sure, obviously I hope so. The heat spreader gives you a bigger contact area between core & HS & thus SHOULD improve heat transfer & reduce core temp. Whether losses between the core and spreader would cancel out any gain would be the big question. Won't know for sure till it is tried by a few people.

[siliconacoustics]

Chief,
You are correct that there is no possible thermal benefit from a copper heat spreader when used with a carefully attached copper heat sink, and in fact there is an increased thermal resistance because of the added thermal interface layer.

The conductivity of even the best thermal interface material pales in comparison to the conductivity of aluminum or copper, so the fewer layers you have of it the better.

My understanding is that the primary purpose of an integrated heat spreader (IHS) is to reduce localized hot spots on the die, and increase user friendliness by creating a package that is less susceptible to mechanical damage or thermal damage from sloppy installation.

Intel marketing material seems to consistently suggest that an IHS helps to better cool the processor. But the Intel technical material I have read suggests that the opposite is true, and that an IHS can actually double the overall thermal resistance.

Check out this Intel document for discussion about how the IHS makes cooling a CPU more difficult during burn-in testing.

http://www.intel.com/technology/itj/q32 ... hermal.pdf

Just search for “IHS” in this document to skip to the relevant parts.

Jonathan

[MikeC]

Interesting Jon, thanks for the insight. Was worth a shot; I guess my question about whether the added layer would negate any gains is YES.

[Jolt]

Hehe, now that was a bit dissapointing...

At least, I hope that they are using a good quality thermal interface material between the core and the heatspreader.

btw. This should mean that you could achieve better cooling if you removed the heatspreader from the P4, right?
Maybe that's a project that should be evaluated....

[ChiefWeasel]

This guy here pulled off his IHS:
http://www.hardforums.com/showthread.ph ... atspreader

Theres a few other threads about it at the [H] if you search for 'IHS' or 'heatspreader' in the title, on the Strictly Intel forum.

[MikeC]

From the various posts, looks like the minimum temp improvement by removing the IHS is 1C; the max is maybe 5~6C. Average is under 3C? Is this worth it? Probably not for most folks, unless you're trying to run one of the fastest ones quietly.