Until AMD dramatically lowers its Athlon 64 prices or offers lower speed grades, the Socket A platform will remain quite popular among enthusiasts looking for the best bang for their buck. But what chipset will enthusiasts flock to for their Socket A needs? For some time, NVIDIA’s nForce2 has been the Socket A chipset to have. However, the nForce2 is getting a little long in the tooth, and its lack of integrated Serial ATA support becomes more noticeable with each passing day.
The KT600, VIA’s latest Socket A chipset, just happens to have integrated south bridge Serial ATA. But that’s not all. The KT600 chipset officially supports a 400MHz front-side bus and features a tweaked memory controller that promises to wring every last drop of performance from its single memory channel. KT600 boards are making their way onto the market, and we’ve gathered eight of them for a massive mobo round-up. Just to keep things spicy, we’ve also thrown an nForce2 Ultra 400 board into the mix to see how VIA’s latest Socket A creation stacks up against the well-entrenched competition.
Can VIA’s KT600 run with or even beat NVIDIA’s venerable nForce2? Which KT600 board is right for you? Read on as we compare boards from ABIT, ASUS, AZZA, EPoX, FIC, MSI, SOLTEK, and SOYO to find out.
![]() VIA’s KT600 north bridge |
![]() … and the VT8237 south bridge |
The chipset
VIA’s KT600 chipset isn’t a radical departure from the KT400A, but VIA has managed a nip here and a tuck there to improve performance and to offer more features. The KT600 north bridge is primarily responsible for improved performance, while the chipset’s VT8237 south bridge has all the new features. Before we cover all that’s new and interesting in the KT600, let’s see how it’s all laid out in VIA’s lovely block diagram:

Source: VIA
The KT600 block diagram nicely segments different elements of the KT600, which makes the chipset a lot easier for me to explain. Here are some of the highlights, starting from the top.
- A 400MHz front-side bus – Despite its name, VIA’s KT400A chipset never officially supported a 400MHz system bus. KT400 boards could often be overclocked, but not every board was stable at 400MHz. The KT600 officially supports 400MHz front-side bus speeds and provides all the necessary memory, AGP, and PCI dividers to keep the rest of the system running in-spec with AMD’s fastest Athlon XPs.
- Single-channel memory controller – VIA has tweaked the KT600’s memory controller to squeeze as much performance as possible from a single channel of DDR400 memory. The KT600 won’t be able to match the raw theoretical peak bandwidth of dual-channel DDR400 memory controllers, but it will have the distinct advantage of being competitive with only a single DIMM installed. Dual-channel solutions like the nForce2 require 2 DIMMs for optimal performance. Another often-overlooked benefit of single-channel memory controllers is how easy it is to expand a system’s existing memory configurationDIMMs can be added one at a time rather than in pairs.
- Serial ATA – The VT8237 south bridge has native support for up to four Serial ATA devices, though accessing two of those devices requires an external PHY chip. Because it’s located right on the south bridge, the VT8237’s Serial ATA controller doesn’t have to compete for limited PCI bus bandwidth with other devices. Serial ATA drives have full access to the KT600’s 533MB/sec 8X V-Link interconnect.
- V-RAID – VIA is serving up software RAID 0, 1, 0+1, and JBOD support for the KT600’s VT8237’s Serial ATA controller, which should delight mirroring and striping fanatics alike. Because V-RAID only works with south bridge-connected hard drives, it’s not saddled with PCI bus bandwidth limitations that can plague third-party RAID chips. Using software built right into the driver, users can create and delete RAID arrays on the fly, designate spare disks, and even do array repairs. Unfortunately, V-RAID only works with Serial ATA drive connected to the VT8237 south bridge; “parallel” ATA RAID isn’t supported. Because V-RAID only supports Serial ATA drives, RAID 0+1 support is limited to KT600 boards that expose the VT8237’s extra two Serial ATA ports via an auxiliary PHY chip.
