Well that review was useless. Between the 980 reviews and this you've changed the length of the benchmarks making the results non-comparable, and then you test the Sapphire R9 390X against stock cards.
1) Keep tests consistent or
2) Add in 3rd party cards to the graphs