MSI’s K9N SLI Platinum motherboard

The nForce 570 SLI doesn’t have quite as many PCI Express lanes as the 590, and it’s missing a buzzword here and there, but you still get SLI support, six Serial ATA RAID ports, dual hardware-accelerated Gigabit Ethernet controllers, and High Definition Audio. You pay a lot less, too. nForce 590 SLI boards sell for close to $200 online, while those based on the 570 SLI can be had for closer to $125.

MSI’s K9N SLI Platinum is currently one of the most affordable nForce 570 SLI boards on the market, and we’ve snagged one to see how it compares with high-end offerings based on the nForce 590 SLI. Can this mid-range board hold its own against competition that costs 50% more? Keep reading; the answer might surprise you.

Board specs
Before diving into the K9N SLI Platinum’s spec sheet, I should take a moment to highlight the main differences between the nForce 590 SLI and 570 SLI chipsets. There aren’t many, so it will really only take a moment.

Perhaps the biggest difference between the 590 and 570 SLI is the number of available PCI Express lanes. The 590 SLI bristles with 46 lanes of connectivity, allowing it to supply a pair of graphics cards with 16 lanes of bandwidth each in SLI. With only 28 lanes of PCI Express at its disposal, the 570 SLI can only afford eight lanes of bandwidth to each graphics card in SLI. From a bandwidth perspective, that’s not a trivial limitation. However, we’ve yet to see dual x16 SLI configurations offer tangible performance benefits over dual x8 setups.

LinkBoost is the other big feature missing from the nForce 570 SLI, but it’s not so much missing as not applicable. Like most chipsets, the nForce 590 SLI is split between north and south bridge components. An interconnect joins the these components, and LinkBoost gives users the option of increasing the speed of that interconnect by 25%. The nForce 570 SLI is a single-chip solution, so there’s no chipset interconnect, making LinkBoost moot. To be fair, LinkBoost also boosts the bandwidth available to the nForce 590 SLI’s PCI Express x16 slots by 25%, but even Nvidia admits that this change has little impact on real world performance.

So there you have it. The major differences between the nForce 590 SLI and 570 SLI chipsets are ones that are unlikely to impact performance even under the best of circumstances. In fact, the similarities between the two chipsets are so striking that we suspect that the nForce 590 SLI’s south bridge component is little more than an nForce 570 SLI chip in disguise. We’ve asked Nvidia if this is the case, and they would only say that the two chips are “very similar.”

The K9N SLI Platinum’s spec sheet is largely defined by the capabilities of the nForce 570 SLI chipset. All six of the board’s 300MB/s Serial ATA ports are fed by the Nvidia chip, which supports RAID 0, 1, 0+1, and 5 arrays. The nForce 570 SLI is also responsible for the board’s dual hardware-accelerated Gigabit Ethernet controllers. These GigE MACs are identical to those in the nForce 590 SLI, and they sport Nvidia’s DualNet GigE teaming and FirstPacket outbound packet prioritization capabilities.

On the audio front, the nForce 570 SLI gives the K9N SLI a High Definition Audio controller that MSI predictably pairs with a Realtek codec chip. The ALC883 is classified as a value HD audio codec, and with a 95dB signal-to-noise ratio, it’s easy to see why. Realtek’s more popular ALC882 codec has a 103dB SNR, although we’ll have to see if our audio quality tests can distinguish any difference between the two. Either way, the crab claims yet another enthusiast motherboard.

With the core logic chipset providing plenty of networking and storage options, MSI doesn’t bother with auxiliary Serial ATA or networking chips on the K9N SLI. The board’s only nod to additional peripherals is VIA’s near-ubiquitous VT6307 Firewire chip. Remember VIA? They used to make enthusiast chipsets.

The board
Contrary to its name, the K9N SLI Platinum is not, in fact, decked out in platinum. Instead, it’s built from a black board that’s dotted with multi-colored ports and slots. The color-coding ruins any hope for a consistent color palette, so the board doesn’t end up looking terribly unique. What’s worse, the manual doesn’t actually refer to any of the ports and slots by their color, so you lose out on both looks and utility.