The KT600 chipset also inherits USB 2.0, six-channel Vinyl Audio, 10/100 Fast Ethernet, and 8X V-Link technology from the KT400A. Here’s a quick synopsis of all the KT600’s features:
| CPU support | Socket A-based AMD Athlon XP processors |
| Chipset | VIA KT600 |
| Interconnect | 8X V-Link (533MB/sec) |
| AGP | 4X/8X AGP 8X (1.5V only) |
| Memory | Maximum of 4GB of DDR400/333/266 SDRAM |
| Storage I/O | 2 channels ATA/133 |
| Serial ATA | 2 channels Serial ATA 150 |
| RAID | RAID 0, 1, and 0+1 for Serial ATA drives |
| USB | 8 USB 2.0/1.1 ports |
| Audio | Vinyl “Six-TRAC” 6-channel AC’97 audio |
| Ethernet | 10/100 Fast Ethernet |
We’ve rounded up eight KT600 boards for testing, and despite the fact that they’re all based on the same chipset, there’s actually quite a bit of variation between the boards. Here’s a quick cheat sheet on the main differences:
| PCI slots | IDE RAID | SATA RAID | Networking | Audio Codec | Digital audio ports | Auxiliary power | HSF mounting holes | North bridge cooling | USB 2.0 ports | Firewire ports | Price | |
| ABIT KV7 | 5 | None | None | VIA VT8237 | VIA VT1616 | TOSLink S/PDIF output | Yes | Yes | Active | 4 backplate, 4 auxiliary | None | $74 |
| ASUS A7V600 | 6 | None | None | 3Com Marvell 950-MV00 | Analog Devices AD1980 | Coaxial S/PDIF output | No | Yes | Passive | 4 backplate, 4 auxiliary | None | $75 |
| AZZA KT600 ALX | 6 | None | None | VIA VT8237 | Realtek ALC650 | None | No | No | Passive | 2 backplate, 6 auxiliary | None | NA |
| EPoX EP-8KRA2+ | 5 | Highpoint HPT372 | None | VIA VT8237 | Realtek ALC650 | None | No | No | Passive | 4 backplate, 4 auxiliary | 3 auxiliary | $88 |
| FIC KT-600 PRO | 5 | None | None | VIA VT8237 | Realtek ALC655 | None | Yes | No | Passive | 4 backplate, 4 auxiliary | 3 auxiliary | NA |
| MSI KT6 Delta-FIS2R | 5 | Promise PDC20378 | Broadcom BCM5788 | C-Media CMI9739 | Coaxial, TOSLink S/PDIF outputs | Yes | No | Active | 6 backplate,2 auxiliary | 3 auxiliary | $128 | |
| SOLTEK SL-KT600R | 6 | None | None | VIA VT8237 | VIA VT1616 | None | Yes | No | Passive | 2 backplate, 6 auxiliary | None | $78 |
| SOYO SY-KT600 Dragon Ultra | 5 | None | Silicon Image Sil3112 | VIA VT8237 | C-Media CMI8738* | Coaxial, TOSLink S/PDIF inputs and outputs | No | Yes | Active | 4 backplate, 4 auxiliary** | 1 backplate, 2 auxiliary | $146 |
*The C-Media CMI8738 is a full blown audio chip with an integrated codec
**The SY-KT600 Dragon Ultra’s 3.5″ memory card reader consumes two of the board’s auxiliary USB ports I’m going to explore the layout of each board in a moment, but before I get into that, it’s worth taking a look at some of the different chips used on each board. These chips help differentiate some boards from their competition by augmenting the KT600 chipset’s storage, networking, audio, and even connectivity features.
![]() Highpoint’s HPT372 IDE RAID chip |
![]() Promise’s PDC20378 S/PATA RAID chip |
Silicon Image’s Sil3112 SATA RAID chip |
On the storage front, EPoX, MSI, and SOYO all incorporate third-party RAID controllers. EPoX is the only manufacturer to include an IDE-only RAID chip, Highpoint’s HPT372, which offers support for single “parallel” ATA drives in addition to RAID 0, 1, and 0+1 arrays. This IDE RAID option nicely makes up for the fact that VIA’s V-RAID is Serial ATA only. Instead of complimenting V-RAID with IDE RAID, SOYO chose to equip its KT600 board an additional Serial ATA RAID option. The SY-KT600 Dragon Ultra uses Silicon Image’s Sil 3112 Serial ATA RAID controller to power a pair of Serial ATA ports for RAID 0 or RAID 1 arrays.
Instead of choosing between Serial ATA and IDE RAID, MSI employs Promise’s PDC20378 RAID chip to support both on the KT6 Delta-FIS2R. The PDC20378 supports RAID 0, 1, and 0+1 arrays across not only two Serial ATA drives, but also two “parallel” ATA drives. Though the chip supports Serial ATA drives on individual channels, “parallel” ATA drives must share a single IDE channel, which will invariably degrade performance in arrays with two PATA drives. Still, the ability to span RAID arrays across multiple drive types is pretty slick.
![]() 3Com’s Marvell 940-MV00 GigE chip |
![]() Broadcom’s BCM5788 GigE chip |
Though VIA’s VT8237 south bridge’s integrated 10/100 Fast Ethernet support should be fast enough for most home and business networks, ASUS and MSI pimp up their KT600 boards with a little extra networking love in the form of Gigabit Ethernet. ASUS uses 3Com’s Marvell 940-MV00 Gigabit chip, while MSI integrates Broadcom’s BCM5788. Unfortunately, both chips are forced to share PCI bandwidth with other add-in cards, and in the KT6 Delta-FIS2R’s case, with an integrated Serial ATA RAID controller. It’s unlikely that either GigE implementation will have performance approaching that of PCI-X or CSA-based Gigabit Ethernet solutions. So far, only a handful of boards have offered different storage and networking chips. However, nearly every board in this comparison deals with audio a little differently. Check out this huge selection of audio and codec chips:
![]() Analog Devices’ AD1980 codec |
![]() C-Media’s CMI8738 audio chip |
![]() C-Media’s CMI9739a codec |
![]() Realtek’s ALC650 codec |
![]() Realtek’s ALC655 codec |
![]() VIA’s VT1616 codec |
Yikes! That’s a lot of chips. Let’s deal with the odd man out first. SOYO’s SY-KT600 Dragon Ultra is the only board in this comparison to forgo VIA’s integrated VT8237 south bridge audio in favor of a third-party audio chip. The Dragon Ultra uses C-Media’s CMI8738, which packs a six-channel audio controller and an accompanying codec into a single chip. The CMI8738 supports 16-bit, 48kHz audio recording and playback and is actually used in a number of low-end discrete sound cards.
