Too many times have we seen the effects of cheap memory. Although inviting at first, it often results in stability issues, if not now but in the future. Some memory is a made of a mish-mash of different brands of chip. They often skimp on costs by having an ultra thin layer of gold for the connections, meaning that after as little as 2 installations the connections can wear away causing the memory to fail. Most cheap memory is high latency, with very little flexibility for overclocking.

In a recent memory labs in Custom PC magazine, they came to the conclusion that low latency memory was expensive and offered very little extra performance for the money. In doing our own benchmarks for another publication, results were similar, with SysMark only performing up to 2.5% faster. But that doesn't mean low latency memory has no reason for existence.

People who want to buy low latency memory are generally the same people who buy high end graphics cards. They don't offer particularly good value for money, but they do offer the very best performance available at that point in time. To cater for this particular niche market, we decided to use games as our benchmarking focus.

We have used 4 different types of memory, in different configurations. Hopefully, we can highlight the performance difference between low and high latency memory and the variants in between.

The test system was:

- MSI K8T Neo
- Athlon 64 3200+ (Socket 754)
- Gainward Geforce 5950 Ultra

Test were performed using the following benchmarks:

- 3D Mark 03 (Build 340) @ 800x600
- Far Cry, Maximum Settings @ 640x480 using custom made demo
- Call of Duty @ 800x600 using custom made demo

The settings have been deliberatly kept low, to prevent the graphics card or CPU causing a bottleneck.

<< Back | 1 | | 2 | | 3 | | 4 | Next >>