MSI’s also made a few layout decisions of which we’re not too fond. Among them is the placement of the board’s auxiliary 12V power connector, which creates unnecessary cable clutter around the CPU. We’re also not crazy about having the board’s four-pin Molex connector, which is recommended for use with SLI configurations, so close to the top PCI Express x16 slot. This position makes getting at the plug a pain for those of us with short, stubby fingers.

On a more positive note, MSI neatly tucks the 24-pin primary power connector out of the way along the right edge of the board. The power plug’s surrounded by the board’s ATA and floppy connectors, which are also well placed.

Things are a little tight around the K9N SLI’s socket, where DIMM slots and taller capacitors flank the AM2 heatsink retention mechanism. There’s plenty of room for our Zalman CNPS9500, but those with wider heatsinks may run into clearance issues, especially with taller memory modules like Corsair’s Pro and Xpert DIMMs.

From this angle we also have a good view of the board’s chunky VRM cooler. The cooler bucks recent trends and does not connect to other onboard heatsinks via a heatpipe. Elaborate heatpipe cooling certainly has its place, but it’s apparently not needed on the K9N SLI.

Active chipset cooling isn’t necessary, either. The K9N SLI makes do with a relatively modest passive chipset heatsink that’s about twice the width of a standard cooler, but no taller than a low-profile design. That leaves plenty of clearance for double-wide graphics cards, although longer cards do pose a couple of other problems for the board.

When installed in the board’s top PCI Express slot, longer double-wide cards like the GeForce 7900 GTX will block access to one of the board’s Serial ATA ports. Five SATA ports should still be plenty for most folks, but it seems silly that such a clearance issue exists at all, especially when there appears to be enough available board real estate to avoid it completely.

More annoying than losing access to one Serial ATA port is the limited clearance between the top PCI Express x16 slot and the DIMM retention tabs. There’s just enough room for the retention tabs to flip down, but with longer graphics cards installed, it’s not easy to actually get at the tabs themselves.

Fortunately, it is easy to get at the board’s CMOS reset switch. MSI puts a handy red button next to the CMOS chip, and pressing it’s all you need to do to reset the BIOS. That beats fumbling with a tiny little jumper any day.

The K9N SLI’s slot layout shares some of the blame for the DIMM slot clearance issue. There’s only so much vertical area to work with in the ATX form factor, and MSI squeezes two PCI Express x16, two PCI-E x1, and three standard PCI slots onto the board. This generous array of expansion slots ensures that two PCI and one PCI Express x1 slot are available even when a pair of double-wide graphics cards are running in SLI. The presence of three full PCI slots should delight those disappointed by the dearth of PCI-E x1 peripherals, as well.

Around back, there’s a little something for everyone from legacy-loving luddites to digital audio snobs. PS/2, serial, and parallel ports all make an appearance on the backplane, but we’d gladly trade the latter two for a few extra USB ports. The port cluster only has four USB ports, and although the board comes with a PCI back plate that supplies and additional two ports, users are on their own when it comes to tapping the additional four onboard USB headers.

MSI has taken an interesting approach to supporting digital audio output with the K9N SLI by equipping the board with not one, but two digital outputs. S/PDIF output is supported in coaxial and TOS-Link flavors for maximum compatibility, but it comes at the expense of a digital S/PDIF input. If you’re serious enough about audio to need digital S/PDIF input, chances are you won’t be using integrated audio, anyway.

BIOS and tweaking software
Mid-range boards like the K9N SLI Platinum are prime candidates for overclocking, and MSI has equipped the BIOS with all the bells and whistles you’ll need to push an Athlon 64 beyond its stock speed.

HyperTransport clock speeds are available up to a whopping 425MHz, and CPU voltages go up to 1.7V. There’s no provision for increasing the chipset’s voltage, but memory voltage options are available up to 2.45V.

If you prefer a more dumbed down approach to overclocking, the K9N SLI’s BIOS also has a handful of automatic overclocking options that will increase the processor speed by between 1% and 15%. Most enthusiasts should be comfortable enough manipulating the processor multiplier, HT clock, and memory speed, though.

