Test Setup

The SPD of the modules is exacly what you would expect. The model number reflects that of each module - CM2X2048-8500C5D. This indicates that these modules are rated for 1066MHz at CL5, also reflected in the EPP#1. We will see later on, how far we can take the DIMMs during testing.

Performance

Placing the memory on the board does not pose any problems. The heatsinks are thin enough to fill all DIMM slots if you wish. Installing the Airflow is quite easy as well. It does take up quite some space in your case, so you want to make sure that the area above your DIMMs is free of cables and the likes.

The first step was to test the memory at its specified settings to set a baseline for which to compare overclocked benchmark results. From there, the maximum memory clock was sought at stock voltage and timings to see just how much overclocking potential this kit has out of the box. As the table shows, the result was phenomenal considering the high density of these modules. 1026 MHz was the max overclock reached at CL4 with stock voltage.

The remaining tests show how well this kit works with a range of voltages at specific timings. The kit was tested at CL3, CL4 and 5 timings to find the max stable memory clock at 1.8, 2.0, 2.2, 2.3 and 2.4V each.

After the factory specified settings were tested it was time to see how far beyond their rated speed these modules would go while remaining stable at each voltage setting. Using the intended CL rating of 5, yet not altering the voltage from 2.1V, they managed to break 1100 MHz stable. However, if one feels the need to push the envelope, they are also capable of 1200 MHz at CL5 and 2.4V.





As the voltage scaling graph shows, this particular memory kit seems to scale quite nicely until after 2.2V. After 2.2V, clock speed increases were very low at 2.3V and 2.4V, posing the question as to whether the mild boosts are worth the risk of running higher voltages. I should finally note that, while these results are a good indication of expected performance, no two kits will perform identically and that these results are not guaranteed.