Corsair Vengeance Pro DDR3 2666MHz 16GB Reviewccokeman - September 11, 2013
» Discuss this article (3)
Corsair Vengeance Pro DDR3 2666MHz 16GB Testing:
Memory is often hard to separate from one kit to another in gaming, but when it comes to number crunching and computing, some memory provides an extra boost in comparison. To see just what kind of performance this kit has to offer, I will be running the modules through a series of benches to see just how they compare. There will be 8GB and 16GB kits ranging in speed from 2133MHz - 2400MHz, tested at native speeds as well as overclocked. Overclocking of course will be dependent on exactly how far the testing rig will allow, but I'll push it as far as I can. The testing setup used for these benchmarks is listed below, where Turbo Boost has been disabled to eliminate uncontrolled clock changes that may skew the results. The CPU will be run with default Boost clock speed of 3.9GHz for baseline testing and bumped up to 4.2 GHz for OC testing, or as close as possible to that speed. All current updates and patches are installed for Windows 7 Professional 64-bit and the latest driver for the NVIDIA GTX 770 will be used.
- Processor: Intel Fourth Generation Core i7 4770K
- CPU Cooling: Corsair Hydro Series H100
- Motherboard: ASUS Maximus VI Extreme
- Memory: Corsair Vengeance Pro 2666MHz 16GB
- Video Card: NVIDIA GTX 770
- Power Supply: Thermaltake Smart 750W
- Hard Drive: Corsair Force GT 240GB SSD
- Optical Drive: Lite-On Blu-ray
- Case: Corsair Obsidian 650D
- OS: Windows 7 Professional 64-bit
CPU-Z: This application visually shows the settings that we have chosen in the BIOS. Items shown in this application include CPU speed and bus settings, motherboard manufacturer, BIOS revisions, memory timings, and SPD chip information.
Task Manager: We use this utility to show physical memory, kernel memory, page file, and processor usage.
- Processor: Intel Fourth Generation Core i7 4770K @ 4200MHz
- Memory: Corsair Vengeance Pro 11-13-13-35 2828MHz
Overclocking on Intel Haswell processors is quite a bit different from what we are used to in the last few generations, where the CPU clock speeds did not influence the clock speed the memory controller could handle. Entry to Haswell memory overclocking 101 shows that while the CPU may handle a nice overclock of up 4.7GHz to 5.0GHz on really impressive examples, the memory may not scale up past 1866MHz or 2133MHz at those CPU clock speeds, even though the memory modules are rated much higher. What I found on both of my CPUs was that anything greater than around 4250MHz on the CPU would not run the memory at 2600MHz or higher. That being said, keeping the CPU as close to 4200MHz was imperative if I wanted to push these modules from Corsair higher than the rated 2666MHz.
With the rated speed that high I was expecting to pull off a 300MHz screen shot but was not able to do so. The highest I was able to reach was just over 2900MHz using the default XMP profile with some modified settings and a boost in voltage to 1.75v. To hit that range I was forced to use the 167MHz strap and lower the bclk to 161MHz. At this speed the memory sub timings were bad enough that performance dropped off markedly, negating any real performance gains. So back to the drawing board to see how high I could get while still retaining the primary timings at the XMP defaults.
Letting the board retain the memory subtimings set by the Maximus VI Extreme at 4.2GHz on the core and 2800Mhz on the memory, the performance was still sub par when the board would train the memory timings so it was back to the drawing board to manually tweak the settings enough to improve performance overall. In the end I was able to run 11-13-13-35 using 1.75v on the DIMMs and leaving the System Agent, Digital IO, and Analog IO to the board to control. Ultimately 2828MHz using the 2800MHz divider and a 101MHz bclk gave me a solid 162MHz boost in clock speed that delivered measurable gains in memory performance.
Maximum Memory Speed:
The maximum memory speed for each set of overclocked modules is indicative of how well the modules ran on this test system. As such, your results may differ in either a positive or negative way based on the capabilities of your hardware. In other words, your mileage may vary!
The benchmarks used in this review include the following:
- PCMark 8
- Geekbench 2.1
- Hyper Pi 0.99
- SiSoft Sandra 2013
- Metro: Last Light