Test Setup & Methodologies
Test Setup & Methodologies
Click on image to enlarge
Test Platform:
Testing will occur on a Highspeed PC Standard Top Deck Tech Station, and not in a traditional case. This allows easier access to the motherboard for the constant poking and prodding that is required during the reviewing process. The setup remained as pictured during the duration of the benchmarking and stability overclocking process.
Keep in mind that not all processors are going to be able to handle running memory at DDR3-2000, even despite the slightly loose 9-9-9 timings. This is because Core i7 processors feature an integrated memory controller (IMC), and that component will ultimately dictate how high your memory modules can reach and at what timings.
Overclocking Methodology
Although the F3-16000CL9T-6GBTD already comes very highly clocked we are definitely going to find out what this triple-channel DDR3 memory kit is truly capable of. The overclocking section is the part of our reviews that we take the most pride in, and we spend an
excessive number of hours testing, tweaking, failing, and succeeding in order to give you the best possible insight into each product's overclocking capabilities. After all, if you are anything like us, the overclocking section is the first (and often last!) part that you read when checking out a product review.
For the purposes of this review, our overclocking efforts will primarily focus on four different timings configurations (6-7-6/7-8-7/8-9-8/9-9-9) and three different voltage settings (1.55V-1.65V-1.70V). The QPI/DRAM voltage was increased up to a relatively high 1.53750V BIOS (1.50V measured) in order to ensure as best as possible that the integrated memory controller would not be a bottleneck. The CPU frequency was kept as close as possible to the stock 2660Mhz, the Uncore was twice the memory frequency, and QPI Link was kept near 4800-5000Mhz.
During our overclocking adventures we put an emphasis on stability. While the question “What is stable?” could be debated endlessly, we have devised a methodology that combines a wide range of programs that test the stability of the entire system.
Here is the suite of applications that will be run in order to validate each of the overclocks:
- Eight 32MB instances of SuperPi (ran at the same time) via HyperPi 0.99b
- 3+ hours of Eight-Threaded Prime 95 v25.9 using the Stress Testing Blend
- 3+ hours of Quad HCI MemTest in Windows using all available memory
- Multiple loops of 3DMark Vantage (30 minutes of looping the full tests each)
- 1 hour of game play in Left 4 Dead & World in Conflict @ 1680x1050
Altogether, the above suite should provide enough stress testing to ensure a completely stable overclock, however we are always up for new suggestions. As always, no two systems are ever alike, so your results may vary. Also, <b>overclock at your own risk!</b> The Core i7 platform is brand new, and if you try to mimic our results there is always the possibility that you could damage any and all of your components. If you aren’t fully confident in what you are doing, feel free to stop by
our forums and our helpful community will be glad to offer some assistance.
Benchmark Methodology
For this review, our benchmarking section will focus solely on the G.Skill kit, in stock and overclocked configurations. The reasoning behind this approach is that there is an infinitesimal clock-for-clock performance difference between two memory kits that are equally-clocked and with identical timings, thus declaring one product a winner based on a 1% performance advantages seems pointless to us. Since the Core i7 platform and triple-channel memory kits are somewhat new, we have decided to test the memory at various clock speeds and timings in order to demonstrate the effect they have on overall system performance.
We have outlined the three benchmarked configurations in the sample graph above. These particular memory speeds and timings were chosen to demonstrate this memory kit's flexibility, and because we are confident that most samples of this particular memory kit will be able to achieve similar results. The glowing
red results will indicate the performance at the default timings. For the DDR3-1600 6-7-6 and DDR3-1866 7-8-7 configurations, the CPU is running at stock 2.66Ghz (20 x 133), with a 4800Mhz QPI Link. By default, the Uncore frequency is twice the memory frequency, so it was set to 3200Mhz and 3733Mhz respectively. For the DDR3-2000 9-9-9 configuration, the CPU is clocked at a slightly elevated 2.72Ghz (19 x 143), with a 5148Mhz QPI Link, and we used the 2:14 memory multiplier to achieve the high DDR3-2000 memory speed. At this memory speed, the Uncore is running at 4000Mhz. To further review the specifics of each benchmarked configuration, click on the thumbnails below:
As was mentioned in the overclocking section, we used Everest bandwidth benchmarks and a few other performance tests to determine which settings were ultimately faster, not just higher. 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) Intel 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 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. That is about all you need to know methodology wise, so let's see what kind of numbers this memory kit has achieved in our overclocking section and benchmarking suite.