I don't want to rain on any parades.
I have to gravitate toward ThermalRight's coolers because they have consistently outgunned the competition.
There are two things you look for in recent reviews of these products.
First, is the item of interest rated by measured thermal resistance in degrees-C/Watt?
That value will tell you almost exactly how much the CPU temperature will rise given the thermal wattage of the processor -- whether stock speeds or over-clocked.
For instance, an over-clocked processor with a stock TDP of 95W may have a thermal power of 120W. If the thermal resistance is 0.11 C/W, then the idle-to-load temperature spread should be 13.2C degrees.
While the Ultra 120 Plus or Extreme has not been released, you can estimate its thermal resistance from a truly objective comparison review, when the original Ultra 120 is thrown in as a reference. The original cooler was measured by Citarella at Overclockers.com to have a C/W of 0.11. Given the temperatures exhibited in this review:
AnandTech on ThermalRight Ultra 120 Extreme
We can estimate the thermal resistance for the Ultra 120 Extreme from Citarella's Ultra 120 review and his "heatsink roundup" comparison summary, showing the original Ultra 120 at a thermal resistance value of 0.11 C/W.
The X6800 used in the review has a TDP of 75W. So we look at the numbers with all stock settings at the processor's rated 2.933 Ghz. Therefore:
Delta-C/W = 1C / 75W or 0.0133 C/W
So if the original Ultra 120 proved a thermal resistance minimum (?) of 0.11 C/W, the new version?s TR-value would probably be about 0.097 C/W.
The second thing to look at in reviews is the frequency of comparisons with several models, and the choice of models for those comparisons which you're lucky enough to find.
With this CoolerMaster Gemini II, I've searched for comparison reviews all over the place, and the ones I find compare the cooler with models known to perform worse than either the TR or Zalman coolers. I can't find enough performance comparisons. But I see the Ultra 120 (and Extreme) in comparisons with several good coolers, and they always come out on the top of the heap.
When it really boils down to efficiency as represented by thermal resistance measurements in a lab, few manufacturers (probably none) publish that factor in their specifications. Seldom -- although occasionally -- do you find lab tests that provide the actual measurements and comparisons (as with Overclockers.com). If the testbed is controlled for processor TDP and clock settings, just providing idle and load temperatures is sufficient to show relative efficiency, and you can estimate thermal resistance using that data in conjunction with other known facts.
I believe, that if you did find this data on thermal resistance readily, there would be a shake-out in the market for these devices with fewer players left -- provided that enthusiasts would all be savvy enough to see that the cooler with the lowest thermal resistance performs the best cooling. And of course, we know that people may choose these things because they "look" like they'll really do the job -- without attention paid to the data we're discussing here.
The only other reason, besides efficiency, for choosing one cooler over another, would be the orientation, weight, fan-noise and fan-size. With the Gemini, it has the advantages of coolers like TR's SI-120 for directing air at the motherboard. But with the appropriate ducting and other uses of fans, a cooler like the Ultra 120 doesn't need to do that. Even if the design doesn't kill two birds with one stone, there are other ways to kill the two birds -- to skin the cat so to speak -- because there are other aspects of case-cooling that require intake and exhaust.