Overclocking

As cooling solutions do a better job of keeping the CPU at a lower temperature, it is reasonable to expect the overclocking capabilities of the CPU will increase. In each test of a cooler we measure the highest stable overclock of a standard X6800 processor under the following conditions:

CPU Multiplier: 14x (Stock 11x)
CPU voltage: 1.5875V
FSB Voltage: 1.30V
Memory Voltage: 1.90V
nForce SPP Voltage: 1.35V
nForce MCP Voltage: 1.7V
HT nForce SPP <-> MCP:
Auto

Memory is set to Auto timings on the 680i and memory speed is linked to the FSB for the overclocking tests. This removes memory as any kind of impediment to the maximum stable overclock. Linked settings on the 680i are a 1066FSB to a memory speed of DDR2-800. As FSB is raised the linked memory speed increases in proportion. The same processor is used in all cooling tests to ensure comparable results.

Highest Stable Overclock (MHz)

The Noctua NH-U12F with the stock low-noise fan reached 3.83GHz in the overclocking test. This was below the top coolers evaluated so far, which generally reached 3.90GHz with the same CPU and configuration. The SilenX high-output, low-noise fan did not improve these results, which was something of a surprise. However, using two Noctua NF-S12 fans in a push-pull configuration allowed the overclock to reach 3.88GHz, just below the typical OC for the best coolers we have tested.

The 3.88GHz is so close to 3.90GHz as to be an insignificant difference in performance. However, if you refer back to stress temperatures at 3.88GHz, you will clearly see that the Noctua with push-pull fans is not cooling nearly as well as the very best coolers, like the two top Thermalrights and the Tuniq Tower 120. This is still excellent performance in overclocking for a cooler that was clearly designed for quiet first, but we hoped for a bit better performance at the top given the similarities in appearance between the Noctua and the Thermalrights.

The best air cooler tested topped out at 3.94GHz. The rest of the best air coolers reached 3.90GHz. 3.88GHz is excellent performance, but no challenge for the best tested so far.

Scaling of Cooling Performance Noise
Comments Locked

30 Comments

View All Comments

  • Hulk - Wednesday, April 25, 2007 - link

    1. Equip all fans with a variable speed controller and adjust the speed so that the decibel level is exactly 40dB, 45dB, etc... Record the temps. This would isolate the thermal transfer efficiency of the cooler. For better isolation of cooler thermal efficiency you could use the same fan on all coolers.

    2. This time adjust the fan speed to acheive the same load temperature. This time record the fan RPM and the noise level. Of course some units might not even make the temp at full speed but that would be okay. As it is right now you have to take into account cooling performance when looking at noise. It be nice to see how much noise they make when they are all drawing away the same amount of heat from the processor.

    3. In order to really give these coolers a workout how about adding an overclocked quad core to the tests?

    I enjoy reading these reviews. Great job!

  • yacoub - Wednesday, April 25, 2007 - link

    No word on fitment again. Fitment is really the number one thing people want to know: Will this fit on my board? If it doesn't fit their board, it doesn't really matter to them how well it cools. Would be great to hear if it does or doesn't clear a variety of boards you have around the testing labs, particularly a couple 680i, 650i, P965, and 975. Pick one that's fairly reference in design and a couple of the ones people tend to purchase that often have passive cooling heatsink/pipe configurations on the northbridge and MOSFETS, and let us know how well the darn thing clears it all, especially if it requires a back brace and some boards have stuff on the back beneath the CPU socket (like the MSi P6N-SLI Platinum) and whether or not it still fits.

    Thanks.
  • Wesley Fink - Thursday, April 26, 2007 - link

    The Noctua is narrower than most heatpipe towers, but as wide as a thermalright. The shape is much like a thermalright. The top plate can be installed in two directions and the cooler can be turned 90 degrees if necessary, so mounting is very flexible.

    The Noctua fit the EVGA 680i (barely, due to width and very tall chipset cooler near the socket), Asus, Striker, Asus Commando, and Asus P5W-DH Deluxe. We did not have the MSI board you specify in the lab to check, but the back plate has an open center and is designed to clear back components.

  • yacoub - Wednesday, April 25, 2007 - link

    Any word on fitment on an MSi P6N-SLI Platinum? This is important as the board supposedly has a few items on the back such that certain backplates do not fit too well and it would be important to know if this hsf can clear them alright.
  • mostlyprudent - Wednesday, April 25, 2007 - link

    Very nice. I like the balance of low noise and above average performance. Any plans to look at the Thermalright HR-01 fanless cooler with the ducting that can connect to a rear case fan? I would love to see the cooling abilities of such a configuration.
  • xsilver - Wednesday, April 25, 2007 - link

    hey wes; how many more hsf setups are still in the pipeline for review? care to list?
    is the thermalright ultra 90 one of them?
    or anything else that is of the "cheaper" level?
  • Wesley Fink - Wednesday, April 25, 2007 - link

    We have a few more top-line HSFs and then we will do an "under $30" HSF roundup. We also have a wide assortment of 120mm fans in the labs for a fan roundup.
  • xsilver - Wednesday, April 25, 2007 - link

    oh also, was there any plans to put some "classic" coolers in the chart as a control reference?
    Im thinking thermalright xp-90/120
    zalman 7000/7700

    how do these type of coolers compare to the ones currently being reviewed? no full review is really needed but putting them in the charts would be nice.
  • puffpio - Wednesday, April 25, 2007 - link

    The 'Final Words' page is actually a duplicate of the 'Noise' page
  • Wesley Fink - Wednesday, April 25, 2007 - link

    The posting error on Final Words has been corrected.

Log in

Don't have an account? Sign up now