Just because the rest of the KT600 boards we’re looking we’re looking at use VIA’s integrated south bridge audio doesn’t mean that all the implementations are identical. Nearly all of the boards route the VT8237’s audio streams through a different codec chip. VIA actually makes a VT1616 codec chip to accompany the VT8237 south bridge, but only ABIT’s KV7 and SOLTEK’s SL-KT600R use VIA’s codec. The VT1616 is a part of VIA’s “Vinyl Audio” brand and supports 18-bit audio recording and playback at 48kHz.
Of course, no motherboard audio round-up would be complete without a little ALC action from Realtek. I’ve seen Realtek’s ALC650 codec chip on more motherboards than I can count, and the popular chip is featured on both the AZZA and EPoX boards. FIC uses a close relative of the ALC650, the ALC655, on its KT-600 PRO. The ALC655 is actually a step down from the ALC650 in terms of supported sampling rates and resolutions. The ALC650 supports 18-bit recording and 20-bit playback at 48kHz, while the ALC655 only supports 16-bit audio up to 48kHz.
Rather than use popular codec chips from Realtek or VIA, ASUS and MSI opt for less common chips from Analog Devices and C-Media. ASUS’ A7V600 uses Analog Devices’ AD1980 codec, which supports 16-bit audio recording and 20-bit playback at up to 96kHz. The VT8237 can’t take advantage of the AD1980’s support for 96kHz sampling rates, but the codec should sound great playing back Wham MP3s. MSI’s piece de resistance is C-Media’s CMI9739a codec, which supports 18-bit recording and 20-bit audio playback at 48kHz.
Each board’s audio implementation supports six analog output channels, but as we’ll see later, performance and output quality varies quite a bit. The boards from ABIT, ASUS, MSI, and SOYO also support digital S/PDIF outputs, and SOYO’s board even has digital input ports.
![]() VIA’s VT6307 Firewire chip |
We’re almost done with chips, but not quite. A handful of the KT600 boards we’re looking at feature VIA’s VT6307 Firewire chip for IEEE 1394 connectivity. The VT6307 supports up to three Firewire devices and is implemented in EPoX’s EP-9KRA2+, FIC’s KT-600 PRO, MSI’s KT6 Delta-FIS2R, and SOYO’s SY-KT600 Dragon Ultra. After seeing all those different codec chips, I half expected to see a couple of different Firewire solutions, but the VT6307 seems to be a popular choice among mainboard manufacturers.
So far, we’ve seen quite a bit of variety in the chips used on our KT600 boards, but there are even more differences in the board layouts.
|
Good things come in small packages
ABIT’s been making enthusiast-oriented motherboards for a long time, and they’re currently enjoying renewed success with a strong lineup of new boards. For the KT600, ABIT is rolling out the KV7, a surprisingly small board that still manages to exploit everything VIA’s new chipset has to offer. The KV7 sports ABIT’s new orangish-red color scheme, which is refreshing change from many of the multicolored neon monstrosities on the market. Check it out:
The KV7’s layout is a little tighter than other KT600 boards, but since the KV7 is also a couple of inches shorter than the competition, I think it’s a reasonable compromise. Five full PCI slots grace the board, which also has an auxiliary four-pin power connector. Auxiliary power connectors aren’t required by AMD or VIA, but the extra juice should help the KV7 maintain stability when pushed beyond its limits.
Unfortunately, though the KV7’s four-pin power connector is smartly placed near the top edge of the board, the main power connector is halfway down the board and close enough to the CPU socket to create cable clutter that could impede air flow.
The area around the KV7’s CPU socket is a little tight, but the board does have four mounting holes for larger heat sink mounting assemblies. The extra holes are no longer a part of AMD’s Socket A spec, but it’s nice to see ABIT include them for those looking to use more secure heat sink mounting mechanisms and larger, heavier heat sinks.
Despite its tighter layout, there’s enough room around the KV7’s AGP and DIMM slots to allow memory modules to be removed while a graphics card is installed. Considering the fact that many larger boards have problems with AGP/DIMM slot clearance, ABIT gets extra props for doing it right on a smaller board.