Speaking of memory, the K9N SLI’s BIOS has all the usual timing options, including control over the DRAM command rate. Unfortunately, there’s nothing MSI can do about the Athlon 64’s limited memory bus divider options. The Athlon 64 calculates the memory bus speed as a factor of its core clock, and for some speed grades, it’s impossible to have DDR2-800 memory running at exactly 800MHz. For example, the closest the Athlon 64 X2 5000+ can bring its memory bus to 800MHz is 742MHz.

At least MSI does a good job with the K9N SLI’s processor fan speed control. Temperature-based automatic fan speed control is supported for three- and four-pin fans, and users can choose between four target CPU temperatures. Ideally, we’d like to be able to define arbitrary target temperatures and fan voltages for each onboard fan header, but few BIOSes go that far.

For those who prefer to do their BIOS tweaking from Windows, MSI supplies a version of its CoreCenter software with the board.

However, unless you crave limited functionality and an ugly, awkward interface, you shouldn’t install it. Instead, skip right to Nvidia’s latest nTune system utility, which sports a much better interface and a broader array of options than MSI’s software.

MSI hasn’t included the necessary BIOS hooks to make all of nTune’s widgets work, but you still get control over loads of memory timings and the HT clock. We’d really like to see MSI update the K9N SLI’s BIOS to allow nTune to manipulate system voltages, which are currently unavailable.

Fortunately, BIOS hooks aren’t necessary for nTune’s custom rules feature. Custom rules allow users to define and activate system settings profiles based on application activity and variables like processor temperature, so you can have separate profiles automatically launch for games, media playback, and so on.

nTune also has a monitoring app that keeps tabs on various system variables, including clock speeds, voltages, and fan activity. The monitoring app can even log variables to a file if you want it to keep an eye on things.

Overall, nTune is a much better monitoring and tweaking utility than what the vast majority of motherboard manufacturers develop in-house and bundle with their boards. Manufacturers would do well to support it fully rather than dedicating additional resources to developing proprietary apps.

Our testing methods
We’re comparing the K9N SLI Platinum’s performance to that of a couple of full-size Socket AM2 motherboards from Asus and Foxconn, and Shuttle’s nForce 570 Ultra-powered XPC SN27P2.

All tests were run at least twice, and their results were averaged, using the following test systems.

Thanks to Corsair for providing us with memory for our testing. 2GB of RAM seems to be the new standard for most folks, and Corsair hooked us up with some of its 1GB DIMMs for testing.

Due to the peculiarities of the Athlon 64’s on-die memory controller, our systems are actually running their memory at 742MHz. The Athlon 64 doesn’t have enough memory dividers to hit 800MHz exactly at every speed grade, and the closest the 5000+ comes without going over is 742MHz.

Also, with the exception of our XPC SN27P2, all of our test systems were powered by OCZ GameXStream 700W power supply units. Thanks to OCZ for providing these units for our use in testing.

We used the following versions of our test applications:

• SiSoft Sandra Standard 2005 SR3
• WorldBench 5.0
• TCD Labs HD Tach v3.01
• Futuremark 3DMark06 Build 1.02
• Far Cry v1.3
• Splinter Cell Chaos Theory v1.05
• RightMark Audio Analyzer 5.5
• RightMark 3D Sound 2.2
• NTttcp
• Cinebench 2003
• Sphinx 3.3
• trq4demo1 demo
• F.E.A.R. 1.04

The test systems’ Windows desktop was set at 1280×1024 in 32-bit color at an 85Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests. Most of the 3D gaming tests used the Medium detail image quality settings, with the exception that the resolution was set to 640×480 in 32-bit color.

All the tests and methods we employed are publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.

Memory performance

The Athlon 64’s on-die memory controller makes for pretty even performance across our memory bandwidth and latency tests. Differences in performance come down to how each mobo maker has chosen to tune the on-die controller, and as you can see, none have found significantly more performance than others.

Motherboards aren’t always happy to accommodate four DIMMs, so we added a couple of memory modules to see how the K9N SLI reacted.

