Test Setup
CPU : Intel I5 2500K Retail
CPU Cooler: Noctua U12P SE2
Motherboard : ASRock P67 Extreme6
RAM : G.Skill Ripjaws F3-12800CL8D-4GBRM
Video : Sparkle X460 Calibre 1GB
Power Supply : Cooler Master 850W
HDD : Seagate Barracuda 320GB 7200.10
Case: Cooler Master ATCS 840
During the tests, the CPU speed was set at 3.3GHz and Turbo was disabled in UEFI. Testing with the same stock timings was possible until 1866MHz was reached. To make the system stable at this frequency, it was needed to relax the timings a little and the voltage was set first at 1.65V; the next step was to tighten the timings in small steps and for any modification I have ran HCI Memtest (Windows utility) with 4 instances for about a half an hour, to determine if the system was stable or not. When the most stable timings were found, I have started to decrease the voltage until the memory was giving errors again and raised it with one step in UEFI, to obtain the best timings at this frequency, at the smallest voltage possible, this time 1.515V.
I have tried multiple times to overclock the modules at 2133MHz, by setting the voltage to 1.65V or even at 1.7V, relaxing the timings completely and upping the VTT voltage with no success (the system was booting to Windows, but HCI Memtest reported errors after a few minutes).
The CPU-Z tab where the memory timings and frequency are confirmed (at maximum stable overclocked speed):
CPU-Z online validation at 1866MHz 9-10-9-24:
Test Results
Synthetic Benchmarks
3DMark 2001
3DMark 2006
PCMark Vantage
SuperPI XS 1.5 2MB
wPrime 32M
Memory Bandwidth, Latency
AIDA64
CrystalMark Bandwidth
Cinebench R10 Rendering
Cinebench R11.5 Rendering
x264 Encoding
Games
Crysis
Devil May Cry 4
Far Cry 2
Street Fighter 4
Bandwidth and Latency comparison vs previously tested kits at rated speed
Bandwidth and Latency comparison vs previously tested kits at maximum stable overclock