ABIT is the first manufacturer I’ve seen mounting IDE ports facing outward along the outside edge of a motherboard, and I actually quite like the change. Having IDE ports flipped 90 degrees and placed right along the edge of the board helps make routing and folding cumbersome ribbon cables a little easier.
I’m not a huge fan of the KV7’s active north bridge cooler. Badum-ching. Noise isn’t an issue, because the fan is whisper silent, but ABIT could easily get away with a larger passive cooler that’s immune to potential fan failure.
The KV7 serves up a packed port cluster that offers a full suite of audio input and output ports, including a digital S/PDIF port. The KV7’s five analog output ports lets users take advantage of all the board’s analog input and output ports without having to swap cables.
It may seem like a minor point to touch on, but it’s worth mentioning that the KV7 uses Rubycon capacitors exclusively. Rubycon’s high-end capacitors were immune to the cap-busting problems that plagued many enthusiast-oriented motherboards a few years ago, and it’s encouraging to see ABIT using them throughout on the KV7. The fact ABIT is putting high-end capacitors on an inexpensive board like the KV7 is a refreshing change from more typical cost-cutting attitudes.
|
A reference board with a twist
Being the largest motherboard manufacturer in the world means ASUS makes boards based on nearly every chipset under the sun. ASUS doesn’t often do anything wild or daring, but the company’s reputation for producing stable, reliable, and fast boards has made them a favorite among enthusiasts and PC manufacturers alike.
The KT600-powered A7V600’s drab brown finish doesn’t make a good first impression, but a closer look at the board reveals a few unique twists that give the board an edge.
The A7V600 is a massive board, so there’s lots of room for six PCI slots and all sorts of chips and peripheral ports. The board lacks an auxiliary four-pin power connector, but at least its primary power plug is positioned nicely along one edge.
Considering how expansive the board is, it would be nice to see a little more room around the A7V600’s CPU socket. As it is, memory modules need to be removed to easily access the retention tabs for most heat sinks. On a more positive note, the board does include four mounting holes for alternative heat sink mounting mechanisms.
The A7V600 has less AGP/DIMM clearance than any motherboard I’ve ever seen. Part of the blame lies with the fact that the board has to accommodate six full PCI slots, but I’ve seen many boards provide more clearance, even with six PCI slots.
ASUS at least uses the A7V600’s available real estate to position the IDE and Serial ATA ports mid-way up the board’s edge.
The A7V600 uses a massive passive heat sink to keep the KT600 north bridge nice and cool. The heat sink won’t generate any noise, and perhaps more importantly, it won’t be prone to a fan failure that could potentially fry a north bridge chip.
The A7V600’s port cluster is standard, for the most part. The board features three analog audio ports, which means that line-in and mic inputs share jacks with rear and center output channels. A digital S/PDIF output also graces the rear port cluster for those looking to completely bypass the board’s DAC.
Though it’s not bundled directly with the board, the A7V600 has a special expansion header designed to work with ASUS’ Wi-Fi@Home card. Wi-Fi@Home is a wireless network card and Wi-Fi access point rolled into one, which is pretty cool, even if it’s not a freebie.
|
Who?
Maybe I’ve been living under a rock for the last few years, but I’ve honestly never heard of AZZA. After browsing through the company’s North American web site, which doesn’t even list the KT600 ALX, I wasn’t too optimistic about the board’s prospects. The fact that no listings for the KT600 ALX are available on Pricewatch doesn’t say much for the board’s current availability, either.
Still, I’ve always been a fan of underdogs. If anything, the KT600 ALX’s black board suggests this could be the dark horse in our comparison, so I won’t write it off just yet.
At first glance, nothing about the KT600 ALX really jumps out. The board has six PCI slots and lacks an auxiliary power connector, which makes it similar to the A7V600. However, the KT600 ALX’s primary power connector is positioned far from the top edge of the board, so users will have to deal with a little power cord cable clutter.
The KT600 ALX’s socket orientation makes heat sink removal in tight cases a bit difficult, but there’s lots of room around the socket for larger heat sinks. The board also lacks mounting holes for beefier heat sink retention systems.
Despite having six PCI slots, there’s enough room between the KT600 ALX’s AGP and DIMM slots to swap memory modules without having to yank their graphics cards. Those with longer graphics cards (gargantuan GeForce4 Ti 4600s come to mind) may run into a few problems when fiddling with the DIMM slot retention tabs, but there are no clearance problems once everything is installed.
Surprisingly, the KT600 ALX is the first motherboard I’ve seen in ages that doesn’t incorporate an AGP card retention system of any kind. Those who move their rigs to and from LAN parties on a regular basis or run heavier graphics cards should think twice about using the KT600 ALX without some sort of auxiliary AGP card retention mechanism.
AZZA’s placement of the KT600 ALX’s Serial ATA and IDE ports is similar ASUS’ A7V600, which puts the ports mid-way up the board along one edge. Having the ports halfway up the board makes it easier to reach optical drives high up in full tower cases and also keeps cable clutter away from the bottom of the board.
