If you're not locked on an AMD build, then I would advise an i5 haswell instead. Much better arrangement of execution resources for real time workloads like gaming. Unless you have access to a microcenter (really good combo deals on AMD CPUs and boards) there aren't many situations where an FX chip is a better value than a haswell chip for gaming.
-------------
Clock speed and core count are very superficial and misleading specifications in the world of computing hardware; this applies to both CPU's and GPU's. GPU clock speeds are almost irrelevant, as pretty much every GPU these days has a core clock ~800-1200mhz. The specifications that matter, are unfortunately in many ways only comparable within the same family of GPUs...
If core clocks were the most important indicator for GPU performance, then the R7 260X would be AMD's fastest GPU. That's obviously not true at all.
Look here:
http://en.wikipedia.org/wiki/List_of_AMD_graphics_processing_units#Volcanic_Islands_.28Rx_200.29_Series
and here:
http://en.wikipedia.org/wiki/List_of_Nvidia_graphics_processing_units#GeForce_700_Series
The GP/s, GT/s and GB/s VRAM bandwidth specifications on the charts can be used as comparative indicators when comparing GPUs, but keep in mind that real world performance and theoretical throughput are not the same thing. Some architectures close the gap on theoretical better than others, and the conditions of the workload also greatly effect this.
Counting cores and ghz is about as useful as counting cyliders and rpm for an engine. I can build you a miniature replica V12 engine that spins 30,000RPM. Does having 12 cylinders and 30,000 RPM make it more powerful than the 4 cylinder engine in a sedan that redlines at 6000rpm? RPM and Cylinder count are meaningless without more details about the engine, like the displacement, the fuel type, the sophistication of the valve-train, fuel/air management systems, and/or any atmospheric modifications.
--------------
Wattage ratings on a PSU can also be very superficial, which is why GPU manufactures are forced to recommend enormous PSU sizes to power their GPUs. Their PSU sizing recommendation has to accommodate a world where not all PSUs are created equal, and not all wattage ratings are held to the same standards.
There are some 700W PSUs that really can't sustain better than ~500W continuous, and even at those power levels, are going to live a very short life (raidmax has lots of examples of this type of overrated PSU). On the other hand, there are many 450W rated PSUs that could run at 500W for many years without failure. So when the GPU says "700W PSU required," what they are really doing is protecting themselves from a buyer like yourself, who is oblivious to the vast differences in quality among PSUs. The typical bonehead consumer maneuver is to pick the biggest wattage rating for the lowest price, and there are brands placating to that approach. I would advise that you try to avoid this approach, and have a qualified 3rd party help you calculate peak power dissipation of the system and size the PSU accordingly, selecting from among the known great quality PSUs available.
Consider:
The R9 280, is configured not to exceed a power envelope of approximately 250W at stock clocks. An i5, even with VRM losses accounted for rarely exceeds 80-90W power dissipation (unless it's a K model overclocked). The remainder of the system (south bridge/controllers/USB/sound/RAM/Drives etc) is typically ~20-50W depending on system configuration. If you do the math here, you might realize that a 450W PSU that can actually deliver 450W is actually LARGER than necessary for an i5+R9 280 build. You'll find that many professionally designed systems are using 450W PSUs with flagship GPUs like the GTX780Ti. This is perfectly fine as long as the system builder has correctly calculated the peak power dissipation of the machine as configured and selected a PSU that is actually designed to be able to deliver it's label ratings with ease.
---------------
If your goal is to win the bench-marking challenge, scrape together the cheapest MOBO/CPU/RAM/Case/PSU/HD you can and blow the whole budget on a GTX780Ti. You'll win nearly any gaming "benchmark" with this approach because almost all game-specific FPS benchmarks use a single player sequence with low compute overhead to firmly plant the bottleneck on the GPU.
If your goal is to build the better gaming machine for real world gaming and long term reliability, you're going to have to loose the challenge, and choose a good strong i5 CPU to handle the high compute workloads that only present themselves in real-world multi-player conditions (not repeatable for bench-marking sequences), and then pair that i5 with a middle-tier GPU in order to afford to build the rest of the system from quality parts.