NVIDIA Firewall
Another of the features of the nForce 4 is the improved integrated hardware firewall. Just as with the SLI technology, the firewall is built directly into the MCP to reduce system overhead. Unlike your Linksys or Dlink broadband router which relies mostly on NAT for firewall protection, the NVIDIA nForce 4 firewall uses what NVIDIA calls ActiveArmor secure networking engine. This solution, according to NVIDIA, makes this the industries first true hardware-based PC firewall.
This technology also allows the firewall to operate without using the system CPU for overhead, so unlike a software firewall, the nForce 4 firewall will not bog down other computing tasks by requiring the main CPU to process packets. The nForce 4 firewall also supports remote access, configuration, monitoring, command line interface (CLI), and WMI scripts. These features allow you to have a great deal of control over how the firewall operates should you desire, and are all able to be configured through wizards so almost everyone should have little trouble in configuring their firewall.
Another feature that makes the nForce 4 firewall quite useful as opposed to standard software firewalls or firewalls integrated into standard broadband routers is the technology in use to filter packets. The first advantage over software firewalls is that the nForce 4 firewall is activated the instant that the network connection becomes active, so you are protected from the instant that the operating system loads. And unlike the standard broadband router which primarily uses NAT as its method of protection, the nForce 4 firewall uses stateless and stateful methods to filter packets. Though the actual nuts and bolts of stateless and stateful packet filtering are not the focus of this article, what they basically do is filter packets based on rules in place for the TCP or UDP port that they are being transmitted over, their source or destination address, and some other criteria. The actual difference between stateful and stateless packet filtering is essentially that in stateless packet filtering, every single packet that passes through the network adapter must be checked against the rules in place. This can lead to a sever drop in performance when there are a lot of rules that each packet must be checked against. Stateful packet filtering, on the other hand, uses the idea of connections to filter packets. Rather than filter every single packet that passes through the interface, packets are only tested when the connection to a specific host is initially created.
Also included is an Intelligent Application Manager, which allows you to specify applications that are and are not allowed to access the network, much like other software firewalls. This feature uses alert messages much like those in ZoneAlarm and other software firewalls to notify you when applications attempt to access the Internet or your computer.
It may be important to note that on the mainstream/value nForce 4 chipset, while the firewall is fully supported, the ActiveArmor technology is not. This more than likely means that more CPU overhead will be involved when using the hardware firewall on a value nForce 4 motherboard than on one of the higher end Ultra or SLI versions. In other words, on both the nForce 3 250Gb and nForce 4 (value), the firewall is hardware optimized via software. The nForce 4 Ultra and SLI are hardware accelerated via ActiveArmor.
Serial ATA 3Gb/s
NVIDIA has also beefed up the Serial ATA performance on the nForce 4. Much like the way Dual Channel memory functions, NVIDIA has implemented two disk controllers into the MCP which gives data two independent paths to travel to and from system memory, which effectively doubles the bandwidth from 1.5 Gb/s to 3 Gb/s. This should give a decent speed boost to transfers to and from the hard drives, however, that is another feature that we will have to wait and see when we get a review sample in our hands.
To better understand this, the drives need to be capable of operating at the 3Gb/s supported by the controllers. The goal of the dual controller setup is to alieviate the bottlenecks that could occur in a single controller environment. In this scenario, if a drive goes through some major disk thrashing, the controller could get bogged down dealing with said drive, slowing down the other drives in the chain. In a dual controller situation, this becomes less of a problem as the other controller will allow the connected drives to operate at full capacity.
NVIDIA has also enhanced their raid configuration with Drive Alert software to allow for simple configuration, maintenance, and monitoring of your raid array and drive status. This feature also notifies you when a drive fails, and even tells you which SATA port that drive is plugged into so that you can easily replace the failed drive without having to go through the hassle of figuring out what drive actually went bad (click here for sample image). In the linked image, you can see that all your SATA connections are displayed, and the problem connection is highlighted in order to assist in troubleshooting.
Morphing is another interesting technology offered by the new controller setup. With the included software, it is possible to expand a striped array via Windows. In the past, if you had a striped array across two disks, the only way to add another drive was to backup, break the array, rebuild and reinstall Windows. Now, simply add the new drive, which Windows will recognize, and use the software to expand the array. The time it takes will vary depending on the speed of your hardware, but it will be much faster and convenient with the morphing technology.
Benchmarks provided by NVIDIA show 4 disk raid arrays performing around two times as well as competitors with sequential 256k reads and writes. Tests done on 2 disk arrays show performance is only slightly above that of competitor products. However, we do not have any of this hardware in hand for testing, so while it looks good on paper, we don't know how this will perform in real-word applications such as gaming and video editing, which are the types of tasks in which performance really matters to the end user.
Native Command Queuing (NCQ) is fully supported by NVIDIA's MCP. For those of you unfamiliar with how NCQ works, this is a feature that can only be found in native SATA hard drives that follow the SATA II specifications. Unlike Legacy Command Queuing (LCQ), NCQ works by allowing a drive to process multiple commands at the same time. These commands can be rescheduled or reordered on a whim, and can also issue new requests while the drive is retrieving data from the previous request. As we've investigated last week, NCQ will not make a huge difference in todays desktop applications, but the potential is there.
It is also important to note here that the value version of the nForce 4 does not have the dual disk controllers, and is only capable of 1.5 Gb/s bandwidth, unlike the 3 Gb/s on the Ultra and SLI versions. While the Ultra and SLI support PATA, SATA-I and SATA-II, the advanced features of the new controllers only work with PATA and SATA-I on the value nForce 4.
NEXT
