Those beeing critical of the review should stop for a moment, and separate oppinion to the quality of the review itself.
A good review have nothing about agreeing what is interpreted from the analysis. The work in this review is marvelous, and we even get it for "free".
I am still very impressed by the cpu analysis
http://www.anandtech.com/show/6747/htc-one-review/11
Its the first time we get all the info about krait300 one place, excellent low level analysis, and Brian goes a step futher to find out
why the results is that, and how it compares to the known arhc. changes.
He fx. practicalle proofs the L2 hardware prefetcher rocks, and have excellent perf/power ratio. We get to now how effective and important a good prefetcher really can be done the right way, and get som indicating numbers for it.
What review comes even 5% close to that analysis of a phones cpu? The work goes far beyond what can be expected from a review. Its far more than a review, and even borders what i would label research.
Brian reviews is the sort of work, where 1-2 years after they are published you go back and search for knowledge. Its a very seldom quality.