Check out the runtime testimonial thread, Quark AA section, there some XML runtime tests and they look pretty comparable to the XPGs.. don't forget the XML is putting out higher lumens at the same level/on the same battery.
http://www.cpfmarketplace.com/mp/sh...ght-Run-Time-Testimonials-real-world-testing!
I ran a moonlight test on LR61s (AAAA) and my XPG-S2 got twice the runtime of my XML... but my light meter rates the S2 at half the lumens of the XML - 0.17 vs 0.33. The XPG-S2 4 lms spec also metered at 2.5 lms while the XML 2.7 lm spec metered at 3 lms for me. 47s seems to getting more conservative with its specs.
I've also run random side-by-side runtime tests with my SC52 and D25A Clicky Ti, all XMLs on 1xEneloops, and found all three lights to be equally efficient +- 10% or so on a lumen-hr basis - for example at max, the SC52 metered 25% brighter throughout, but the Quark ran 20% longer. The Quark and D25A specs are highly comparable... and let's just say way more "conservative" than the SC52s ;-).
Finally check out Selfbuilt's review of the QAA2X, by far the most efficient light in the 115 lm high range, not to mention still the brightest 2AA he's tested (although the D25A2 Clicky XML would probably be right there as well).
The QAAX is still my all time favorite and after much searching, I just found a NW Pro version to match my NW Tactical version
.