I've done some searching on the forum and google and haven't found a definitive answer for enthusiast cards. I realize that the hypermemory/turbocache system was employed to lower the cost of a card by reducing the on-card memory and utilizing system memory. However, would this option be viable for those who use extreme graphics, such as high resolutions with AA and AF? For those who have large amounts of ram at their disposal, 4GB or higher, would there be a substantial performance hit? Or is this just absolutely offset by mutli-GPU/card solutions with high on-card memory?