Kingston Predator 16GB (2x8GB) 1866 MHz DDR3 RAM/Memory Kit Review (KHX18C9T2K2/16X) - Test Setup & Overclocking

Test Setup

Test System:

CPU: Intel i5-3570K

CPU Cooling:  Cooler Master TPC 812

Motherboard: Gigabyte Z77X-UD3H

GPU: Nvidia GeForce GTX 660

OS HDD: OCZ Agility 3 120GB SSD

Secondary HDD: Kingston HyperX 3K 240GB SSD

OS: Windows 7 Professional 64-Bit

 

 

Overclocking

We took several approaches to overclocking the Kingston HyperX Predator kit.  Since overclocking can be accmoplished by lowering timings and/or clock speed, it can be a tricky business to find the perfect clock so there's no simple formula.  Think of it as a race, where one racer is twice as fast as the other.  The slower one gets to go first (lower latency, less wait time) but runs less fast.  The faster runner (higher clock speed), has to wait a little bit longer to start running (higher latency) but catches up and passes with greater speed (higher clock).  Now lets, for simplicity's sake, assume they arrive at the hypothetical finish line at the same time, both racers have the same relative speed.  So with this analogy we can see that overall RAM speed isn't remarkably simple, and is a "tango" between clock speed and latency.  Generally speaking, the faster runner will finish first, but often to increase that speed, he must take longer "rests," which somewhat offsets his gains.  This is what we see by the diminishing return in memory overclocking by increasing the clock speed.

 

First, we wanted to see how low the timings could be set at the Profile 1 XMP setting, which runs at 1866 MHz and 9-11-9-27 timings .  I was a little surprised to see little flexibility for lowering the timings lower than the rated spec without failed POSTs and BSOD immediately upon boot, even as high as 1.65-1.70V.  The rated timings (9-11-9-27) seemed to be about as low as they would go at the XMP 1866 MHz setting.  Next, we tried for the best overall OC combination of timings and increase in clock speed.  The kit seemed a bit more sensitive to timing reduction, which I wasn't surprised about from out failed attempts to lower the Profile 1 XMP timings.  We were able to achieve the next step up at 2133 Mhz and had to drop the timings to CL11, at 11-11-11-32.  This is pretty good for breaching the 2000 MHz mark and making it stably at 1.65V all to way to 2133 Mhz, so we're very happy with that result.

 

Originally I planned to compare these benchmark results against those from Patriot's Intel Extreme Memory kit, but the numbers were virtually identical. Since there was no statistically significant difference, I decided to stick with just analyzing how these modules performed under different clock conditions and it's maximum clock speed.

 

You have no rights to post comments

Comments   

 
# Sabishii Hito 2012-10-26 08:21
These sticks are based on Hynix CFR 2Gb ICs and should have way more headroom, at least DDR3-2400 11-13-12.
 
 
# Evan 2012-10-26 08:28
Quoting Sabishii Hito:
These sticks are based on Hynix CFR 2Gb ICs and should have way more headroom, at least DDR3-2400 11-13-12.


Even the best chips down have some "down-binned" parts come off the line for the lower frequency kits. But appreciate the feedback, and I'll give it another shot for you (we were unsuccessful through several hours of trial-and-error through seemingly endless timing combinations at 2400 MHz). It is very possible we have one unlucky DIMM holding the kit back. Appreciate it and we will report back if we get something to stick this time! These also deserve a discussion of the Hynix chips, so thanks for that as well.
 
 
# Sabishii Hito 2012-10-26 11:04
Hmm, those buggers might need some more volts slapped in 'em! Although Hynix doesn't seem to scale much over 1.75v
 
 
# Sabishii Hito 2012-10-26 12:21
Yeah these ICs, when going for clocks over 2000 like the timings to be tRCD=tRP+1=CAS+ 2
 
 
# Evan 2012-10-26 12:28
Quoting Sabishii Hito:
Yeah these ICs, when going for clocks over 2000 like the timings to be tRCD=tRP+1=CAS+2


Great feedback and a good rule of thumb! Our mentality for the max OC was a "safe" 24/7 OC for a more typical enthusiast who likes to push without much added risk of damage, so we limited the voltage to 1.7V, I'll try a bit higher and see where it takes us!

Have a great day/weekend!
 
 
# Sabishii Hito 2012-10-27 08:14
My mistake, these actually use Hynix MFR 4Gb ICs, but they like the same kind of timings as their smaller-density siblings the 2Gb CFR chips.
 
 
# gene396302 2012-10-28 08:27
Can you put this on a laptop, it looks like a laptop memory.
 
 
# Evan 2012-10-28 17:13
Quoting gene396302:
Can you put this on a laptop, it looks like a laptop memory.


These are full-sized DIMMs, so they will not fit in a laptop. They appear more square because of the large-profile heatsinks.