Prototype1 :
.....I know the GPU is more important than the CPU in a gaming build but I don't know if the CPU will make a huge difference or not....
That was true 10+ years ago when basically all desktop CPUs fell within a tight range of possible performance, and GPU performance was the big variable. Today, there are very significant differences in compute performance for different workloads available across the spectrum of available CPUs. Unfortunately, most amateur system builders and hardware enthusiasts are still placating to that 10 year old philosophy about gaming rigs.
Today, your CPU choice will effect gaming performance very directly, as almost all popular games are compute intensive.
Perhaps the most important thing to understand, is that the "performance" you get from your GPU, is always an inverse relationship with visual quality. It doesn't matter if you use a $100 GPU or a $500 GPU, both can play games at 60FPS. The difference is that the $100 GPU will do it at 720P with medium visual quality, while the $500 GPU will do it at 1440P with high visual quality settings. Point being, the performance dictated by the GPU, is soft and adjustable, while the performance dictated by the CPU, is largely rigid, and only adjustable by changing the compute performance of the machine. It is in fact, the CPU that is more important than the GPU for performance. If your CPU is bottle-necking you to 30FPS, that's the hard limit no matter what GPU you select.
Be very weary of AMD shills who basically use the same marketing strategy that AMD's own marketing department uses to hide the effect of compute performance on gaming performance. A lineup of CPUs on a chart all getting 27FPS doesn't mean that all these CPUs perform the same in the game in question. This is nothing more than a dirty trick to hide the difference in compute performance under a rug. The rug, is an artificially created GPU bottleneck carefully chosen to erase the difference in compute performance. When the visual quality is adjusted to actually better suit the GPU used in the benchmark to shoot for a 60FPS goal (common), the veil is lifted, gaps in performance begin to show up. In some cases, they are large. Many games that run at 60FPS on haswell i5's, only run at 35-40FPS on 8 core Vishera's.
Another popular trick, is to show a lineup of CPUs all achieving the same very high FPS (90+) in a popular game title like BF3/4. The dirty secret here, is that these are single player conditions with very low compute overhead (again, what we are actually witnessing is the GPU bound performance cap). These results are largely irrelevant as the vast majority of players are interested in multiplayer performance. Sadly, it doesn't matter how many times we explain that these results are misleading, bogus, etc, the same people keep referencing the same misinformation to make their case for the AMD CPU for gaming, which proves that they are not interested in truth, but only in selling the AMD CPU, which proves to me, that they are shills.
You should understand, that I am speaking here as an AMD user. I've been using AMD platform builds for over a decade. My computer is an FX-8350 on Asus 990X EVO. I like the AM3+ platform as it offers lots of neat features that you won't find on competing Intel platforms, like ECC memory support, an IOMMU, extensive performance tuning options for tinkerers, very robust abuse tolerant silicon, and great performance/$ in compute intensive workloads than scale well to intercore parallelism. This platform has a lot of strengths and value, but not as a gaming platform.
Time to get off the fence....
A haswell core has more execution resources than a PD module (intracore parallelism), arranged in a more efficient manner (shorter average instruction pipeline length), with better cache performance and less instruction penalties. Clock for clock, a non-hyperthreaded haswell core generally has about the same execution throughput as an entire piledriver module. When hyperthreading is enabled, the haswell core achieves up to ~25% higher execution throughput than a PileDriver module.
In translation, this means that it doesn't matter if the workload presents as 8 threads, or 32 threads, or 592 thread, an i5 haswell performs pretty similarly to an 8 core vishera (when clocked similarly). It also means that any time the workload presents as less than 8 threads, the i5 is faster. More importantly, any time the workload presents in a manner that can only achieve absolute saturation on 1 or 2 threads at a time (that's pretty much ALL GAMES, including those that scale to 8+ threads!!!) the i5 will be faster.
The hard limits of performance in games are dictated by the CPU, not the GPU.
-----------
Darkbreeze :
An 8350 is more than capable of driving an R9 270x to the fullest of the GPUs capability. Especially if you overclock the CPU. The i5 4690k is more capable than the FX chips with bigger cards, but for that card it probably isn't going to make that much difference anyhow. For other computing tasks however, the i5 at stock speeds will easily beat the FX in most tasks with the exception being software or titles that are truly optimized for more than four cores.
In my testing, research, and direct experience, any CPU can bottleneck any GPU in the right conditions. The FX-8350 certainly does not guarantee to "drive" the R9 270X to the fullest of the GPUs capability. My FX-8350 bottlenecks my GTX460 more often than the other way around. In some games, the FX-8350 bottlenecks a GT520. (FYI, I have experience with more powerful GPUs installed in this system as well, there is no benefit for me).
Even an i7-4790K Overclocked to 5ghz can be the source of a performance limitation in some game titles regardless of the GPU selected.