not likely to fill the 3.0 bus at this level... but what can be an issue are the individual components. The CAS reference was in regards to System Memory. CAS 8 or 9 versus CAS 11 can have an impact. More importantly though is the relationship of the speed and the width. As noted in the examples earlier, design choice can severely reduce data rate if you wound up with a narrower interface. The wider that interface, the more impactful each MHZ increase in speed. 128bits will theoretically net about 15MB/s gain, while 256bit will come out to about 30MB/s. That adds up pretty quick wen the difference in clock speeds are on the order of hundreds of MHz.

The same applies to system memory and the CPU. When you ramp up the core bus speed for the CPU--you are also ramping up the speed of any onboard memory (like L1 Cache). In some systems, it also ramps up peripheral bus speeds also (like PCI). Increasing system memory clock has a considerable impact as well.

The issues with this game are two-fold really. It is still very heavily CPU bound because of the screwy DX9 implementation that doesn't offload as much as it could to the GPU. So all of these things stack together. If you are nerfed 10% on the CPU, 20% on the System Memory, and then 15% somewhere on the GPU---it can all combine to make a strange mix of performance issues, and makes it difficult sometimes to track down the real culprit. Best to optimize everything the best you can.