Generating the charts
Now that we have gathered all necessary data on formulas, relative scaling and data range, we can start building the charts. I will not elaborate too much on this topic as it’s pretty straight-forward what we did, but I will elaborate a bit more on the final stage of the creation of the charts.
After all data had been extracted from the database and the memory rating had been calculated, we had a set of absolute performance indications. The issue is that these absolute values have no meaning other than ‘X is higher than Y’. To standardize the chart, we had to scale the absolute results to a scale of 0 to 100, with 100 being the best and 0 being the worst.
Initially, we’d planned to make all performance relative to the best performing memory kit. In this particular case, the GTX2 memory kit would receive a rating of 100 and all other kits would receive a rating relative to this 100. Now, of course this is not the right way to standardize. Who are we to say that the GTX2 should receive a rating of 100/100? So, instead of using the absolute performance rating of the best scoring memory kit, we used the absolute performance rating of ideal memory kit, within reason. So, all memory kits received a rating relative to the performance index of a hypothetical memory kit that would average DDR3-2000 CL6-6.
We appreciate all feedback on this article and the methodology. If you have questions, or want to discuss our unique approach, you can just drop a post on our forums and we’ll be happy to listen to it.
Thank you for reading!
This article was first published at Hwbot here
and has been reprinted with permission.