jgreco, I see you rank highly in the Multi-core geekbench
Well, there are all those 2687W results, but of course those are all dual sockets, and about a dozen of them seem to be for the same system. The 2697 would appear to be top-of-the-charts for a single socket system.
Were I to pop another 2697v2 on the board, I think it is clear that we could see a result of 60,000+ appear, but sadly, that's not likely to happen. We acquired the dual board because it had an LSI2208 and 10GbE integrated, at an overall low cost.
We've got three E3-1230 nodes I wanted to try consolidating. They usually seem to run shy on memory, sometimes also CPU. I didn't want to take a severe hit in terms of frequency; the 3.2GHz E3 part is sometimes a little uncomfortably slow as it is. As you know, that doesn't translate well to E5 options. The 2687W at 3.1GHz could have been a candidate, but losing four cores and going to a TDP outside the 135W the board was designed for was deemed a bad idea. That really only left the E5-2690, or
maybe a pair of E5-2643's. And I was a bit uncomfortable with that since it would be a 1/3 reduction in cores over the trio of E3-1230's, so it might well have ended up being a pair of E5-2690's.
But by the time we finished with the base platform, it was getting to be April anyways, so I decided to wait a bit and see if Intel was looking to come out with anything compelling for Xeon Ivy.
So in the end, the new E5-2690v2 (10 core, 3GHz) looked extremely compelling, but the 2697v2 (12 core, 2.7 GHz) was also very interesting. Since tech specs hadn't been released, it made for a difficult comparison, but in the end, 12 * 2.7 = 32.4 and 10 * 3.0 = 30, and we'll have to hope the turbo boosts are similar.
Anyways, I still feel that the 2697v2 was a good choice. From a capex point of view, two E5-2690's would have been more expensive, and from an opex point of view, the 2697 is probably the best choice for reduced power consumption and cooling.
