NVIDIA Hybrid SLI and nForce 700a Chipsets

As we've mentioned, a key component to Hybrid SLI technology is a core logic chipset with a compatible mGPU.  To that end, NVIDIA has developed a line of nForce chipsets for the AMD platform that feature an integrated graphics processor.

   
NVIDIA's Drew Henry with the nForce 780a SLI motherboard

The flagship model in the nForce 700a series of chipsets is the 780a SLI.  The 780a SLI features two video outputs (one digital, one analog) and an mGPU with similar functionality to a GeForce 8400 series discreet graphics card.  This mGPU is Windows Vista Premium certified and DX10 capable.  It also features NVIDIA's PureVideo HD engine for full CPU offload of all HD video codecs.  The 700a series of MCPs (norhtbridge), will be manufactured at 65nm, and are mostly single-chip designs.

As you can see in this high-level block diagram, however, the 780a SLI uses NVIDIA's NF200 chip for 32 lanes of PCI Express Gen 2 and 3-way SLI support.  The MCP also features an additional three lanes of PCI Express Gen 1 connectivity, a single GigeE network controller, 12 USB 2.0 ports, Azalia HD audio, up to five PCI slots, and Media Shield and ESA support.

As we noted earlier, with this product launch NVIDIA has also disclosed details regarding NVIDIA patented technology that is built into their NF200 PCI Express Switch chip. Essentially, what NVIDIA has done is build a couple of fast paths inside the switch device, dedicated to optimizing mulit-GPU SLI transaction performance both back to the root CPU complex and peer-to-peer between GPUs. This is critical because there is a single Gen2 X16 PCIe link between the NF200 chip and the 780a MCP.

Specifically, there are two functional blocks as you'll not in the above diagram, denoted as "Broadcast" and "PWShort". The Broadcast block specifically provides a broadcast send mode for the root complex down to all GPUs in the system. This allows efficient transfer of data in one group transaction. PWShort, (which stands for Posted Write Short), is a dedicated cut-through mode for peer-to-peer communications between the GPUs, without the need to tap on upstream bandwidth to the CPU complex.

That about wraps up our coverage of the new announcements from NVIDIA.  We'll have more from NVIDIA and the CES show at large, as the days unfold. Stick around!