Scenario. A duel channel MoBo with 4 slots. A DDR3 board where 1333 is highest without any overclocking. So you want to eventually fill all 4 slots. But the two, 2DIMM kits(shortened to just kit/s from now on) you put in must have some level of similarity.
What different memory tech guys from different RAM manufacturers tell me with total certainty and explain in length.
They say not only do the two kits have to be identical in every conceivable way, they still might not work together at all and if they do, I would have to clock down a rank from their rated speed number, so that if I were to buy two kits 1333 ram, I might need to clock them down to 1066 to get them to work. But if I really want to make sure everything works together, I should buy a 4 piece kit tested to work together, with a premium of over 50% more in price to match.(in some cases close to double price)
I was also told that if I buy a kit now, then try to buy one like 4 months down the road, the design of their memory will have changed so much as to be incompatible.
I talked to someone in a chat-room that sounds like he knows what hes talking about, who says that really all I need to match are the CL and speeds, I can conceivably even have different sizes, brands, densities, of kits, and still have them likely working together as long as the speed and CL match. Just that the more variation, in those other things, there is a small increase in the chance of them not working together.
I was told on a tech forum something in-between these two positions. That everything should be the same, sizes, densities and brands too, but it really wasn't necessary to get a premium priced 4 piece kit.
So where lies the truth here? Please give me all the source material for what you tell me that you can so that I might distinguish your claims as the real truth among these other claims of truth I have heard.
There is no absolute when mixing memory modules; every board bios is just a little different. You'll just have to try it. I got my old mac pro 1098 ram to run at 1600 mixed with my new ripjaws 1600. I wasn't expecting it to work, but surprise, it's running stable so far. No voltage increase was necessary. I have 2 2 gb ripjaws mixed with 2 mac pro 1098 sticks and the i3 overclocked at 160 fsb. One reason it worked is that the timing numbers go up at 1600. The old mac pro ran at 7-7-7-20 at 1098; now running at 9-9-9-24 at 1600.
I was already semi-aware of the nature of latency, though I am pretty sure its more then that, it's that lower latency memory is able to wait a shorter amount of time as well, else there would be no value difference between them.
Malmental, You went at length regarding stuff I didn't really ask about, (I already knew intel has a CPU integrated memory controller) and were too brief and uninformative regarding what I really want to know.
Asking about something I didn't before but you touched on, could I get 1600 out of a 4 set? Could I do so with two of the same 2 kits rated at 1600?
Now to the main subject, what counts as mismatched memory? (insert the detail went into in my first post when I asked this question) Then, how much of you belief is theoretical and what is your source?