NVIDIA SLI Interview


posted 9/10/2004 by John Yan
other articles by John Yan
One Page Platforms:
GamingNexus: How does the technology work? Is it all hardware based or is there a software component?

Steve Sims: SLI takes the PCI Express multi-GPU configuration to the next level because it features intelligent patent-pending hardware and software technology. We have dedicated scalability logic in each GPU and a digital interface between GPU's to help enable this. . In addition we have a full software suite that enables dynamic load balancing and advanced rendering algorithms that provide the best image quality and performance.

GamingNexus: On the surface, how is the process compared to Alienware’s where it load balances the rendering between the two cards.

Steve Sims: The Alienware solution they announced does not take advantage of any of the dedicated scalability logic, inter-GPU digital interface or dynamic load balancing software of SLI.

GamingNexus: Is this something you couldn’t do with AGP or did you just see a new opportunity with the new bus technology?

Steve Sims: The AGP bus was not conducive to an SLI configuration. AGP is not really setup to feed multiple GPU’s data at the speeds necessary to make SLI work efficiently

GamingNexus: Will this work with any new PCI Express board or will this require specialized motherboards?

Steve Sims: SLI technology requires a PCI Express motherboard with two x16 physical connectors. The graphics cards plug into these connectors. The graphics cards can work with whatever routing because x16 PCI Express connectors can auto-negotiate down to x8, or x4 electrical. “SLI Ready” PCI Express motherboards will also be available soon from: Abit, ASUS Computer International, Epox, Gigabyte Technology, Co., Ltd., SuperMicro, Tyan, and Micro-Star International. We will also be supporting SLI on our nForce platforms in the near future.

GamingNexus: Did you use any knowledge from the 3Dfx method or is this all a completely new design scheme?

Steve Sims: This technology was developed entirely at NVIDIA. Many of the engineers are former 3dfx employees, though.
Page 2 of 3