Benchmark MethodologyIn all of the benchmark graphs including the sample graph to the right. For recent DDR3 reviews we have seen comparisons to DDR2-1200 5-5-5, DDR2-1000 4-4-4, and DDR2-1120 4-4-4-8. We figured it is about time to quit beating a dead horse and cut out the DDR2 comparisons. It has been seen on all three occasions that despite DDR3 providing higher bandwidth and equal latencies, this does not equate to a substantial increase in either synthetic or real world benchmark results. So today we will simply be taking a look at a couple memory setups in the benchmarks.
As always, the grey bars in the graphs (1 / 2) will indicate the rated frequency and timings of the Corsair Dominator memory. For these grey results, all but the primary memory timings are left on AUTO in the BIOS. This will best simulate what the average user would have for a setup on a 790i board. We have gone with two sets of 'stock' settings. One, as if the average user left the CPU at stock and simply un-linked the memory and set 1800 in the BIOS. The other is with a mild overclock to 400x9 with the memory set to 1800 as well. Naturally, the first grey bar isn't exactly DDR3-1800 but as close as un-linked memory would allow with a 333MHz bus speed (1333 FSB). This motherboard is nice because it allows this simple method in order to run the memory at the frequency it is rated for without having to overclock the system bus or only to overclock the system bus marginally like we have with the second set of results.
The last three red results in the graphs (2 / 3 / 4) are reserved for the 'non-stock' settings. With the 790i motherboard we have complete freedom with what we want to compare so we took this opportunity to explore the memory performance of the 790i chipset a little bit with a direct comparison of 6-6-5 at close to its maximum stable overclock, 7-6-5 at near its max, and 8-7-6 at its highest stable clocks we can achieve. The CPU frequency, FSB, and all other variables remain the same, all that changes is the memory frequency and timings, thanks to the un-linked memory ability. This essentially provides a direct comparison between the three memory frequencies at the height of their potential with this kit. That is of course as long as the different ratios play no role in the performance.
For all of the benchmarks, appropriate lengths are taken to ensure an equal comparison through methodical setup, installation, and testing. The following outlines our testing methodology:
a/ Windows is installed using a full format.
b/ NVIDIA Chipset drivers and accessory hardware drivers (audio, network, GPU) are installed followed by a defragment and a reboot.
c/ Programs and games are then installed followed by another defragment.
d/ Windows updates are then completed installing all available updates including .NET Framework followed by a defragment.
e/ Benchmarks are each ran three times after a clean reboot for every iteration of the benchmark unless otherwise stated, the results are then averaged.
We have listed the benchmark versions on each graph as results can vary between updates. Normally we would briefly discuss the effect tRD has on performance but this review is taking place on an NVIDIA chipset and not an Intel like previous memory reviews so there is nothing to discuss really. What you see is what you get, there is no monkey business or adjustments going on in the background that need to be brought into the light. | 
LinkBacks (?)
