I finally got finished the new round of testing and the results were a bit different this time. With an overclocked Q6600 in the system instead of the E6750, my heat load rose from around 125 watts to around 177 watts according to
this page, and I don't derate to 88% like is suggested on that page due to that page not being updated since the introduction of the Core Architecture processors (it's based on the Netburst Architecture). I also monitored and logged case temps as read by the onboard temp sensors on the GA-P35-DS3L mobo with Speedfan. I know that those sensors aren't the most accurate in the world but they are better than nothing and should show any differences in monitored temps during the tests even if the temps are not absolutely right. I used an average of the case temps as logged for figuring my C/W values. Also, I used the linked page to give me some C/W values for the Ultra 120 with each different fan installed.
Also, there was a new fan test addition in this round of testing. ancalime was nice enough to buy and ship me a Zalman ZM-F3 case fan for this round of testing and I do appreciate it man.
BTW, it held up very well in testing too.
Once again, the heatsink was not removed or otherwise disturbed during the testing; just the fans swapped out. Here are the results I recorded in this round of tests.
Some further data for you to know:
I had 3 fans have a worker thread error during the tests, and they correspond to the fans with the worst average temps recorded. The Noctua NF-P12 had worker thread #3 error out 15 minutes into the Prime95 torture test, running large in-place FFT's. The Scythe SY1225SL12M Slipstream had worker thread #3 error out 14 minutes into the torture test (same test as before). And the Yate Loon D12SM-12 had worker thread #4 error out at 17 minutes into the torture test. When these errors occurred, I cut off the temp monitoring at that point, since the processor wasn't outputting maximum heat and averaged the temps from that point. However, I've been running the Noctua fan on the U-120 for the last 30 hours or so while running Seti with no problems at all, so I imagine that in normal usage even these three fans would be adequate for cooling on an U-120 at this heat load.
There were a few surprises in this round of testing for me. The first was how much difference the extra 50 watts or so of heat dump into the Ultra 120 made it kind of marginal for my cooling needs. I forsee another Ultra 120 Extreme purchase in my future as soon as I can afford to buy one. Another surprise was how much better the noisy Sunon fan performed with the U-120 when it was struggling to get rid of the extra heat from the Q6600. It was by far the best performer of the fans this time around, but is still too damn noisy for me to live with in 24/7 usage. Much better choices for 24/7 usage would be the Panaflo M1BX, Panaflo L1BX, Zalman ZM-F3, Delta EFB1212LE and the Scythe SFF21F S-Flex. As a matter of fact, the Scythe even beat out the two Panaflo fans by a little bit, which surprised the heck out of me. I din't think a 25mm thick fan could outperform a 38mm thick fan in the same flow ranges and the M1BX actually is rated for about 20% higher cfm. The Zalman also had a little better C/W value than the two Panaflo fans.
I will do some experimenting tomorrow and see if I can record a 10 second clip of each fan running and post it here for everyone. That way you can actually hear the noise (or lack of) from each different fan.
