Both pairs will run at CL3 4-4-8 which is the SPD setting of our Corsair TwinX2048-4000PT. At 11x200FSB Corsair ran at 3-3-3-8 when settings were left on Auto (SPD). I adjusted the Mushkin kit to run at CL3 3-3-8 / 11x200FSB ensuring all settings were equal. Our test rig is water-cooled, our case will remain open exposed to the ambient temp of the room. The 4V jumper on the DFI motherboard will remain in the default position (Off). DFI Lanparty is truly an Overclockers platform providing a small Reset and Power switch on the board, thus allowing the end-user to eschew connecting the front panel. The DFI LP nF4 UT's VDIMM reaches 3.4V prior to enabling the 4V jumpers. Whether or not high voltages will have any effect other then an adverse one of the IC's spec'd for this product remains to be seen. The thumbnails below represent speeds/latencies tested for each memory.

Corsair 4000PT
11x200HTT 3-3-3-8 | 10x250HTT 3-4-4-8 | 9x265HTT 3-4-4-8


Mushkin (1024MB)
PC4400 11x200HTT | 3-3-3-8, 10x250HTT | 3-4-4-8, 9x265HTT 3-4-4-8



Bandwidth Benchmarks









The results are predictable for the Sandra Bandwidth benchmark as a 1024MB kit should out-perform a 2048MB kit in this particular application. The reason being the amount of "addressing" or work done in the memory controller given the application. As speeds increase, it seems Corsair begins to pull ahead and interestingly in Everest "Write" test Corsair fared well. Good news for Corsair!