- These tests are not truly scientific, but they do use the on-CPU diode of the AthlonXP, accessed via MBM5 on an Asus A7V333-R using BIOS 1012.
- The CPU is a 1600+ AGOIA (Palomino, like all 1600+'s) overclocked to a 166/166 bus, for 1750MHz.
- Voltage is 1.775V.
- Arctic Silver 3 (just applied it today, so not "settled" yet)
- The temperature listed is the delta between case temp (measured by the mobo, not sure precisely where) and the on-core diode reading.
- The temperatures are under 100% full load using the hottest-running distributed-computing client I could find: ECC2-109. This client makes use of the MMX units, like Distributed.net's hot-running CSC client did.
Ok, first the baseline cooler: the
$7 Coolermaster DP5-6I31C-A1. 38dB, 5400rpm, with a lapping job sufficient to flatten the base:
27°C above case temperature
Next result: the $20 Thermalright SK-7 with an 18dB ~18cfm NMB fixed-speed
80mm fan. Pleasant to listen to.
CPU stabilized at 34°C above case temperature, then CRASHED.
Next: same as above but using the excellent Panaflo L1A 80mm fan. Pleasant to listen to.
CPU stabilized at 28°C above case temperature. So here we have about the same cooling as the whiny Coolermaster, but with low noise. This reinforces my opinion that the Panaflos have a good design that overcomes backpressure well.
Next up is the YS Tech adjustable. It's supposedly capable of a range from 10cfm to almost 50cfm. Because the system crashed using an 18cfm NMB, I think it's fair to say it isn't going to be stable with the YS Tech cranked down to minimum either. I might do a full-range report on the YS Tech with the 1600+ at its stock speed, however. Stay tuned for more details in a few...
SK-7 with YS Tech at maximum RPM: Still more pleasant to the ears than the Coolermaster.
Oops, did my math wrong just now. 22°C above case temperature, not 12°C over. D'oh!
SK-7 with YS Tech cranked up about 2/3 of the way, to where I just begin to "hear" it over the NMB case fans a bit:
25°C above case temperature.
Hmm, I think that's a good sampler anyway. Given the results, I may just go with the ol' Panaflo L1A here.