The KT600 ALX uses a passive heat sink to keep its north bridge chip cool, which is just fine by me. Passive heat sinks may not have the sex appeal of funky fan-driven cooling systems, but with enough surface area, passive sinks get the job done nicely.
AZZA doesn’t take any chances with the KT600 ALX’s port cluster, which only gives access to two of the board’s eight USB 2.0 ports. The remaining USB ports must be accessed via PCI back plate headers, but that blocks access to PCI slots. With only three analog audio ports, the KT600 ALX’s center and rear outputs share ports with its line-in and microphone inputs.
|
Making being green look easy
Though EPoX makes plenty of enthusiast-oriented motherboards, we’ve actually never reviewed one here at TR, so I was anxious to get my hands on the EP-8KRA2+. If I had to succinctly describe the EP-8KRA2+, I’d have to call it old school. Not only does the board bundle in plain old IDE RAID, it’s also one of the few KT600 boards to dress in a classic shade of Printed Circuit Board Green. Green may not be the flashiest color for modding enthusiasts looking to show something off through a case window, but it’s remarkably distinctive in the sea of multicolored boards currently on the market.
The EP-8KRA2+ is laid out on a full-size ATX board with plenty of room for extra IDE ports and six PCI slots. The board lacks an auxiliary power connector, and its single power plug is located a little too close to the CPU socket for my picky cable routing tastes, but overall things are spaced out nicely.
There’s plenty of room around the EP-8KRA2+’s CPU socket for larger heat sinks, but no retention mechanism holes to help keep bigger, heavier cooling systems secure. Although Athlon XP processors feature internal thermistors, the EP-8KRA2+’s CPU socket sports its own temperature sensor just in case.
Despite the board’s six PCI slots, there’s enough room between the EP-8KRA2+’s AGP and DIMM slots to facilitate easy DIMM swapping. Only those running longer graphics cards should have to pull their graphics card to swap memory modules.
The EP-8KRA2+’s four IDE ports are nicely lined up along the edge of the board, and the Serial ATA ports aren’t far away. With all four IDE ports in use, things will get a little crowded around the bottom of the board, but that’s what Zip Ties are for.
A passive north bridge heat sink keeps the EP-8KRA2+’s KT600 chip running cool, silent, and free of worry over potential fan failures. I’ve actually had a couple of north bridge fans die on me over the years, and while the failures have never released the magic smoke or fried any system components, they were detrimental to system stability under load.
The EP-8KRA2+’s port cluster is pretty standard fare. With only three analog audio ports, rear and center outputs must share ports with line-in and mic inputs.
Like many of EPoX’s motherboards, the EP-8KRA2+ has a two-digit post code display that makes troubleshooting a breeze. Personally, I’ve never enjoyed deciphering BIOS beep codes to get to the bottom of motherboard boot problems, so the post code display really does it for me. This feature probably saves EPoX’s tech support team plenty of time, too. I can imagine the joy tech support jockeys must feel knowing they won’t have to ask troubled users to hold the phone up to beeping motherboards that refuse to boot.
|
Capitalization’s all the rage
FIC’s take on VIA’s KT600 chipset is the appropriately named KT-600 PRO. The KT-600 PRO isn’t a wild departure from the KT600’s core feature set, but that’s not necessarily a bad thing. Unfortunately, I can’t find any KT-600 PROs for sale online, but I’d expect them to be among the cheaper KT600 options once they’re finally available.
Rather than dressing up the KT-600 PRO with a wild splash of color, FIC opts for a more conservative look for the board. Color is used sparingly to identify different expansion slots, and though I don’t find the palette particularly attractive, my fashion sense is questionable at best.
FIC gets high marks for putting the KT-600 PRO’s primary power connector along one edge of the board, but the auxiliary power plug is situated directly below the CPU socket where cabling can potentially create clutter and impede air flow. Since only half of the KT600 boards we’re looking at today support any auxiliary power at all, I can’t complain too much about FIC’s plug placement.
The KT-600 PRO’s socket region is tight but clean. Memory modules in the nearest DIMM slot can interfere with some heat sink retention clips, which is a minor annoyance, but one that’s easily rectified by temporarily removing the DIMM during installation. The board also lacks mounting holes for alternative heat sink retention mechanisms, but since the holes are no longer a part of AMD’s socket spec, it’s hard to fault FIC too much for not including them.
With only five PCI slots, the KT-600 PRO easily has enough clearance between its AGP and DIMM slots. The board uses an AGP card retention mechanism that’s quite a bit simpler than the hinged clips found on most boards, and in my experience, easier to use as well.
Like most of the KT600 boards we’re looking at, the KT-600 PRO’s Serial ATA and IDE ports are located along the edge of the board where they should create the least amount of cable clutter. The floppy port is buried all the way at the bottom of the board, but since floppy drives are rarely needed these days, it’s not a big deal.
