JackNaylorPE :
Yes I am quite familiar with the article and many more. In the THG tests where they saw 5% in gaming, the impact is negated by the fact that the GFX card is the usual bottleneck ... and, just like RAM Speed and CAS, when you look at minimum FPS or you look at SLI, these differences are greatly magnified. Tweaktown saw a 70% difference when only 2 sticks were intalled in quad channel boards.
A 70% difference at
what though? Following through to the reference and going to page 5 as directed by Wikipedia leads us to synthetic memory benchmark results in SiSoft Sandra. So sure, if all you do is run SiSoft Sandra benchmarks all day on your system for some reason, you should see higher numbers with quad-channel memory. Otherwise, you won't likely notice any difference. The Wikipedia page also pointed out that the 70% only applied to that one benchmark, and that the other benchmarks they performed didn't show any major difference between dual and quad channel modes. Here's a more relevant article, which tests more than just synthetic benchmarks...
https://www.pcworld.com/article/2982965/components/quad-channel-ram-vs-dual-channel-ram-the-shocking-truth-about-their-performance.html
Once again, PC World's SiSoft Sandra synthetic benchmark shows a huge difference, but that doesn't translate to much of a difference in actual software. At their Handbrake video encoding test, quad channel saw absolutely no performance advantage, and was technically around 1% slower, though that probably falls into the margin of error. Testing in PCMark also showed no performance advantage in simulated real world workloads. Winrar's built-in compression test showed quad channel being faster by just 2%, and only 7-zip showed a somewhat notable advantage, with compression being 7% faster. The few games they tested also showed no advantage, even at reduced resolution and graphics settings to push framerates up around 200 fps and prevent the graphics card from being the limiting factor. Tomb Raider got identical performance with either memory setup, Bioshock was actually 5% slower in quad-channel mode, and Dirt Showdown was about 1% faster in quad channel, which again, is probably within the margin of error.
Sure, there's undoubtedly other memory-intensive software like 7-zip that shows some performance gains from having a quad channel setup, but the differences are at the level where people probably shouldn't care much, unless they're running some sort of specialized server software or something where they know it will make a difference. The differences might be a bit greater going from single channel to dual, but again, plenty of software and games won't be affected at all, and in the limited scenarios where differences exist, those differences usually won't be huge.
I would certainly go for dual channel over single, but I also wouldn't go with a motherboard with only two RAM slots. For larger boards like this, I agree that there isn't much reason to go with only one stick. Adding another stick later should usually work fine if they match, but of course there could potentially be differences between production runs that some motherboards might not like. In the event that that happens, you won't likely be any worse off than if you had to replace a pair of sticks anyway though.
JackNaylorPE :
What I am mainly commenting on however, is that it was extremely rare to see a 1 - stick build here in the forums and I have answered at least 10 in the last 2 days.
I hadn't been paying much attention to that, but maybe some specific budget build with one stick of RAM was posted on a popular site recently. Or maybe there was some relatively good sale on a single stick of RAM somewhere.