|Patriot Viper-II DDR3 Lynnfield Memory Kit|
|Reviews - Featured Reviews: Memory|
|Written by Mathew Williams - Edited by Olin Coles|
|Thursday, 26 November 2009|
Page 4 of 5
Performance Test Results
Today's review features four different benchmark applications: Passmark Performance Test, Lavalys EVEREST, Sisoftware Sandra, and Crysis. The first three are synthetic benchmark suites designed to assess memory performance. Each one has a unique approach to memory testing, which should provide a good assortment from which to determine performance trends. The last application, Crysis, offers some insight into how memory performance affects gaming on our Lynnfield system. It's important to keep in mind, though, that CPU speed may also be a factor in our test results. While we did our best to eliminate it as a variable, it was impossible to keep the CPU frequency the same while still testing the memory kit at the necessary settings.
In Passmark PerformanceTest, the first two memory speeds we tested, 1066MHz and 1333MHz, don't show much variation. It's likely that the difference in timings offset any gain in frequency. Moving on to the Viper II Sector 5's default frequency of 2000MHz, we see a slight increase in performance in all three benchmarks. As expected, overclocking the memory results in a further increase.
Lavalys EVEREST paints a slightly different picture than Passmark. While the higher frequencies take the lead in the Read and Copy benchmarks, they drop slightly behind in the Write benchmark. This kind of result can usually be attributed to a blottle-necked memory controller. In this case, though, there may have also been a compatibility issue between Windows 7, EVEREST, and our hardware.
The results of Sisoftware Sandra are much more consistent than EVEREST. They appear to scale directly with the memory frequency of our Sector 5 DDR3 kit. That's not surprising as these two benchmarks are intended to measure memory bandwidth. As with all synthetic benchmarks, though, it's important to keep in mind that performance will vary in real world applications.
Our real world gaming benchmark for this review is Crysis. Three resolutions and quality settings were chosen to demonstrate the effect of memory speed as the demand on the video card increases. The results are fairly clear. At the lowest resolution, the higher memory speeds can increase framerates by nearly 15%. At higher resolutions, though, the difference is negligible. It's likely that we've reached the limit of our HD 4870 and no increase in memory bandwidth can help.