FIC uses a tiny heat sink on the KT-600 PRO’s north bridge, and I have some serious reservations about how well it can keep the chip cool. Passive heat sinks on the other KT600 boards we’re looking at have at least three times the surface area of FIC’s diminutive north bridge cooler, which gets quite hot even when the board is running on an open workbench.
The KT-600 PRO’s port cluster offers a full array of standard ports, four USB ports, and a trio of audio jacks. With only three audio ports, the board’s center and rear output channels must share plugs with the line-in and mic inputs, but that’s probably not going to be a big deal for most users.
|
Painting the town red
Thus far, we haven’t seen any particularly daring or sexy KT600 implementations, but that doesn’t mean that the chipset can’t have a wild side. MSI is tossing caution to the wind with its KT6 Delta-FIS2R, which squarely targets the enthusiast market with an array of high-end extras.
If the KT6 Delta-FIS2R’s laundry list of extra features doesn’t make you hot, its brilliant red color scheme will. I’m not crazy about the pastel green DIMM slots, but the board still looks fast.
Despite being peppered with expansion ports and extra chips, the KT6 Delta-FIS2R’s layout is remarkably clean and tidy. The board’s primary power connector is about as close to the top edge as one can get, but the auxiliary plug’s position isn’t ideal.
A single annoying capacitor keeps the KT6 Delta-FIS2R from having a spacious CPU socket area, but it’s not a huge deal in the grand scheme of things. A bigger problem could be the socket’s orientation, which puts the heat sink retention clip along the top edge of the board. In tight cases with little or no vertical clearance, the board may have to be removed to gain access to the heat sink retention mechanism.
Despite its crowded layout, there’s plenty of room between the KT6 Delta-FIS2R’s AGP and DIMM slots. Memory modules can be removed and installed without having to pull the graphics card. Though a couple of capacitors look imposing to the right of the AGP slot, they won’t get in the way, even with NVIDIA’s notoriously massive GeForce4 Ti 4600 and FX 5800 Ultra graphics cards.
If we count its floppy port, the KT6 Delta-FIS2R serves up a total of eight storage-related expansion portspretty impressive for a board that costs less than $130. IDE and Serial ATA ports are spaced out nicely along one edge of the board, but the Promise controller’s IDE port is a little low to reach optical drives located at the top of full-tower cases.
MSI goes with active cooling for the KT6 Delta-FIS2R’s north bridge chip, and the variable speed fan is pretty slick. However, I’m still not a huge fan of active north bridge cooling, as I’ve mentioned.
The KT6 Delta-FIS2R’s port cluster is full of USB ports, but I can’t help but wish that MSI included a couple of those ports on some sort of front-panel drive bay insert. Chances are, if you’re using all six USB ports, it would be more convenient to have at least a couple of those ports at the front rather than at the rear of a case.
The KT6 Delta-FIS2R also comes with an auxiliary audio port header to fill out the board’s analog outputs and add a couple of digital ports. The board doesn’t have to share analog output channels with its line-in and mic inputs, and users also have access to two S/PDIF digital output options.
Thus far, the KT6 Delta-FIS2R has had no problem differentiating itself from the pack, but MSI has one more trick up its sleeve: the Core Cell chip.
So what exactly is the Core Cell? According to MSI, the chip is a diagnostic tool that tunes the board to an “optimal” state by balancing performance with power consumption and noise. The Core Cell manipulates the speeds of the north bridge and CPU fans in an attempt to keep temperatures constant and noise to a minimum. In testing, the KT6 Delta-FIS2R wasn’t noticeably quieter than any of the other KT600 boards, but the benchmarking sweatshop is usually purring with the sound of multiple systems running full bore, so minute differences in noise levels are hard to notice.
MSI also claims the Core Cell can improve overclocking potential and stability over competing designs. We’ll see how the KT6 Delta-FIS2R fares with overclocking a little later in this comparison.
|
Fresh out of witty comments
SOLTEK’s small form factor systems have been generating a bit of buzz lately, but the company also has a KT600 board to show off. The SL-KT600R’s most striking feature is its lack of integrated Ethernet, which is a surprising omission. The SL-KT600 line actually comes in a number of different flavors that mix and match peripherals, but it boggles my mind why SOLTEK would even consider building a board that didn’t take advantage of the VT8237 south bridge’s “free” 10/100 Fast Ethernet support.
The SL-KT600R’s color scheme also sets it apart from the crowd.
The SL-KT600R packs six PCI slots and an auxiliary ATX12V power connector. Unfortunately, both power plugs are located far from the board’s top edge, which makes for a minor cable routing nuisance. At least users will be able to bind the two motherboard power cables together to reduce slightly reduce clutter.
There’s a reasonable amount of room around the SL-KT600R’s CPU socket, but the board doesn’t have mounting holes for custom heat sink retention systems.
