All kits we test are put through the same battery of benchmarks, to ensure that all future tests on the same platform have accurate and quality results for comparison purposes. We’ve recently added SuperPi to our list of benchmarks, as we felt it better represented memory performance than the communications suite in PCMark Vantage, which relied more heavily on GPU processing than memory usage. We will be testing the TridentX memory against the recently reviewed Samsung MV-3V4G3D DDR3-1600 8GB Memory Kit.
The first test we utilize is AIDA64’s Cache and Memory benchmark. This benchmark tests the read, write, and copy speeds of the memory, as well as reading the latency (in ns).
The next hurdle the TridentX kit has to overcome is three tests from SiSoft Sandra. The first measures similar things as AIDA64’s test prior to this, but tests in a different manner, therefore providing different results.
Following on from bandwidth testing, we test the memory latency, which Sandra (very helpfully) gives you a readout of the memory’s time-to-react in ns. It also gives the memory a “speed factor” representative of its performance in testing.
Finally a cache performance test.
With Sandra’s work all said done, we move on to MaxxMem² – PreView. A simple tool and relative newcomer to the testing scene, MaxxMem² is a very closely related test to AIDA64’s Cache and Memory benchmark. Again, however, it tests in a different manner and therefore its results are invaluable.
Finally, the latest edition to ThinkComputers’ benchmark lineup, SuperPi calculates Pi and uses your CPU and Memory to do it as fast as possible. The faster the better, so the lowest score is the one that represents higher performance.
Note: Due to the fact we used BCLK to reach our 2496Mhz overclock, this also overclocked the CPU. This would have contributed substantially to the faster result on SuperPi, and therefore, the speed increase should not be attributed to only the memory itself.
Jan 29, 2015 0