I think it is both simpler and more complex because there are different issues.Does the bios just check to see if the slots are filled or if Dual Channel is working?
The "Dual Channel" thing happens if both memory controllers have memory, period. Whether it is the same size module, number of chips or whatever, it is in dual channel mode for at least a certain amount of memory and there will be a small performance increase because of that, and yes, folks have demonstrated it. The demonstrations are not impressive because the performance increase for dual channel is small, and the benchmarking we are talking about is with something like Sandra, which is about the best we can do for accuracy. The figure I usually see in the forums works out to about 2-10% increase in a bandwidth test, with two equal modules closer to the 10%, and closer to the 2% for 3 dimms of the same size.
At the same time, adding more modules, such as 3 vrs 2 and more memory chips obviously has an impact on how high a FSB you can reach, and sometimes how low you can set the timings just as it always has. This is really noticeable and is the subject of many complaint threads when that magic 200FSB all of a sudden turns into a 190, or even a 185 when a large third module is added, or if two 512's are exchanged for two 256's. There are usually 2 or 3 threads active on this subject alone in Asus at any particular time.
So balancing the memory betwen controllers is not necessary for dual channel performance increases, however, as the quote from _Durandal_ says, it seems optimal if it is.
JMO
YMMV
Ever so cool that there is a memory forum, I can move nearly all threads now.![]()

Reply With Quote

. *Hint* it's the sixth post down
.


