^^ Here's the problem: The ability to use more cores does NOT automatically make your program faster.
Look at it this way: You have a single core chip that has to do 8 things over a 1 second period. As long as those things all get done over that timespan, there is NO performance benefit in using more cores. It's when the work being done is greater then the cores ability to complete it that you begin to benefit by adding more cores.
Likewise, adding cores at the expense of single core performance can also cause the opposite problem. If ANY single core gets more work then it can complete, then the program in question has to sit around waiting for that core to catch up.
These two explanations show why Intel is typically faster (or in the worst case, as fast) as AMD in gaming: Even in titles that do scale beyond 4 cores, Intel's quad core chips are fast enough where no single core is getting overloaded. As a result, there is no CPU bottleneck, and your FPS is bound by single core performance. That's why you sometimes see benchmarks where the FX-8350 produces the same FPS as a core-i3 3220.
Where AMD has the advantage is reduced latency. While the i3 line, over a period of a second, is able to get all its work done, on a frame by frame basis, frametimes tend to be inconsistent. As a result, the 8350 would tend to be *smoother* in gaming over the i3, even if both output the same FPS.
Given DX12 should drive down CPU load, Intel actually would benefit more then AMD, as it's core i3 line should be much more attractive, as the above problem should occur less often. Sure, AMD will have pretty core loading, but that doesn't translate to performance.
This is the same exact argument I used to argue Bulldozer was a bad design back before it released. And it's still perfectly valid today.
So yes, you'll get a performance benefit via a better API. But no, the ability to use more cores won't vault AMD ahead of Intel. If anything, the i3 line will become more competitive with the FX-8xxx lineup by virtue of making the i3 perform better.