The explanation of the newer platform being slower comes from one architectural feature of the nForce4 SLI X16. To ensure high-speed communication between the chipset’s North and South bridges, NVIDIA had to reduce the width of the HyperTransport bus the chipset connects to the CPU with and to use two 8-bit channels instead of the ordinary nForce4’s two 16-bit channels there. The smaller bandwidth of the CPU-chipset link does not show up in games but the performance of applications that actively use the system bus, like data archiving utilities for example, may suffer a lot. We are not even sure the nForce4 SLI X16 will offer any advantage in the next generation of PC games, despite its two full-width PCI Express x16 slots, because the narrow HyperTransport bus will probable be a bottleneck. Being the first chipset with the PCI Express x16 + x16 formula, the NVIDIA nForce4 SLI X16 is interesting and unique, but it is no better than the ordinary nForce4 SLI, at least in current applications.