Like every other Athlon 64 motherboard we’ve encountered, you have to raise the DRAM command rate to 2T to get the board to boot with four DIMMs installed. With that done, the K9N SLI’s four-DIMM performance is right in line with the other boards.

WorldBench

The K9N SLI does well in WorldBench, pulling up right behind Shuttle’s XPC SN27P2. Note that in this test, our nForce 570 platforms are just a wee bit faster than those based on the nForce 590 SLI.

Gaming

MSI leads our first round of gaming tests, but not by much.

High resolution gaming performance
Our first round of gaming tests was conducted with more modest in-game detail levels and display resolutions, but we’ve cranked things up for a second round. These tests use high resolutions, high detail levels, and anisotropic filtering and antialiasing. We’ve also tested the Asus, Foxconn, and MSI boards with a pair of 7900 GTXs in SLI. The SN27P2 doesn’t support SLI, so it will have to make do with a single card.

Unfortunately, the K9N SLI isn’t fond of Nvidia’s official 91.31 ForceWare graphics drivers when it comes to SLI. Activating SLI with those drivers produces a garbled screen image followed by a system crash. We’ve actually heard of similar behavior with other nForce 570-based motherboards, so this problem doesn’t appear to be unique to the K9N SLI. In any case, installing the beta 91.33 ForceWare drivers available at Nvidia’s nZone site resolved the issue.

The K9N SLI certainly doesn’t need 16 lanes of bandwidth for each of its PCI Express x16 slots to perform well with SLI. It has no trouble keeping up with—and even beating—boards based on Nvidia’s nForce 590 SLI chipset. Keep in mind that we are using slightly different graphics drivers on the MSI board, though.

Cinebench rendering

Cinebench performance is consistent.

Sphinx speech recognition

Scores in Sphinx are close, too.

Audio performance

Boards with Realtek codecs perform similarly in RightMark 3D Sound, but it’s the Asus M2N32-SLI Deluxe that consumes the fewest CPU cycles in this test. The Asus board uses a codec chip from Analog Devices, suggesting that maybe the crab isn’t all that it’s cracked up to be.

Audio quality
We used an M-Audio Revolution 7.1 sound card for recording in RightMark’s audio quality tests. Analog output ports were used on all systems. To keep things simple, I’ve translated RightMark’s word-based quality scale to numbers. Higher scores reflect better audio quality, and the scale tops out at 6, which corresponds to an “Excellent” rating in RightMark.

The K9N SLI’s performance in RightMark Audio Analyzer is unremarkable, but certainly not as poor as one might expect from a “value” codec chip like the ALC883. That’s not to say that it sounds good; it just doesn’t sound much worse than other onboard audio implementations, which are mediocre at best.

ATA performance
ATA performance was tested with a Seagate Barracuda 7200.7 ATA/133 hard drive using HD Tach 3.01’s 8MB zone setting.

The K9N SLI doesn’t differentiate itself in our ATA performance tests.

Serial ATA performance
Moving to Serial ATA, we tested performance with a Western Digital Raptor WD360GD SATA hard drive. Again, we used HD Tach 3.01’s 8MB zone test.

Serial ATA performance is consistent across the board.

USB performance
Our USB transfer speed tests were conducted with a USB 2.0/Firewire external hard drive enclosure connected to a 7200RPM Seagate Barracuda 7200.7 hard drive. We tested with HD Tach 3.01’s 8MB zone setting.

The K9N SLI’s USB performance is right where it should be.

Firewire performance
Our Firewire transfer speed tests were conducted with the same external enclosure and hard drive as our USB transfer speed tests. It’s just a 1394a Firewire enclosure, so it won’t benefit from the higher speeds supported by the C51XEM2AA’s 1394b Firewire chip.

Windows XP doesn’t fully support 1394b Firewire, anyway. In fact, with Service Pack 2, XP throttles the performance of 1394b Firewire devices to below 1394a speeds. This issue is detailed in this Microsoft support document, which provides a patch for SP2 users. We’ve tested both of the C51XEM2AA’s Firewire ports with and without the patch installed.

