Result Summary and Power Consumption
All in all, there were some impressive results. Some kits did better than others at 5-5-5 timings, others were able to clock quite high at tighter 4-4-4 timings.
At 5-5-5, the Ballistix kit came out way ahead at a blistering 1300MHz. The Reaper HPC kit also did well, clocking up to an impressive 1220MHz. The Reaper kit only needed 2.3V to acheive this frequency. The Ballistix kit needed all the vDIMM the Asus P5K test board could deliver to acheive 1300MHz. It was able to clock to 1200MHz at a much more reasonable 2.2V. The mainstream XMS2 kit was actually a pleasant surprise reaching 1180MHz at only 2.2V. The Hyper-X and Dominator kits didn’t have a lot of headroom beyond their default 1067MHz at 1120 and 1100MHz respectively.
Things look a little different at tigter 4-4-4 timings, with the Hyper-X kit doing especially well at 1020MHz. The Ballistix kit was still off the wall at the tighter 4-4-4 timings, hitting an amazing 1180MHz. The Dominator kit also did very well at 4-4-4 reaching an impressive 975MHz. The XMS2 and Reaper HPC kits that did so well beyond 1067MHz and 5-5-5 timings didn’t scale as well at 4-4-4.
At very tight 3-3-3 timings, we still see the Ballistix kit at the top, hitting 900MHz. A similar trend continues with the
Power Consumption Tests
I decided to run some power consumption tests on the various kits I had in the lab. For the first test, I kept the vDIMM constant at 2.0V and the CPU at 2700MHz. The Memory frequency was set to 750MHz. Each kit was left looping test number five in Memtest86+ for about 5 minutes before a reading was taken using an AC power meter.
The reading while running Memtest86+ was so steady that the results are totally reproducable. There is indeed slight differences in power consumption between each kit. Interestingly, the two best performing kits were the least efficient, drawing 2-4W more than the others. The single sided Corsair XMS2 kit was the most efficient at a full 5W less than the Ballistix kit. This makes sense as it seems to employ higher density ICs.
For the second test, I run the Crucial Ballistix at varying frequencies and vDIMM, while keeping CPU/FSB frequency constant. As expected, vDIMM increases can significantly increase overall system power consumption. Although 20W may not seem like a lot, it is the equivilant of almost two compact flourescant bulbs, or the power consumption of a VIA C7 based CPU and mini-ITX board. Those who are looking to build super-efficient PCs should try to keep vDIMM at the JDEC standard of 1.8V.
Articles RSS