Hey guys, I M working hard on several mods for my Athlon XP rig and also CPU cooling I was wondering if putting a heat spreader on my Barton would have a positive/negative effect on the temps (die won't crack) or would a shim have the same effect?
Use a shim if you want to protect the core, think of every layer between die and heatsink as a nice warm snuggly blanket.
Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.
I don't think (s)he'd see a huge increase in temps (if any at all) with a copper heat spreader, provided a good thermal paste is used.
You can delid old s478 CPUs and use that, I think the cap is copper and should work with the athlon core's height. Just lap it to get rid of intel markings, I think it'd be cool... 😎
Outrigger: an ongoing adventure with the OR840
QuForce FX 5800: turn your Quadro into a GeForce
The thermal conductivity of all thermal pastes is utterly crap... utterly....compared to any metal, it just so happens though that it's 10x better than air, which is what you've got left in there if you don't displace it with thermal paste. However, while one thermal interface is essential, two aren't.
Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.
Oh, the memories of the shim arguments, some swore by them, others that they were useless or more likely to cause damage.
An extra layer and extra thermal interface would probably not be good - look at the complaints over Intel's TIM.
With the small size of the core, the quality of the TIM is probably more significant than with larger contact areas
wrote:The thermal conductivity of all thermal pastes is utterly crap... utterly....compared to any metal, it just so happens though that it's 10x better than air, which is what you've got left in there if you don't displace it with thermal paste. However, while one thermal interface is essential, two aren't.
That's one side of the coin. The other is actually cooling more area from the heat spreader. TIMs have an order of magnitude less heat conductivity than copper, true. At the same time your heatsink will be acting on an area 4 or 5 times bigger. Note I didn't say lower temps would be achieved, I said similar or equal.
wrote:An extra layer and extra thermal interface would probably not be good - look at the complaints over Intel's TIM.
With the small size of the core, the quality of the TIM is probably more significant than with larger contact areas
1st statement is a different problem altogether. People delidding today still recap, they just replace the TIM.
2nd statement is debatable: that's the entire purpose of a heat spreader. EDIT: sorry, misread your comment. Agree that high quality TIM is important.
Outrigger: an ongoing adventure with the OR840
QuForce FX 5800: turn your Quadro into a GeForce
Just use a shim to keep the heatsink from tipping when putting it on. I used them back in the day for Athlon chips and I use them today for Athlon chips.
A heatspreader will get you absolutely nothing if used with a CPU that was originally a bare die. All you are doing is adding an extra layer of TIM and an extra layer of metal in between the core and the heatsink.
As long as your heatsink is a good nice flat surface where the CPU die meets up, it will cool better with the bare die than with a heatspreader added into the mix.
Shim, a heatspreader that's not perfect will either leave to much gap or you just run the rick of cracking the die anyway
wrote:or you just run the rick of cracking the die anyway
This is particularly true if the heatsink mounting system doesn't allow any adjustment and it increases pressure on the die.
Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.
I use a Tualatin heat spreader on my Pentium M. It looks just like a P4, and the IHS lets me use standard S478 coolers. I've got that sucker overclocked to 2.72GHz, and even with a $10 Cooler Master, load temperatures never exceed 36C. But for an Athlon XP? I can't think of any good reason you'd need a heat spreader.
94 MHz NEC VR4300 | SGI Reality CoPro | 8MB RDRAM | Each game gets its own SSD - nooice!
wrote:I use a Tualatin heat spreader on my Pentium M. It looks just like a P4, and the IHS lets me use standard S478 coolers. I've got that sucker overclocked to 2.72GHz, and even with a $10 Cooler Master, load temperatures never exceed 36C. But for an Athlon XP? I can't think of any good reason you'd need a heat spreader.
I got 3 temp sensors on my cpu,, 1 for the socket, 1 for the die and one for the package. The die is usually 10 degrees hotter then the socket and the HS, even though I have a Salman cnps-7000 to cool it. Since I am currently preparing my chieftec dragon for a custom loop was wondering if a heat spreader would solve the athlon high temperatures combined with a custom loop. Seems this would not be needed and a shim be plentiful (I use arctic silver mx5 thermalpaste for all my components)
If you have an old Socket A chip and K6-2 to experiment with, try taking the heat spreader off of the K6. It'll probably increase temps, but it'd be interesting to try. They look very similar without the IHS.
Now for some blitting from the back buffer.
wrote:If you have an old Socket A chip and K6-2 to experiment with, try taking the heat spreader off of the K6. It'll probably increase temps, but it'd be interesting to try. They look very similar without the IHS.
I actually did this back in the day. I didn't have any real testing equipment but if you have a good cooler, removing the heatspeader off of a K6/K62/K63 will lower temperatures. It is basic physics. The TIM in between the die and the heatspreader is going to hurt heat transfer. There is no way around that. The least temperature transfer hindering way if you have a heatspreader is to have the die soldered to the heatspreader like Intel does on some of their CPUs.
wrote:wrote:I use a Tualatin heat spreader on my Pentium M. It looks just like a P4, and the IHS lets me use standard S478 coolers. I've got that sucker overclocked to 2.72GHz, and even with a $10 Cooler Master, load temperatures never exceed 36C. But for an Athlon XP? I can't think of any good reason you'd need a heat spreader.
I got 3 temp sensors on my cpu,, 1 for the socket, 1 for the die and one for the package. The die is usually 10 degrees hotter then the socket and the HS, even though I have a Salman cnps-7000 to cool it. Since I am currently preparing my chieftec dragon for a custom loop was wondering if a heat spreader would solve the athlon high temperatures combined with a custom loop. Seems this would not be needed and a shim be plentiful (I use arctic silver mx5 thermalpaste for all my components)
This is normal. You are never going to get the CPU die to be the same temp as the heatsink or the socket, especially under load.
Direct die liquid cooling will get you closer, but the die is still going to be a bit warmer than the liquid that is cooling it.
