Shouldn't RAM of the exact make and model all work the same?

[MoeB74]

Please let me know if it's just me or if everyone feels the same way. Apparently I'm being told by people who work for G.Skill that their memory modules may work differently even if you buy the same exact make and model. Does that honestly make sense to anyone else? I mean there's who R&D departments that are specifically in place to engineer plans on how to make products perform a certain way. IMHO, there's no way that if a RAM is suppose to support a certain speed, that there should be any performance difference if I buy 1 set or 2 sets.

I was actually told that the 2x8gb ram is only tested at speeds using the 1 set and not if you had 2 sets. Does that actually track with anyone? Do you mean to tell me that a big company or is trying to make a profit wouldn't plan on someone buying 2 sets? Especially when there's specifically ads to sell them like that around online. Please someone let me know if this is all RAM companies or just G.SKill. That way I can know never to buy

[emissary42]

Without dropping any names for obvious reasons, I can tell that it is not only the case with G.SKILL. You would get similar answers from other manufacturers and find cases for that in their supports forums. I do understand it is not very obvious, but as someone with a bit of a technical background, that also is to be expected. So let me try to explain it:

The thing is no two chips even from the same part of a single wafer are exactly the same. So what happens in production is a chain of selection processes that determines IC quality and matches similar chips for the use on modules of certain specs. Further down in the chain modules are matched to kits. However naturally there still are small differences, even between two modules of the same matched kit and from one kit to another. That is commonly known as sample variation. They are however tested to work together at their specs and that is what the manufacturer warranty covers. However it is literally impossible for memory manufacturers to verify inter-kit operability because of the sheer number of tests that would be required even among a single batch of kits (hundreds of thousands).

Also for many specifications even the type of IC used on them can change with their availability/out of necessity, not only over time. Very recently there even was a transition from one PCB to another one for a lot of Non-RGB models to improve compatibility with newer motherboards, because motherboard manufacturers nowadays tend to optimize their DRAM training for short trace PCBs. Both of those cases also can have implications on compatibility between different kits.

To add to that, there are other obstacles, that do not originate from the memory itself. For example even the compatibility between a two module kit and the respective four modules kit of the same specification can be different, because the motherboard or CPU memory controller have lower frequency limits with four slots populated. So even two perfectly matched kits that each on their own work flawlessly at their specs might not be able to do that when used together, because of that.

tl;dr Do not mix kits and buy a single one of the chosen overall capacity, like recommended in the G.SKILL memory FAQ. Also make sure to check the QVL for motherboard compatibility.