Single page Print

Nvidia's nForce 780i SLI chipset

Old dog, new tricks

If you want to combine one of Intel's swanky new 45nm processors with an SLI multi-GPU graphics configuration, there's a problem. You see, while nForce 600-series chipsets are compatible with Intel's latest Penryn-based CPUs, current motherboard implementations are not. The incompatibility runs deeper than what can be addressed with a simple BIOS update, as well—a board-level circuit change is required.

Motherboard makers could respin their nForce 600-series designs with the necessary changes, but the nForce 600 series is more than a year old now, so it's due for a refresh anyway. That refresh comes in the form of the nForce 700 series, led by the flagship nForce 780i SLI.

In many ways, the nForce 700 series is identical to the 600 series that preceded it. However, Nvidia has added a few new wrinkles to the equation, including support for second-generation PCI Express, three-way SLI configurations, and its proposed Enthusiast System Architecture. There's a new reference motherboard design for the nForce 780i SLI, too—one that will work with Intel's upcoming 45nm processors.

So what's actually new in this nForce 700 series and the 780i reference board design? More importantly, is this refresh enough to keep the nForce 780i SLI competitive with Intel's recently released X38 Express chipset? Keep reading to find out.

The new chipset
Traditional chipsets have two components: a north bridge, which Nvidia calls a System Platform Processor (SPP), and a south bridge, which Nvidia refers to as a Media Communications Processor (MCP). The 780i SLI follows this logic with the usual SPP and MCP components, but Nvidia adds a third chip into the mix to handle PCI Express 2.0. This nForce 200 chip connects directly to the north bridge and supplies 32 lanes of PCIe 2.0 connectivity that can be split evenly between a pair of physical x16 slots or even divided between four cards with eight lanes each.

The nForce 780i SLI block diagram. Source: Nvidia
PCI Express 2.0 calls for a signaling rate of 5.0GT/s, and the nForce 200 delivers that in full to each of its 32 lanes, yielding 1GB/s of bandwidth per lane. The nForce 200 is also capable of routing signals directly between graphics cards connected to it, so in two-way SLI configurations, it handles any SLI traffic not passed over the bridge connector without hitting the north bridge. Three-way SLI setups hang the third graphics card off the south bridge, requiring that traffic not routed through the SLI bridge connector hop between the nForce 200 and the MCP through the north bridge.

SLI threesomes are likely to be rare, but single-card and even two-way SLI configs are not, which brings us to the nForce 200's connection with the 780i SPP. With the nForce 200 serving up 32 lanes of PCIe 2.0 at 1GB/s of bandwidth per lane, you'd expect the chip's link with the north bridge to be pretty beefy. And you'd be wrong. The link in question is described as an "Nvidia interface" with 16 lanes running at a 4.5GT/s signaling rate. This yields aggregate interconnect bandwidth of only 14.4GB/s—well short of the 32GB/s that the nForce 200 supplies to graphics cards connected to it. According to Nvidia, 14.4GB/s is still enough to "enable full performance of PCI Express 2.0 graphics cards." We'll be putting that claim to the test in a future article; time constraints didn't permit doing so for this review.

The nForce 780i SLI SPP (top) and its nForce 200 companion chip
Using the nForce 200 seems like a convoluted way to bring PCIe 2.0 connectivity to the 780i SLI. New chipsets from AMD and Intel put PCIe 2.0 right into the north bridge and offer full end-to-end 5.0GT/s signaling rates without the need for a third chip. So why is Nvidia using the nForce 200? I suspect it's because the nForce 780i SLI SPP isn't really a new chip at all. Nvidia MCP General Manager Drew Henry told us the 780i SLI SPP is an "optimized version of a chip we've used before," suggesting that it's really a relabeled nForce 680i SLI SPP.

If you recall the last couple of Nvidia SPP chips, you'll remember a feature called LinkBoost, which cranked up the link speed for the chipset's PCI Express lanes. Nvidia was adamant that this wasn't overclocking since the chipset had been fully validated to run at higher speeds. I think we're seeing an extreme version of LinkBoost in action here, with the 780i SPP simply being a 680i SPP whose 16-lane PCIe 1.1 link has been coaxed into running at 4.5GT/s and validated at that speed. This approach would be fitting considering that second-generation PCI Express is really just gen one cranked up to a faster signaling rate. But it's a shame Nvidia didn't manage to nail 5.0GT/s on the button.

