Review Index:
Feedback

NVIDIA nForce Professional 2200 MCP

Author: Ryan Shrout
Subject: Chipsets
Manufacturer: NVIDIA
Tagged:

nForce Professional 2200 and 2050

NVIDIA nForce Professional 2200 MCP


The nForce Professional 2200 MCP is the actual logic chipset that is used to communicate between the processors and the I/O on the motherboard.  Just like the single chip design that is seen on the nForce4 platform, the nForce Pro 2200 supports PCI Express, AMD's 1 GHz HyperTransport bus and various storage options.



The NVIDIA nForce Professional 2200 chip supports the 940-pin AMD Opteron processors in both single and multi-processor configurations.  NVIDIA is also pushing that this chipset is completely ready for the dual-core Opterons that will be released later this year, so servers and workstations built around it will be able to easily swap out current processors for future ones and get the benefits of dual-cores. 


The PCI Express support on the chipset consists of four separate controllers and a total of 20 lanes of PCI Express.  That means that board vendors can choose to design a board with up to four PCI Express slots of varying width; this is just like the nForce4's PCIe support.  Theoretically, you could have four x4 width PCIe lanes two x8 lanes and a x4 lane.  Most common so far has been the single x16 and single x4 PCIe configuration. 


The nForce Pro 2200 also has the same built in Gigabit networking MAC that the nForce3 and nForce4 used, but this iteration of it has an updated TCP/IP offloading engine.  The engine supposedly decreases the CPU utilization of the network engine even more so than on previous chips and the high-end workstations and servers that are using a ton of bandwidth will see more benefit to this feature than the majority of end users. 


The storage controller on the MCP is similar, with four SATA channels and two PATA channels with RAID 0, 1 and 0+1 across SATA and PATA.  Also included on this new chip is support for NCQ hard drives (native command queuing) that allows for hard drives to change the order of their read requests on the fly to increase efficiency and performance.  This is the first AMD-based chipset to support this option. 


NVIDIA nForce Professional 2050 I/O Chip


The nForce Pro 2050 chip is an additional chip that will not work on its own and must be coupled with a 2200 MCP to function.  It does not include additional logic for communicating with the processor, but as many as three 2050 chips can be added to a board with a 2200 MCP for a very robust, and very expensive, motherboard.



What the nForce 2050 chip adds to a motherboard is 20 additional lanes of PCI Express, an additional NVIDIA Gigabit MAC and four additional channels of SATA-2.  It does not add any USB 2.0 connections or IDE channels, only what I have listed.  This ability to add additional PCI Express lanes is the first step in the new change over to scalable interfaces in modern computers: where as the PCI bus was a shared electrical circuit, the PCI Express bus is independent and thus allows for these kinds of chips to be developed.  It is similar to how we saw the HyperTransport bus dramatically change the way the AMD processors and platform evolve over the past two years or so.


The down side to all of this functionality is that boards with this many traces are going to require a much more complicated motherboard design and probably 6 or 8-layer PCB.  These are going to drive up the cost of these products, which is exactly why NVIDIA is bringing them into the workstation and server market that can buffer the price increases. 

No comments posted yet.

Post new comment

The content of this field is kept private and will not be shown publicly.
  • Lines and paragraphs break automatically.
  • Allowed HTML tags: <a> <em> <strong> <cite> <code> <ul> <ol> <li> <dl> <dt> <dd> <blockquote><p><br>
  • Web page addresses and e-mail addresses turn into links automatically.

More information about formatting options

By submitting this form, you accept the Mollom privacy policy.