Memory Guide

The Great DDR3 1600MHz Memory Showdown

The Great DDR3 1600MHz Memory Showdown



Test Setup

Test Setup

Since we had a mix of dual and triple-channel kits, we have two separate platforms to test them on. Since the latest AMD systems are still riding on a dual-channel memory controller design, we test the dual-channel kits on AMD's latest Phenom II processor. The triple-channel memory kits obviously had to make use of Intel's Core i7 triple channel memory controller. The dual-channel kits have a slight edge in terms of total memory (2 x 2GB) compared to the triple-channel ones (3 x 1 GB). We figured these memory capacities are the more common configurations for most needs these days.

As we had highlighted in the previous few pages, the dual-channel kits came in a range of memory timings and voltages. This is not the case for the triple-channel versions, which have uniform specifications for the most part. All memory modules tested here are rated for 1600MHz.

We have also conducted our benchmarks and other tests at the modules' specified timings and voltages. As a point of reference, we have included the results of DDR3-1333 memory modules for both the dual and triple-channel kits. Since the DDR3-1333 memory modules usually come with faster memory timings, especially in the case of the triple-channel kit, we have also 'simulated' the results for the reference memory modules to match the timings of our triple channel memory modules by adjusting the timings manually. This will give you an idea how much better 1600MHz memory are in certain tests with regards to the increased bandwidth and latency. These reference memory modules are:

  • 3 x 1GB Kingston HyperX DDR3-1333 @ 7-7-7-20 (Triple-channel kit)
  • Simulated 3 x 1GB Kingston HyperX DDR3-1333 @ 8-8-8-24 (Triple-channel kit)
  • 2 x 2GB Corsair XMS3 DHX DDR3-1333 @ 9-9-9-24 (Dual-channel kit)

For the overclocking segment, we had to increase the voltages accordingly and the timings too depending on the frequencies. Finally, for the temperature testing, we looped the memory modules through 3DMark06 before measuring the external temperature of the heat spreader using a laser thermometer.

For the AMD system, we had the benchmark system had the following configuration for the other components besides memory:

  • AMD Phenom II X4 955 (3.2GHz)
  • MSI 790FX-GD70
  • Seagate 7200.10 SATA HDD 200GB (single partition)
  • Zotac GeForce 9800 GTX 512MB (ForceWare 186.18)
  • Windows Vista Ultimate with SP2

For the Intel system:

  • Intel Core i7-965 (3.2GHz)
  • Gigabyte EX58-EXTREME (Intel INF 9.1.0.1012 and Intel Matrix Storage Manager 8.8.0.1009)
  • Seagate 7200.10 SATA HDD 200GB (single partition)
  • Zotac GeForce 9800 GTX 512MB (ForceWare 186.18)
  • Windows Vista Ultimate with SP2

Benchmarks

The benchmarks used consisted of both synthetic tests and two modern 3D games:

  • PCMark Vantage ---- PCMarks and Memories Suites
  • SPECviewperf 10 --- proe-04 and sw-01 viewset tests
  • SiSoft Sandra 2009 SP3c --- Memory Bandwidth and Latency tests
  • Crysis 1.21 --- CPU Test 2, 1024 x 768, Medium settings for all except Physics at High
  • Far Cry 2 --- 1280 x 1024, Ranch Small Map, No AA, DX10, High Settings for Render and Performance