The nForce 780i 570 SLI MCP
Things get even more retro at the south bridge, where we find that the nForce 780i SLI MCP is in fact an nForce 570 SLI MCP. It even says so right on the chip. This chip's orgins can be traced way back to May of 2006, when it first debuted as the south bridge component of the nForce 590 SLI chipset for Socket AM2 processors. Since then, it's been passed around more than a wasted girl with low self esteem at a frat party, seeing action in nForce 570, 590i, and 680i chipsets. In fact, the 570 SLI MCP's promiscuity adaptability won it best chipset distinction in our TR Awards for 2006. The only thing new this time around appears to be the 570's ESA certification.

Despite tapping SPP and MCP components that aren't entirely new, the 780 SLI's feature set compares favorably with Intel's Core 2 chipsets, including the latest X38 Express.

X38 Express P35 Express nForce 780i SLI SPP nForce 680i SLI SPP
Front-side bus 1333/1066MHz 1333/1066MHz 1333/1066MHz 1333/1066MHz
Memory controller DDR2-800
DDR2-800 DDR2-800
PCI Express 1.1 lanes 0 16 2 18
PCI Express 2.0 lanes 32 0 32 0
Multi-GPU support CrossFire CrossFire* SLI SLI
Chipset interconnect DMI DMI HyperTransport HyperTransport
Peak interconnect bandwidth 2GB/s 2GB/s 8GB/s 8GB/s

Front-side bus speeds are supported up to 1333MHz, allowing the 780i to handle all but Intel's upcoming Core 2 Extreme QX9770, which will call for a 1600MHz FSB. Nvidia says it will have products out to support a 1600MHz front-side bus when processors requiring that speed are available on the market next year. Expect those products to jump onto the DDR3 memory bandwagon, as well. DDR3 prices will hopefully sink to more reasonable levels by then, because with current market pricing, DDR2 is a much better value.

We should note that while the 780i's memory controller is listed at DDR2-800—the fastest JEDEC-approved spec—Nvidia says its memory controller is capable of handling DIMMs running at up to 1200MHz. The nForce 780i SLI supports Enhanced Performance Profiles, as well, so if you have EPP memory modules rated for 1200MHz, 780i-bassed boards should detect and automatically run them at that speed.

Like the 680i SLI chipset that preceded it, the 780i offers gobs of interconnect bandwidth between its SPP and MCP components. A HyperTransport link serves up a whopping four times the bandwidth available through Intel's DMI interconnect, but given how Nvidia has mapped out three-way SLI, it really needs the fatter pipe.

ICH9R nForce 780i SLI MCP nForce 680i SLI MCP
PCI Express 1.1 lanes 6 28 28
Serial ATA ports 6 6 6
Peak SATA data rate 300MB/s 300MB/s 300MB/s
Native Command Queuing Y Y Y
RAID 0/1 Y Y Y
RAID 0+1/10 Y Y Y
Matrix RAID Y N N
ATA channels 0 1 1
Max audio channels 8 8 8
Audio standard HDA HDA HDA
Ethernet 10/100/1000 2 x 10/100/1000 2 x 10/100/1000
USB ports 12 10 10

That's because the third card in an SLI trio hangs off the south bridge's PCI Express lanes, chewing up a significant chunk of interconnect bandwidth. And the nForce 570 780i MCP isn't exactly short on other bandwidth-hungry peripherals, either. With a pair of Gigabit Ethernet controllers, six Serial ATA RAID ports, and an additional 12 PCI Express 1.1 lanes not dedicated to graphics, the MCP bristles with connectivity options. Heck, it even has an ATA channel.

About the only things the 780i MCP loses to Intel's new ICH9R south bridge are a couple of extra USB ports and support for Matrix RAID, which allows multiple RAID partitions to span a single set of drives. The 780i MCP also lacks support for the Advanced Host Controller Interface (AHCI), although unlike the ICH9R, it's able to implement features like hot swap for SATA drives and Native Command Queuing without.