Heat Sink Designs a Schuck ?

[jboy]

I just purchased an Alpha PAL 6035, nice looking copper bottom cooler w/a Delta fan-highly rated by every cooler rating site I've visited, and decided to do a little test between it and the stock cooler that came w/my AMD 1.4 Tbird. Since presumably what were paying for is advanced heatsink design when we buy one of these things, I decided to test the Alpha w/the AMD cooler fan, to see how much more efficient a heatsink design the Alpha was vs. the AMD. I also decided to test the AMD sink w/the Delta fan, to see how much of a difference that made. The results-(and I ran these tests twice to make sure there were no anomalies, pulling my MB for both tests to visually confirm proper heatsink seating), same room temps and circumstances for all tests-(one hour video render which max's out cpu's 100%), ?? The AMD heatsink beat the Alpha in the test that counted, and was real close in the other. . Final temperature results after one hour render w/the stock AMD cooling fan; Alpha-55C, AMD-57C. Final temp results after same test using the Delta fan on both. Alpha-50C, AMD 47C. Has anybody else tried this.? I'm wondering if all this heatsink design stuff is just a lot of smoke being blown up our collective butts. In thinking about this, shouldn't all heatsinks be tested without fans ? We all know what cooling effect a given volume of air being blown over a heated surface will have, so why test heatsink/fans in combination ? Admittedly my little test wasn't exhaustive, I was using the MB sensor on a Soyo Dragon to measure things, but I'm sure my results are real. I'm real interested to hear if anyone else has tried such tests, and if, as I'm beginning to suspect, cooling efficiencies for most of these heat sink designs is more a function of the fans being used, than the design of the sink itself. Observations anyone ?

[Intraveinous]

Are your conclusions based on temps read from an in-socket thermistor? You really can't trust it at all, it's a measure of the temp of the air inside the socket rather than the temp of the die of the processor. Only real way to test is with a thermistor mounted inside the heatsink right above the chip (like Joe does) or with the built in thermistor that the Intel PIII and PIV, and the Athlon XP have.
Peace
John

[jboy]

I understand that board sensors are not as accurate as on die ones in an absolute sense, but arent they still accurate as strictly comparative indications of relative efficiencies between two different sinks ? After all, they both completely block the square on the Socket A, and trap the same amount of air, if the air under one sink is cooler than the air under another, then that sink must be transferring heat out thru the sink more efficiently, no ? I'm not trying to be argumentative, just trying to figure out what's going on. Are on board sensors known to be both erratic and inaccurate, or just inaccurate ?

[Sonny]

Originally posted by jboy
...but arent they still accurate as strictly comparative indications of relative efficiencies between two different sinks?No they will not. Because HSF designs are different. The air flowing thru the sink can influence the thermistor giving you false readings. Check the front page about why thermistors can not be trusted.