A few notes on testing: At first installation, the modules would not function properly using slots 1 and 3. Burn-in at lower speed got close, but ultimately we had to change module positions to slots 2 and 4 before these would perform flawlessly. While unnerving for the novice builder, this sort of thing is a frequent enough occurrence with certain board-memory combinations. For fair comparison's sake, our TWIN2X-5400C4 modules also received similar burn-in, at which point these were able to achieve slightly better timings than previously reported.
Already stable in slots 2 and 4 at its rated timings and voltage, we raised voltage to 2.2v as per OCZ's documentation to allow added speed and further latency enhancement. Again, for fair comparison, the TWIN2X modules were matched to 2.2v. TWIN2X was perfectly capable of operating in slots 1 and 3, the default selection for most builds. Both pairs were then tested to find their maximum clock speed, followed by their minimum stable latency values at industry standard speeds, with the following results:
Best Stable Speed and Timings at 2.2v
Device
533
667
800
867
OCZ Technology PC2-6400EL XTC
3-3-2-6 1T
3-4-3-8 2T
4-4-4-12 2T
5-5-5-15 2T
Corsair TWIN2X-5400C4
3-3-2-6 1T
3-4-3-8 1T
4-4-4-12 2T
5-5-5-15 2T
Amazingly, both products achieved the same highest speed and latency values when set to the same voltage! The only notable difference was Corsair's ability to use 1T Command Rate at 667 data rate, which we attribute to its use of slots 1 and 3 rather than 2 and 4.
Using these settings and dual-channel mode (a 128-bit pathway), the Pentium 4 530's 64-bit 800FSB became the system bottleneck. Sandra offered the following bandwidth benchmark results:
We look forward to testing future AMD DDR2 platforms, with an integrated memory controller and internal HT bus freeing us of FSB bandwidth limitations. Now on to DirectX and Open GL testing!
Events
IE 7
Firefox 2.0