Longer graphics cards can create clearance problems for the board’s DIMM slot tabs.
All the SL-KT600R’s IDE and Serial ATA ports are clustered together towards the edge of the board, but thin, flexible Serial ATA cables ensure that the area doesn’t get too messy.
SOLTEK keeps the SL-KT600R’s north bridge cool with a tall, passive heat sink that’s packed with fins to radiate heat.
Without an RJ-45 jack, the SL-KT600R’s port cluster looks pretty bare. Three analog audio ports force sharing between line-in and mic inputs and rear/center output channels, but the board does have a game port if you’re looking to run an older joystick or game pad.
|
Saving the best for last?
The SY-KT600 Dragon Ultra is the last of the KT600 boards we’ll be looking at, and though you’re probably bored to tears of DIMM tab clearance issues and north bridge heat sinks, SOYO’s KT600 offering is distinctive enough to spice things up. The SY-KT600 Dragon Ultra is also the most expensive KT600 board we’re looking at today, and SOYO has packed it with enough extras to justify its higher price.
The first thing that jumps out about the SY-KT600 Dragon Ultra is the board’s silver and purple color scheme. While the look is likely a little garish for some, it’s certainly unique, and will nicely match the interior of an aluminum case.
SOYO has a lot going on with the SY-KT600 Dragon Ultra, but the board’s layout is still neat and relatively spacious. With all the board’s extras, it’s a shame SOYO didn’t include an auxiliary power connector. The primary power plug could be better placed closer to the upper edge of the board, too.
The SY-KT600 Dragon Ultra’s CPU socket has mounting holes for alternate CPU retention mechanisms, but the board’s DIMM slots are a little too close to the socket for comfort. When filled, the board’s first DIMM slot complicates access to traditional heat sink retention clips. I also worry that memory modules mounted in the closest DIMM slot could pick up additional heat radiated from the CPU heat sink.
With only five PCI slots, the SY-KT600 Dragon Ultra has plenty of clearance between its AGP slot and DIMM retention tabs. The SY-KT600 Dragon Ultra has the distinction of being the only KT600 board I’m aware of with an AGP Pro slot, though it’s unlikely workstation users will be flocking to VIA’s latest Socket A platform. Unfortunately, the Dragon Ultra doesn’t have an AGP card retention mech, which makes the board less suitable for gamers who regularly drag their systems to and from LAN parties.
SOYO positions the SY-KT600 Dragon Ultra’s multiple storage ports along the edge of the board, and everything is nicely spaced. The board’s four Serial ATA ports are quite close to each other, but thanks to flexible Serial ATA cables, it’s not a big deal.
The SY-KT600 Dragon Ultra’s north bridge heat sink relies on active cooling to keep the KT600 running smoothly, but by now you’ve no doubt noted my preference for larger passive heat sinks.
Interestingly enough, SOYO slaps a passive heat sink on the board’s VT8237 south bridge. The SY-KT600 Dragon Ultra is the only KT600 board in the round-up with a south bridge heat sink, which makes me wonder if the fins aren’t just for show. More effective chip cooling certainly can’t hurt, though.
SOYO packs the SY-KT600 Dragon Ultra’s port cluster with everything from USB to audio to Firewire, but there’s more.
The board also sports an expansion header that’s riddled with extra audio ports. With the expansion header in place, the SY-KT600 Dragon Ultra offers three analog outputs, two analog inputs, and digital S/PDIF input and output ports.
Nope, we’re not quite done with ports yet. SOYO also bundles its Sigma box with the SY-KT600 Dragon Ultra, which gives users access to a couple of USB and Firewire ports in either a 3.5 or 5.25″ drive bay. The Sigma box also comes with a memory card reader and a selection of face plates to match black and silver cases.
Overall, the Sigma box is the only significant bundled extra of any of the KT600 boards we’re looking at today. Opulent bundles seem to be all the rage for high-end enthusiast-oriented motherboards, but it looks like manufacturers aren’t looking to squeeze in a bunch of extras with their KT600 offerings.
Because SOYO’s spec for the SY-KT600 Dragon Ultra clearly caters to enthusiasts, I’m not sure why the board relies on antiquated dip switches for multiplier adjustment. To be fair, some of the KT600 boards in this comparison don’t offer multiplier manipulation at all, but I can’t understand why SOYO didn’t work multiplier options into the board’s BIOS.
