OCZ DDR2 PC2-6400 CL3 FlexXLC Edition review - Testing explained
Author: Grga Curkovic
Date: 18 May 2007

We ran the modules on our Black Pearl testbed that is completely water cooled. The testbed was assembled of the following components:
- EVGA NFORCE 680i SLI Black Pearl
- Intel Core2 Duo E6700
- OCZ GeForce 8800GTX
- OCZ PowerXStream 1000W
- WD Raptor 150GB
- Gigabyte 3D Galaxy II water cooling set

During all tests the modules were in a separate water loop to ensure adequate cooling. We tried touching the modules several times during our tests and at no time did they feel warm to the touch so we?re guessing that the temperature was around 40 degrees C. We decided to find out how fast can these modules perform on different aggressive timings since they are obviously low latency modules. We didn?t want to relax the timings above 4-4-4 since we think that?s not the way these modules are meant to be used. Upon finding the top speed for each timings we performed 2 sets of benchmarks, one with the FSB as close to 266MHZ as possible and the cpu as close as possible to 2.67GHz, and one with FSB and memory working synchronous and the CPU thinking somewhere around 2.7-2.8GHz. We did this in order to show that much higher performance can be obtained by using low latency modules working synchronous with the FSB. Unfortunately that wasn?t possible in case of 4-4-4 since our board couldn?t handle FSB over 495 MHz. We first started with the default settings and ran a set of benchmarks, and then set the FSB to 400MHz and did another set of benchmarks. Though the memory was working at same settings we saw a huge difference in the results. After those initial tests we started looking for the top speed with those timings and found that these modules can go as high as 435MHz. Again we ran a batch of tests with default FSB clocks and with synchronous FSB. Then we changed the settings to 3-3-3 and ran a batch of test at 400MHz.

After that we started searching for the highest stable clock with those timings and found them to work up to 433MHz, just 2 MHz less than with 3-4-4. Big Tony of OCZ told us to try 4-3-3 so those were our next timings and we were surprised how well the modules performed with those timings. The top stable speed was an incredible 474 MHz, and that was doable with only 2.2V, not 2.4V what the EPP says. The last but not the least was the 4-4-4 test. The top stable speed was 1110MHz, but in this case we were unable to set the FSB to synchronous mode since our board can only clock up to 495 MHz. This is an amazing result, we have never seen any modules that were capable of such high speeds at 4-4-4.