The System

The aluminum case is reducing the temperature of some peripherals substantially and also the mobo temperature by around 2C, but I have NOT seen ANY influence on the CPU temperature. So, for this test we can therefore neglect the impact of the aluminum case. Also: for this test ALL 9 case fans were disengaged, and the case was kept open. The purpose of this comparison is mainly to show what you gain and not so much to test the maximum MHz with this combo. It was possible to run this specific 1.2 Giga Athlon at 1450 (10x145) and run benchmarks with the Taisol combo. For absolute stable operation the Taisol combo was tested at (10x141) 1410 MHz and the PAL6035 combo at (10x138) 1380Mhz.

Taisol Test Settings

PAL 6035 Test Settings

Temperature Taking
According to AMD at 1200 MHz an Athlon has a heat output of around 66 Watt! And at 1400 MHz the same CPU has an output of 76 Watt!! That is a lot of heat to be handled with and that is the reason why heat sink performance is of greater concern for AMD CPU’s than for the cooler running Pentium III. Temperatures for both HSF combos were taken after 30 minutes CPU idling and after 30 minutes running the "CPU Stability Test" 6.0 from Vuorio. Room temperature was between 19C and the mobo temperature between 23C. Temperatures were taken for the following 3 conditions:

The handy (I love this warming up function) Vuorio CPU Stability Test is available here.

All temperatures were taken and recorded by the Asus Probe V2.12. I know that his is not the most accurate way of taking the CPU temperature. However, the purpose was more to show the difference between the 2 adapters and the 2 heatsinks and for this relative measurement the Asus Probe might suffice.

For the sake of good order, I tried also to couple 2x80mm fans to see whether an additional impact could be achieved. That was not so, the CPU temperature didn’t go down a single C! I assume that the fans would have to run perfectly synchronous to have an added impact on the air-stream, and that of course is never the case.

The Athlon 1200 EV266 running at 1410 MHz

Lowest taken temperature: 90mm combo on Taisol, after 30 minutes CPU idling

System Summary for the lowest taken temperature: 90mm combo on Taisol, after 30 minutes CPU idling

Taisol Results

Alpha Results

Combined Results

Interpretation of the Results
The Taisol CGK761CU-BIG is proving a 20% better cooling performance under load "Stability Test" when used with a 90mm fan + adaptor combo instead of a 60mm/6000 rpm Delta fan mounted directly on the sink. This is of course mainly resulting from the much bigger amount of air the 90mm fan is moving. But I guess to a small part it’s also resulting from the enlarged total heat sink surface through the additional aluminum surface of the adaptor. It should be not forgotten that I compared the 80mm and 90mm fans with the infamous 60mm Delta fan, which is the strongest available 60mm fan: the gain would be much more when compared with standard / stock 60mm fans moving only around 20CFM.

With the PAL6035 (and I guess it would be similar with a FOP32) you can lower the CPU temperature up to 10% by using a 90mm fan + adaptor compared with the strongest available 60mm fan. The advantage of the 90mm/80mm adaptor+ fan versus the 60mm Delta is smaller on the PAL6035, because the smaller heatsink surface of the PAL6035 is becoming the limiting factor. You cannot blow more heat out of the heatsink than the heat sink is capable of absorbing from the CPU core! The prevailing factors in this regard (thermal resistance, surface, CFM etc) are explained by Ed Citarella here and also here. In plain words: to take full advantage of the additional CFM provided by a 80mm or 90mm fan, you need one of the larger heatsinks like the Taisol or the newer Global Wins. The performance difference between an 80 and 90mm adaptor/fan combo seems to be minor, but is still clearly visible. And I definitely recommend the 90mm combo.

Conclusion
A 90mm Delta fan on a performance heatsink like the Taisol CGK761CU-BIG can cool down your highly overclocked Athlon Thunderbird under full load to around 28C. This is 20% better than what is possible with a high performance 60mm Delta fan. With the Taisol and the 90mm fan this Athlon 1.2 Giga could load WinMe up to 1450Mhz. Absolutely stable operation with this HSF combo was possible at (10x141) 1410 MHz. To run an Athlon 1200@1410 MHz at 28C under full load is, I assume, not very far away from water-cooling. I think these adaptors are an interesting option to try before resorting to the more dangerous and more expensive water-cooling.