VR Zone points out the increased memory bandwidth of AMD's and NVIDIA's new high-end GPUs is more for GPGPU computing than for gaming:
Well, GPUs have very very high peak FP compute rates, but the real performance in FP-heavy code is greatly affected by the available memory bandwidth. While a game may get a 5% FPS boost from a 33% memory bus width jump, a vector FP run may gain over 20% from the same enhancement, all other things staying the same. So, whatever the vendor's primary intentions are for widening the graphics local memory buses on new GPUs, the extra bandwidth - and yes, extra capacity headroom from being able to add more chips and keep things in local RAM - will have the greatest benefit exactly for GPU compute apps. The impact? More HPC-like stuff running oh so fast on your PC, and more true HPC apps running faster on GPGPU arrays.