MSI avoids Windows XP’s 1394b woes by sticking with Firewire 400, and that works out pretty well for the K9N SLI.

Ethernet performance
We evaluated Ethernet performance using the NTttcp tool from Microsoft’s Windows DDK. The docs say this program “provides the customer with a multi-threaded, asynchronous performance benchmark for measuring achievable data transfer rate.”

We used the following command line options on the server machine:

ntttcps -m 4,0,192.168.1.25 -a

..and the same basic thing on each of our test systems acting as clients:

ntttcpr -m 4,0,192.168.1.25 -a

Our server was a Windows XP Pro system based on Asus’ P5WD2 Premium motherboard with a Pentium 4 3.4GHz Extreme Edition (800MHz front-side bus, Hyper-Threading enabled) and PCI Express-attached Gigabit Ethernet. A crossover CAT6 cable was used to connect the server to each system.

The boards were tested with jumbo frames disabled.

Nvidia appears to have left its ActiveArmor acceleration woes in the dust with its latest nForce 500 series chipsets, and the K9N SLI performs exceptionally well in our Ethernet throughput and CPU utilization tests. However, Nvidia’s hardware TCP/IP acceleration isn’t compatible with software firewalls, so you’ll have to choose between the two.

Power consumption
We measured system power consumption, sans monitor and speakers, at the wall outlet using a Watts Up power meter. Power consumption was measured at idle and under a load consisting of a multi-threaded Cinebench 2003 render running in parallel with the “rthdribl” high dynamic range lighting demo.

Our Asus, Foxconn, and MSI motherboards were tested with an OCZ GameXStream 700W power supply, while the SN27P2 used its own integrated Silent X 400W unit.

The nForce 590 SLI is a rather power-hungry chipset, and the nForce 570 SLI isn’t all that much more frugal. Perhaps much of the 590 SLI’s power drain is tied to its 570 SLI south bridge component.

Overclocking
The Athlon 64’s memory divider mechanism makes memory overclocking complicated at best and infuriating at worst. To avoid driving ourselves to the brink of frustrated insanity, we’ve limited the bulk of our overclocking tests to seeking out the highest stable HyperTransport clock for each board.

For our HyperTransport overclocking tests, we backed off on our CPU, memory, and HyperTransport processor link multipliers to remove them as potential problems. Next, we started turning up the HyperTransport clock, testing for stability along the way.

We made it up to a 230MHz HyperTransport clock with little complaint from the K9N SLI, but she would go no higher. 240MHz wouldn’t post, and without control over the chipset or HT voltage, there wasn’t much we could do to coax more from the board.

As always, keep in mind that overclocking success is never guaranteed. Results can depend as much on the mix of system components as they can on the characteristics of individual samples, and your mileage may vary.

Conclusions
If the results of our performance tests make one thing clear, it’s that the nForce 570 SLI chipset has no problem keeping up with its more expensive sibling. The loss of true dual 16-lane SLI doesn’t have an impact on the performance of a pair of GeForce 7900 GTX graphics cards, and LinkBoost isn’t missed. In fact, there are a few instances where the K9N SLI Platinum is actually a little faster than nForce 590 SLI boards from Asus and Foxconn. Considering that those boards cost at least 50% more, the K9N SLI starts to look pretty appealing.

For the K9N SLI Platinum to stay appealing, you have to be willing to look past a few layout issues. Clearance problems between longer graphics cards and Serial ATA ports and DIMM slot retention tabs aren’t uncommon, but they’re still annoying, and can certainly be avoided.

Layout issues aside, the K9N SLI’s only other failing is in the overclocking department. We don’t normally ding boards for not performing well beyond stock speeds because overclocking success can vary from board to board. Still, a measly 30MHz overclock doesn’t instill much confidence in the Platinum’s potential.

Of course, if you don’t plan to overclock, the K9N SLI Platinum is still a fine board, especially when you consider its affordable $125 street price. Building around its layout quirks may take a little extra time, but that doesn’t diminish the solid feature set and excellent performance.