Testing Method & Test Setup

To be objective in the overclock testing, we use the most popular enthusiast platform, which currently is Socket 1155. Obviously, our choice for CPU fell for Ivy Bridge as it provides a much wider memory testing range.

Motherboard	ASUS Maximus V Gene (BIOS 0086)
CPU	Intel Core i7-3770K
Graphic card	XFX 8600 GT
Memory	Kingston HyperX T1 KHX24C11T1K2/8X
HDD	Samsung 40 GB
PSU	Silverstone OP1000
OS	Windows 7, 64 bit SP1

As usually, we call things stable only if we don’t detect any errors using eight 750MB instances of HCI Memtest in the first 150% of progress.


An important note needs to be made before we proceed. You see, SPD is there to set the subsettings so that users have less parameters to worry about, but on Kingston’s recent kits it fails to get that job done. For example, letting the platform set subtimings automatically results in a no-POST situation already at 900Mz. Therefore, if you are planning to use the HyperX outside the prescribed profiles, we recommend you to base your subtiming values on those we found optimal on the screenshot above.

Results

As usual with Hynix memory, second and third primary timings, tRCD and tRP apply certain frequency limitations irrespective of other settings. So, given that frequency is a function of voltage using a constant CAS value, the optimum primary timing formula depends mainly on the voltage range you’re looking to run. In the sub 1.55V region, it’s possible to run timings almost even (CL+1 = tRCD = tRP+1) and in the 1.55-1.75V range you’d need to raise tRCD and tRP by one value ending with CL+2 = tRCD = tRP+1.
Looking at the picture in general, it seems we weren’t lucky with our sample as it failed to achieve its rated speeds coming about 10MHz short using either manual settings or the XMP. This was unexpected as most Hynix CFR based memory should be able to clear Kingston’s spec without issues so we won’t state that every sample of DDR3-2400 HyperX will be as bad as ours. However, this serves as an alarming indication that Kingston are not yet fluent with production of high frequency DDR3 memory.

Page 1 - Introduction	Page 4 - Results
Page 2 - Closer Look	Page 5 - Conclusion
Page 3 - Photo Gallery

Discuss this article in the forums