As the table above reflects, we'll be testing this memory on two different platforms. We know the DFI is capable of 4.0V VDIMM, but since the Abit IC7-G board only offers 2.8V max, I'll be utilizing another OCZ product, the DDR Booster , in order to push enough voltage to attain stable 2-2-2 latencies. And just to ensure there's no confusion, from here on out when I mention X-X-X-X latencies, I mean it as CAS, tRCD, tRP, tRAS.

All benchmarks were performed on a clean load of Windows XP (SP2) running DirectX 9.0c with all current critical hot fixes and drivers available as of March 30, 2005 . Each round of testing was run after performing a fresh reboot of the machine which was then left to idle for a minimum 5 minutes. Tests were run a total of three times and then averaged together to obtain the final, reported results.

Both the AMD and Intel tests were broken down into stock and maximum achievable speeds at both 2-3-3-8 and 2-2-2-8 latencies. 3.2V was definitely required on both systems to get the memory to boot properly at 200MHz with 2-2-2 latencies. It took a bit of tweaking BIOS settings, but eventually I was able to hit a maximum stable speed reached on the Intel of 239MHz @ 3.4 and 249MHz @ 3.4 on the Athlon 64. I believe the memory could have gone further on the Intel system, but no matter what amount of voltage applied (up to 3.8V), it just wouldn't stay stable, or in many cases boot into Windows. I believe this to be more of a compatibility issue with the IC7-G since there is a commonly known issue with VTT not tracking to ½ VDIMM like it is supposed to above 2.8V which can lead to stability issues. As far as the Athlon 64 system, I spent half a day, literally, tweaking around with the seemingly hundreds of obscure settings you can control with the DFI nf4 SLI-DR board and just couldn't get it stable above 249MHz. I started with the recommended settings from OCZ's forums and worked my way from there, trying the 325, 326, and 331 beta BIOS updates along the way, all to no avail. Much like the Intel system, I believe there is more to be had from this memory as there are a lot of folks hitting 250-260MHz with the same board.

Now as far as stability is concerned in these tests, it doesn't just mean the machine lasted long enough to complete the benchmarks. Stability in this matter means that the system remained stable for a minimum of 24 hours while running an instance of Folding@Home.

The CPU-Z screens below show the default layout and speeds that I set in BIOS on the initial powering up. The only thing I thought of as odd was that the SPD timings show as 3-3-3-8 instead of 2-3-3-8. More than anything, this is likely for compatibility, after all, you need to make sure everybody's machine will boot the first time with your hardware.