Following up on our initial Clarkdale review, we decided to take a closer look at which kind of speeds to expect from using various cooling methods on the Intel Core i5 processor. Thus we have tested maximum load stable speeds with the box cooler, a custom air setup, a liquid solution and phase change cooling.

As our lab in Copenhagen had no good H55 board other than the reference Intel platform, testing was done using the P55-based Asus Maximus III Formula, also cancelling out the possible use of the i5?s integrated on-die GPU.


Images of our various setups during testing.

For this small experiment we used a system comprised of an Intel Core i5 661 Engineering Sample, an Asus Maximus III Formula, 2×2 GB OCZ Animal DDR3-1333 CL7, a HIS Radeon HD 5770, a 128 GB Patriot Torqx M28 solid state drive and last but not least the new 1.6 kw Ultra X4 power supply.

For stability measurement the system had to pass at least 15 minutes of maximum processor stress test using the Linpack benchmark found in OCCT. Linpack is a software library running numerical linear algebra to determine a system?s floating point computing power, and can be particularly stressful demanding lots of intensive processor operations thus increasing overall load.

Our results are not the limit of what one can expect from overclocking a Core i5 661 and only serve as a guideline, but please, keep in mind that overclocking voids warranty and should only be attempted if you know what you?re doing.

Throughout all of the tests turbo mode as well as hyperthreading was enabled. Much to our surprise a solid 4 GHz are indeed possible with the Core i5 standard cooler, although temperatures rapidly climb to just over 60 degrees Celsius under load. Keep in mind that you?ll need good airflow inside your case for this.

Switching to the Thermalright MUX-120 with two ordinary 120mm fans in push-pull configuration made it possible to reach 4.4 GHz, and adding a high-powered Delta Scream fan instead proved that 4.6 GHz was not an issue either ? the temperature under full load was 47 degrees Celsius.
 

Various speeds reached with different types of cooling.

Strapping on our liquid cooling setup (EK Supreme, Laing DDC Ultra, 3x120mm dual-flow radiator) brought the speed up to about 4.8 GHz, yet 100 MHz less was needed for a stable Linpack run. Mounting a single stage unit however made things a whole lot more interesting as the Core i5 661 was flying through the number crunching at 5.1 GHz and slightly below 1.6 vCore. Although, that processor voltage might be a little much for daily operation though.

Conclusion
Intel’s Dual-core certainly packs the punch. Given the price at hand, it certainly has the overclocking potential and if Intel went with a discrete CPU design, this processor had the potential of becoming another Celeron 300A. Unfortunately, if you don’t plan on using the integrated graphics a good quad-core simply sounds like a better deal. This is not to say that Clarkdale isn’t without merit – we liked the overclocked results and the overclocking experience. Associated ecosystem needs to rise up to the challenge and provide a good overclockable motherboards based on native H55 chipsets. Companies such as EVGA recently launched their Clarkdale enthusiast boards and we’ll follow them in the future and report GPU overclocking experience. In meanwhile, you can check Sean’s Core i5 661 review here.