Memory Performance Results

As mentioned, I am using a fairly decent CPU overclock for all testing because greater CPU overlocks allow greater memory performance increases to be utilized. I have selected many different benchmarks that show these differences, but not all workloads are going to see the same gains as these hand-picked benchmarks show. To show the increases, I started with two different memory kits, one from Samsung, and one from Patriot. The Samsung kit is a 1600 MHz kit that features normal JEDEC timings for that speed, rated at 11-11-11-28-1T. The second kit is a much older high-performance kit that was intended to be used with P55 chipsets, but missed out on full support with a SandBridge CPU since those CPUs didn't natively support Patriot kit's 2000 MHz XMP speed with 9-11-9-27-3T timings. IvyBridge differs from SandyBridge in many ways, but one of the most important changes to gamers and overclockers is the addition of many more memory dividers, such as 2000 MHz, which makes this old kit useful. I have spent several months testing these kits with various boards and both have worked great. I also test using two Gigabyte Windforce HD7950 3 GB video cards that run in Crossfire mode at their default speeds of 900 MHz for the core, and 1250 MHz for the memory. This helps eliminate any sort of GPU bottleneck that might have been introduced while, at the same time, showing to be sensitive to memory performance changes. Whether this is due to extra CPU or memory load is not known, or relevant. Over the next several weeks I'll be covering many different memory kits. While there are only a few results now, you can expect that to change soon! Let's take a look at what performance increases the G.SKILL kit offers:

AIDA64 Performance

The G.SKILL TridentX 2400 MHz kit performed well in AIDA64, showing a decent increase from both JEDEC and 2000 MHz testing. The gains here are almost linear with the clock changes, 400 MHz for each step with the jump up to the 2000 MHz DIMMs providing the largest read increase.


Copy, on the other hand, drops off after 2000 MHz with just a small increase. Latency also doesn't scale quite linearly, but the gains are still there.

SiSoft Sandra Performance

SiSoft Sandra provides an interesting perspective over AIDA64, showing that different workloads will experience different results. Even so, the G.SKILL kit still increases performance quite considerably with IvyBridge, perhaps indicating why Intel opened it up so much in comparison to Sandybridge.

SuperPi

SuperPi has always been quite sensitive to changes in memory performance, and we still see the same is true with IvyBrdige, although there are only mere seconds between each result. Still, there is no denying that the G.SKILL kit gives the better results.

WinRAR

Winrar is also fairly sensitive to memory performance changes, due to the type of workload provided by compressing and decompressing data. The benchmark provided with the 64-bit version of WinRAR showed a pretty large increase from 1600 to 2000, but when going from 2000 MHz to 2400 MHz, the increase was only half of the previous gains.

wPrime

wPRime is much more focused on straight CPU performance, yet memory plays its role as well. Here the gains provided by the G.SKILL kit are very small.