The Tests
The setup is as follows:

• Intel Core 2 Duo 6700 with Arctic Freezer Pro 7
• EVGA nForce 680i motherboard
• Crucial Ballistix Tracer DDR2 PC2-8500
• EVGA GeForce 7800GT
• Antec NeoHE 550w PSU
• Western Digital 150GB Raptor Windows XP Professional SP2
• NVIDIA Forceware 97.92 WHQL

Per our norm it was our plan to test these sticks at three different settings, the first off which consists of stock timings of 3-3-3-10 and stock clock of 800MHz. From there we usually see how low we can drop the timings and maintain the stock 800MHz clock speed. After that we try to crank up the speed as high as we can without having to significantly loosen the timings.

That's how we usually do it, but we ran into some problems with this method right away. After the memory ran successfully at stock settings, we tried to lower the timings and the HyperX was having none of that. Not that it’s a huge issue, as the RAM already runs at an incredibly low CAS 3. We couldn’t budge the timings much at all, so we went on to the next step which would be overclocking the RAM at stock timings; but we got minimal returns there also. At best we were able to pull 820MHz but that was at 2.45 volts on the memory. It seems this RAM may be binned for the best timings, and that the clock speed is already at its maximum. This is a bit different than what we usually see, where chips are binned for maximum speed and overclocking potential with much less attention paid to timings.

As a result, we are testing at super-low stock latencies, and at 4-4-4-12 with a clock speed of 1000MHz (a 200MHz overclock). We will put this head-to-head with our numbers from the Crucial BallistiX Tracer 1066MHz which has been our top overclocker so far. This should give us a great example of high speed vs. low latency and which one is ultimately more important.

We will be testing with the following applications:

• PCMark 05 Memory Test
• Half-Life 2: Episode 1
• Counter-Strike: Source
• Call of Duty 2
• Elder Scrolls IV: Oblivion
• ScienceMark 2.0 Memory Bandwidth

Higher bandwidth from higher memory speed reigns supreme in PCmark05’s synthetic tests. We see the faster Crucial RAM speeds pull out ahead, but the HyperX performs very well.


Similar results are found with ScienceMark. The higher speeds give out better scores but as we all know, synthetic benchmarks are not the main reason we are here. The HyperX’s lower timings seem to close the gap a bit more here than they do with PCMark.