Overclocking

Timing and Frequency Scaling

I also test modules for voltage requirements at different CAS settings and speeds to show how well a manufacturer has binned their modules and how well they've optimized them for the best possible performance. Would tighter timings and a lower speed make this module perform better? Can we increase or decrease voltage to get more performance? When overclocking, there are many different considerations that must be taken and this test helps with a few of those choices. For this test, I start with the default timings and a voltage of 1.50 V before figuring out the maximum speed I can achieve. I then go through the procedure with each working CAS Latency setting and each voltage interval. Different sticks are rated for different voltages, so the range I use depends on the stock voltage and how high each voltage takes me. With this set, I used a range of 1.5-1.75 V.


Kingston, like some other companies, programs their SPD profiles with a lot of information, which makes testing for scaling a bit redundant. First, there are several different JEDEC profiles, as you will find on pretty much any stick out there. You can view this information easily by running AIDA64. I found the profiles listed above at the top of the listing.


Then there is the 2400 MHz XMP profile labeled as the "Enthusiast" profile in AIDA64. Not only does it contain settings for 2400 MHz, but it also has settings for lower dividers, which will have timings adjust automatically on some motherboards as you change the divider.


The third set of timings, the "Extreme" profile, starts with 2133 MHz as the main profile and has many more below that.

Both XMP profiles also have configurations for higher CAS settings, to enable those options should your board support it. I could also boot them easily at CAS 6 since there are CAS 6 settings in the XMP profile, although the actual scaling was a bit small, but again, that is common for most DDR3 memory on the market today. As you can see above, quite a few possible timings are already set for you to explore, and each stick should be able to run these timings at the above posted speeds without much issue. What you can get above and beyond that is going to change from set to set.


I tried to keep timings as close to the 2400 MHz XMP listings as possible with my set, since those offered the best performance. I loosened the timings as required once I went up and ultimately landed in the same spot as on my test for maximum frequency. Without even being able to run 2600 MHz, there's not a lot of overhead, but there is a lot of flexibility that is matched by those tight tertiary timings for pretty decent performance, no matter how you want to run your system. I managed to get a wee bit over 2600 MHz out of these sticks, which really is not bad at all.

Maximum Overclock

When testing for maximum overclocks, I reached 2600 MHz with a boosted voltage or loosened timings and managed to eke out just a bit more by adjusting the baseclock. I retained the timings used for "stock" without any problems, but pushing any further proved to be nigh on impossible. It is normal for Samsung-based DIMMs not to scale in frequency since they rather prefer tight timings, which allowed me to run that 2615 MHz with tight tertiary timings for an added performance boost.


You'll find a bunch of benchmarks that illustrate the overclocked performance boost on the following pages. They are, again, broken into System Performance and 3D Performance sections. The section includes the XMP results of the HyperX Predator Anniversary 2400 MHz kit at the maximum overclock posted above, the previously tested G.SKill TridentX 2666 MHz kit with its overclocked numbers, and a reference JEDEC 1600 MHz set of numbers. I will as always let the numbers do the talking.