ces wrote:
MikeC
It appears that you used two different test systems for the Noctua NH-L12 and the Noctua NH-C14.
In your opinion, how would the Noctua NH-L12 (with a single Nexus 120mm top fan) compare with a Noctua NH-C14 (with a single Nexus 120mm bottom fan).
They would both be close to the same height. Because the Noctua NH-C14 without a top fan would need less fan breathing clerance than the Noctua NH-L12 with a simple top fan... perhaps their effective height would be about the same. What do you think about their respective cooling capacity?
We've had two HS test platforms for a while -- a 1366 high power one for the biggies and a 95W TDP AM3 one for the rest. In the past, there were many more biggies, but the need for them is declining, actually, even for low noise cooling of heavyweight CPUs, because cases and HSF both emphasize airflow so much now.
We are going to actually change the AM3 for an 1155 platform, mostly because we want something with even lower power -- 55W at AUX12 is typical for max load on 65W TDP CPUs from Intel these days, and that looks like a good chip to test the really small HSF with (like the ones from Scythe, and a few others). OC that chip by 16%, and we get ~80W at AUX12V, which is about what you see with 95W Sandy & Ivy Bridge, so that will make a nice mid-power HS test platform.
To answer your question, the two heatsinks are not the same size, with different cooling surface area. The larger one will have the higher cooling power, but the real question is whether it will show up in a significant way in actual use. Unless the HSF at a certain airflow/noise level is overloaded by the heat load, then there will be no appreciable difference between the two. Push them to the cooling limits, and the bigger one's advantage will show up. In other words, with a >95W TDP CPU, there will be a clear advantage to the Noctua NH-C14, but if we're looking at <65W, then no. In between those 2 points is a bit of a gray area...
All this assuming the same ambient temps as our lab -- which is too cool; need to add 10C or so to simulate conditions inside a case... but then our test conditions are extreme, too. Prime95 at max load is NOT a normal high load; it is a torture test.
An Aside -- When we do tests at multiple fan speeds (and fans), it makes for quite a complex set of resulting data. In the good old overclocking days (in still in many OC sites), the stress testing is done only at max speed. This provides a single C temp rise per watt of CPU heat number -- C/W. In our tests, we'd have a C/W number for every heatsink at every speed we test it at.
And that really is the way you have to think of it: If I want 15 dBA/1m with no more than 20C rise for my 95W TDP CPU, then what C/W do I need? You could calculate the C/W for various HSF combos we've tested, for each fan speed we tested it at, and answer that question.