Test Setup

Performance & Overclocking

As soon as the memory was installed, CPU-Z was used to check the SPD programming and as it turns out, there are three somewhat exotic speeds present. The DIMM Kit is rated for 1600 MHz, but the maximum set is 1218 MHz. The CL rating is also advertised as 8-8-8-20, while the kit is programmed at 8-8-8-23.

The memory was first set to the intended setting of 1600 MHz CL8-8-8-20 at 1.8V. The FSB of the CPU was set to 400 MHz and the highest divider was used. To make sure that the Intel Core 2 Duo was not the bottleneck, the multiplier was lowered to 6 from the standard 7x. This setting was not changed while testing and the dividers were also kept at the same setting. This will give you a good idea as to how the CPU and memory scale with higher FSB. Only at 500 MHz did we use a unique divider to keep above 400 MHz FSB and thus 2400 MHz for the CPU.

The memory was tested at CL5-5-5-10, 6-6-6-12, 7-7-7-15 and 8-8-8-20. While it is possible to go higher with CL9 or CL10, it does not really serve any purpose, as such loose settings translate into slower performance. All milestones were benchmarked, as well as the highest possible at standard 1.5V or 2.0V - the maximum one should use for DDR3 with sufficient cooling. The Crucial Ballistix kit managed 1600 MHz at CL7-7-7-15 and just shy of 2 GHz at CL 8-8-8-20 with 1944 MHz. This is a very respectable setting, but 2.0V is required to reach this point. As you can see the FSB is fairly high as well with 486 MHz. From past tests we know that this CPU can easily go above 3 GHz and the board manages an FSB just shy of 500 MHz, so this is the limit of the memory.





The voltage scaling graph shows some great properties. The Crucial DDR3 memory is very balanced, as both a higher CL setting and an increase in voltage result in a speed improvement. Only at CL6 do we see an invisible wall, which the memory did not seem to be able to break, no matter how much we set the voltage at or even though we set the divider to 1:2. Please note that each individual module overclocks different, the results here can only be seen as an indicator of performance.