This suggests that either Kaveri is a step backwards in CPU performance (which doesn't fit with reviews - though it wasn't much of a step forward) OR Mantle really is working.
I don't think the Kaveri CPU is a step backwards from its predecessor (Trinity/Richland) although it's only marginally faster due to lower clock speeds. However, it's blatantly obvious that Kaveri *is* a step down in performance compared to 6 core or 8 core CPUs in the FX series assuming you aren't doing single-threaded workloads. We've heard plenty about how BF4 is multi-threaded, so more cores will mean those older FX chips are noticeably superior (not to mention any quad-core Intel part from 2011 onward).
The TechSpot review is 3 months old using version 1.0 of the game for example, I understand that the game has received many patches since then.
Yes, I fully agree that unless Dice has been actively working to slow down BF4 that the DX11 numbers on the R9-290X should be even stronger than the review I posted of the 1.0 version that had significant issues.. that's not helping Mantle's case though.
The TechSpot review also states that it used the beginning of the 4th single player mission to benchmark while we don't know which section of the game AMD used to benchmark. It appears very much to be a case of apples and unknown fruit...
That's a legitimate argument since we only have video from the AMD demo (showing a guy running around on the deck of an oil tanker or something like that) and we don't know exactly what benchmark run was used in the earlier review. Of course, that cuts both ways.. it might help AMD or turn out even worse.