nForce 590: Foxconn C51XEM2AA - PAGE 2Geordan Hankinson,
Tom Karpik - Tuesday, May 23rd, 2006
nForce 590
What is important to know about NVIDIA's 500-series MCP is that it is being dubbed as just that -- simply a new Media and Communications processor, rather than an entirely new chipset. In the case of the 590 SLI, NVIDIA have paired their nForce 590 SLI MCP with a C51 System Platform Processor (SPP, or Northbridge). The C51 first made an appearance in GeForce 6100/6150 IGP motherboards, and relatively little has changed as far as the north bridge is concerned.
At the top end, NVIDIA have added a large number of features over the latest nForce 4 MCP, which include performance enhancing tweaks, better overall ease-of-use, and some new networking performance enhancers. A number of these enhancements make their way down the line from the 590 to the other three lower priced variations, the 570 SLI, 570 Ultra, and 550.
Here are four block diagrams highlighting the key differences between the four MCP's.
Architecturally, the only difference between the 590 SLI and the other members is the addition of the C51 north bridge, which provides the first 16 PCI-Express lanes and a HyperTransport tunnel to talk to the MCP, while the MCP provides the other 20. All of the other members in the 500-series family are single-chip solutions, with the MCP handling all system functions. As far as features go, however, the 590 has support for 'LinkBoost Technology' which we will discuss in detail in a bit.
As with nForce 4, not all 500-series chipsets support all Socket AM2 processors. The 590 SLI, 570 SLI, and 570 Ultra chips do not appear to support Sempron processors, while the 550 will not support FX series chips. This should not pose a problem for many, as anyone buying either an FX or Sempron processor will likely purchase a board that fits with the respective price segment of each processor.
NVIDIA have spent a lot of time focusing on the interests of the demanding enthusiast and have taken a number of steps to try and make the tweaking process as straightforward as possible.
LinkBoost
The first of these "ease of use" features is NVIDIA's LinkBoost technology. LinkBoost increases the frequency of the PCI-E x16 slots and HyperTransport link between the SPP and MCP. While these frequencies can be set manually on most motherboards, LinkBoost is really there for your convenience, and a way through which NVIDIA can guarantee a boost in performance without stability issues.
Unfortunately, there is a small catch with LinkBoost, and that is that it only works with a GeForce 7900-series video card. Throw in any other card into the system, and LinkBoost is disabled -- but you're welcome to try setting the same clocks anyway.
Audio
The audio portion of the new MCP, while not on the same level as Soundstorm, still provides a solid base level experience with High Definition Azalia. Optical-out is supported as well, and finds its way onto the reference board which we will cover on the next page.
Networking
With the 500 series, NVIDIA have also worked at improving networking performance in specific scenarios. This is a one-two punch of features, with the first being dubbed "First Packet".
First Packet is essentially a QoS (Quality of Service)/packet prioritization system. While some higher-end routers from Linksys et. al. allow for customizable prioritizing of packets from game/VOIP packets, First Packet builds this right into the chipset. This will divert packets from other applications competing for upstream bandwidth, and give preference to your game (or any other pre-configured application).
The second networking feature is what NVIDIA dubs "teaming". Teaming is the key feature under NVIDIA's DualNet slogan, which is basically a way of saying that there are two dedicated GigE ports integrated into every 590/570 MCP. While separate GigE ports have their uses, NVIDIA hopes that you will be using them in as a pair, thus increasing redundancy and bandwidth. This is the basic idea behind "teaming". Two ports plus one Gigabit switch equals one TCP/IP interface capable of 2 Gbps presented to the world.
Storage
NVIDIA have updated their MediaShield utility, and have also added support for up to 6 SATA drives in a RAID 5 configuration. MediaShield provides desktop configuration of RAID arrays and also provides information about drive health. It's possible to RAID 5 six disks in two arrays of three disks each now, and MediaShield will keep tabs on the array status and rebuild sectors as necessary.
Memory
Another very notable feature implemented in the 590 SLI chip is Enhanced Performance Profile (EPP) memory support. Our Editor-in-chief explained it in detail in this article, but here is the gist of it, taken from his story.
In a nutshell, EPP are additional settings profiles stored in unused SPD space. The JEDEC specifications actually only require a very small amount of data to be stored in the SPD, which leaves a lot of room for additional information. Corsair and NVIDIA took advantage of this space to do something that is actually quite ingenious: they decided to use it to store tweaked memory settings which encompass parameters far exceeding what JEDEC covers but which advanced overclockers tweak on a regular basis. The EPP profiles include settings for such things as memory voltage, drive strength, tRCD, tRP, tRAS, and many other settings that you probably never touch at all.
The purpose of EPP is to simplify memory tweaking for the casual or beginning enthusiast, and even serve as a starting point from which more advanced overclockers can fine tune their settings.
Let's now take a look at the Foxconn C51XEM2AA motherboard and what it means to have an NVIDIA reference design.