Home / Opinions / DDR2 Memory Round-Up / How We Tested

How We Tested

Something I mentioned during the recent DDR roundup and must mention once again is that no single memory benchmark I’ve discovered has the breadth or depth to give a completely accurate representation of any particular module’s attributes. Much like the programmes we run on our PCs, some benchmarks favour bandwidth while others favour latencies. A memory module that blitzes its way to the top of the pack in, for example, PCMark04, may actually trail by a couple of frames per second in your favourite game title. (If you'd like more detail about the bencharks used please refer to this previous group test.

In an attempt to counter this I ran seven individual benchmarks, each run a minimum of three times, and then I averaged the results to give an overall ‘rank’.

As I’m sure the more technically minded amongst you have realised, this isn’t quite precision maths, but with literally thousands of ways of generating memory performance and data error results, and probably as many ways again of collating them, there had to be some kind of simple metric to work by and this was as good as any.

Our primary interest in this particular roundup was how the various memory modules performed when run in dual-channel mode, and the reasoning for this is simple; the vast majority of the motherboards that will be sold this year will be based on dual channel-memory controllers. Although Corsair and Swissbit supplied single modules, I decided not, on this occasion, to see how the modules supplied in pairs performed when run as single sticks. While this would almost certainly favour memory not optimised for dual-channel operation, it wouldn’t necessarily prove as beneficial for memory that’s supplied as matched pairs. There would also be a capacity mismatch in running a 1024MB module up against a 512MB module.

As all the modules were rated at least PC4300, 266MHz speeds and, because this is available using either Intel’s 925X or 925XE chipsets, this was the speed at which default testing was carried out.

Torture testing was carried out on each module with a two-hour memory burn-in performed both at 266 MHz and at the maximum stable overclocked frequency just to confirm that operation at these speeds was likely to be trouble free. If a module failed it was first re-run at the same frequency before having the frequency lowered if it failed a second time. Sufficient time was given for module temperatures to return to normal before each re-run. Those of you who hold with the theory that memory needs time to “burn in” before it overclocks properly, whether it’s a myth or otherwise, will be pleased to know that the torture testing was done before the Overclocking, despite the small risk of completely destroying them in the process and the possibility I’d have nothing to test at the end of it all!

Overclocking was carried out under a low ambient room temperature of 14 degree centigrade and then again under a raised ambient temperature of 24 centigrade just to try and identify whether or not heat plays a factor in overall reliability. This may seem a little unfair but even at 24c we’re probably still below the kind of temperatures encountered inside most closed-case PCs.

Overclocking was also carried out at 2.1v in all cases with all modules were set manually to timings 4-4-4-12 to reduce the risk of them changing unexpectedly when run at non-standard frequencies.

The reason I list results as being ranked out of ten for modules at 266MHz and out of for eleven modules when overclocked is because I listed Corsair’s 5400UL modules twice, once for overclocked benchmark runs made with a setting of 4-4-4-12 and once again at the rated 3-2-2-8. None of the other modules were able to run at 3-2-2-8 timings, even at 266MHz.

And finally on the subject of timings, for the default 266MHz tests, all modules were run using the “Auto” setting as this best reflects the likely usage of less experienced users nervous at tinkering with memory timings. You may well be able to squeeze out better numbers by tightening the settings a little at these lower frequencies if you’re confident in doing so.

comments powered by Disqus