it was DDR4 3600MHz for Intel and DDR4 3200MHz for AMD. I never understood the reasoning behind that.
As I understood it, the thought process was that those speeds were considered "somewhat easily attainable" XMP memory speed choices for each platform. I don't know how close to reality those speeds actually were for the enthusiast segment of the market, however (ie. the most common memory speed purchased for each platform by enthusiasts).
As you know, AMD (Zen and Zen+) had a more difficult time at higher than 3200 MHz speeds, while Intel was somewhat easy to run at 3600 MHz.
Beyond those speeds, for each platform, the gains were negligible.
Therefore, running those speeds would more accurately reflect the performance an enthusiast would see on their own build, assuming they used similar parts.
I guess the idea was that if Intel had an easier time with higher speed memory, let it use the higher speed memory. It's a platform advantage, so you might as well use it. Otherwise people might say you are artificially hobbling the platform to make the other platform look better (even if it was not your intent).
I can still see why keeping Intel at 3600 MHz makes sense, but only as long as ALL Intel builds used for reviews of motherboards are using 3600 MHz.
When comparing motherboards within a platform, I think it is best to keep everything the same as much as possible.
Highlight the differences for the item under test with as few variables as possible.
A different issue is when you want to start comparing performance between platforms (Intel vs AMD). I do not think choosing 3200 MHz for Intel would be a good choice in this case. So you would not be able to take the results from the MB testing and use them for the platform comparison testing in that case because 3200 MHz memory was used for all Intel tests. You'd need to install 3600 MHz memory to allow Intel to "stretch its legs" and give more accurate enthusiast performance results. If you didn't do that then you'd get nothing but negative feedback from readers saying you've hobbled the Intel platform with slow memory. With Zen2 it may turn out that 3733 MHz is the best memory speed to choose because it maintains the 1:1 ratio and gives the best performance.
I can see pros and cons for each way of doing it.
