|
Subject: Re: Improving the effectiveness of heatsinks?
Name:
 ludicrous
Date: 3/25/2003 9:13:21 PM (GMT-7)
IP Address: 138.67.74.229
In Reply to: Re: Improving the effectiveness of heatsinks? posted by
Dan Druff
Message:
Okay,
What you are evidently doing is working ONLY because you are using a CPU that is pretty old and not excessively hot and has a (relatively) large die area inside inside the package -- and also because you appear to be accessing the cartridge heatplate, not the CPU die itself. The heatplate is spreading the heat out to the heatsink where it is still being picked up and dissipated by whatever portion of the sink you didn't drill away. Ironically, your success probably stems from either a heatsink, or heatplate, that was warped in the middle and hence not making very good contact (which, with an overclocked cartridge PII, would likely cause overheat lockups).
Air cooling is a function of at least three things: The surface area of the cooling device, the air itself (temperature, humidity, and volume of air per unit time -- in this case here we're only really worried about the volume), and the cooling properties of the heatsink.
The primary purpose of a heatsink is to provide a very large surface area for heat removal, because air is not a particularly efficient heat transfer medium and there is a small propagation delay for heat through the sink (this is why copper coolers are more efficient than aluminum: Al is good at rapid heat transfer, but Cu is better; Cu however is more expensive, which is why Al is still used a lot. You can see this same principle at work in higher-end stovetop cookware). By providing a large surface area, you can get by with "just" a 40 or 60 or 80mm fan running in the 4000-7000RPM range (depending on the size of the heatsink and your tolerance for noise).
It is theoretically possible to cool, say, a top-end AthlonXP using only air blowing on the die -- but the volume of air necessary to accomplish that would have to be coming from a compressor nozzle, at sufficient force to punch the die right out of the processor package and possibly out through the back side of the motherboard. Not really joking, either.
This is a long way of saying that "blowing on the die" is a false way of looking at it, because the die (or even its heatplate, as is the case with K6-2s, all Skt370 CPUs, and the P4) does not have a lot of surface area. You may succeeded with your 333@400 (but possibly for other reasons, as described earlier) and may even succeed with your K6-2, but if you perform this trick on a PIII, you will probably suffer an overheat-induced lockup. If a P4, it will go into thermal-throttling mode first, and eventually lock up. If you do this with any Athlon or Duron...it will die, and probably physically burn, also damaging the mainboard in the process.
|