So far, we’ve looked at a lot of hardware, but a motherboard’s BIOS can be just as essential for enthusiasts looking to get their hands a little dirty. Here’s how our boards stack up in terms of BIOS features and tweaking flexibility:
| Multiplier control | Bus speeds | Voltage options | Memory timings | Alarm conditions | Shutdown conditions | |||||||
| Front-side* | Memory | AGP | PCI | Vcore | Vdimm | Vagp | Vchipset | |||||
| ABIT KV7 | Yes | 100-250MHz in 1MHz steps | 266, 333, 400MHz | Auto, 1/2 divider | 1/3, 1/4, 1/5, 1/6 dividers | 1.1-2.325V in 0.025V steps | 2.5-2.8V in 0.05V steps | 1.5-1.65V in 0.05V steps | 2.55-2.95V in 0.05V steps | CAS, Tras, Trcd, Trp, command rate, interleave | CPU temp, fan failure | CPU temp, fan failure |
| ASUS A7V600 | Yes | 100-250MHz in 1MHz steps | 266, 333, 400MHz | Auto | Auto, 1/6 divider | 1.65-1.85V in 0.05V steps | 2.55-2.85V in 0.05V steps | 1.5-1.8V in 0.1V steps | None | CAS, Tras, Trcd, Trp, command rate, interleave | None | None |
| AZZA KT600 ALX | No | 200-233MHz in 1MHz steps | 266, 333, 400MHz | Auto | Auto | None | None | None | None | CAS, Tras, Trcd, Trp, command rate, interleave | CPU temp | CPU temp |
| EPoX EP-8KRA2+ | Yes | 100-250MHz in 1MHz steps | 266, 333, 400MHz | Auto | Auto | 1.4-2.0V in 0.025V steps | +0.1-0.5V in 0.1V steps | None | None | CAS, Tras, Trcd, Trp, command rate, interleave | CPU temp | CPU temp |
| FIC KT-600 PRO | No | 200-250MHz in 1MHz steps | 266, 333, 400MHz | Auto | Auto | None | None | None | None | CAS, Tras, Trcd, Trp, command rate, interleave | None | CPU temp |
| MSI KT6 Delta | Yes | 100-280MHz in 1MHz steps | 200, 333, 400MHz | Auto | Auto | 1.35-2.3V in 0.025V steps | 2.55-3.3V in 0.05V steps | 1.55-2.1V in 0.05V steps | 2.6-2.8V in 0.1V steps | CAS, Tras, Trcd, Trp, command rate, interleave | Chassis intrusion | None |
| SOLTEK SL-KT600R | Yes | 200-233MHz in 1MHz steps | 266, 333, 400MHz | Auto | Auto | 1.1-1.85V in 0.025V steps | 2.5-2.8V in 0.1V steps | 1.5-1.8V in 0.1V steps | None | CAS, command rate, interleave | CPU temp | CPU temp, low fan speed |
| SOYO SY-KT600 Dragon Ultra | Not in BIOS | 100-511MHz in 1MHz steps | 266, 333, 400MHz | Auto | 1/3, 1/4, 1/5, 1/6 dividers | 1.475-1.85V in 0.025V steps | 2.6-2.8V in 0.1V steps | 1.6-1.8V in 0.1V steps | None | CAS, Tras, Trcd, Trp, command rate, interleave | CPU temp, fan failure | None |
*Based on the front-side bus options offered with an Athlon XP 3200+, which uses a 400MHz (2 x 200MHz) front-side bus by default. Overall, ABIT’s KV7 offers a more complete array of bus speeds, dividers, voltages, memory timings, and safety features than any other board in this comparison, though ASUS, EPoX, MSI, SOLTEK, and SOYO also offer well-stocked BIOSes.
Neither ASUS nor MSI offers BIOS-managed fan failure or CPU temperature alarm or shutdown conditions, but both boards have hardware-based solutions to keep processors from overheating. Personally, I prefer that safety features be integrated into a board’s BIOS to facilitate user tweaking, but I’ll take hardware-based overheating protection over none at all.
From an enthusiast’s perspective, AZZA and FIC get failing marks for their KT600 boards. Both boards support for front-side bus overclocking, but neither offers BIOS-level multiplier control. AZZA and FIC also fail to include any voltage manipulation options, but at least both offer a full suite of memory timing options.
All tests were run three times, and the results were averaged, using the following test setups.
| DFI LAN Party NFII Ultra | ABIT KV7 | ASUS A7V600 | AZZA KT600 ALC | EPoX EP-8KRA2+ | FIC KT-600 PRO | MSI KT6 Delta | SOLTEK SL-KT600R | SOYO SY-KT600 Dragon Ultra | |
| Processor | Athlon XP 3200+ | ||||||||
| Front-side bus | 400MHz (2 x 200MHz) | ||||||||
| North bridge | nForce2 Ultra 400 SPP | VIA KT600 | |||||||
| South bridge | nForce2 MCP-T | VT8237 | |||||||
| Chipset driver | nForce unified 2.45 | VIA Hyperion 4.49 | |||||||
| Memory size | 512MB (2 DIMMs) | 512MB (1 DIMM) | |||||||
| Memory type | Corsair XMS3500 PC3000 DDR SDRAM | ||||||||
| Graphics | ATI Radeon 9700 Pro | ||||||||
| Graphics driver | CATALYST 3.6 | ||||||||
| Storage |
Maxtor 740X-6L 40GB 7200RPM ATA/133 hard drive |
||||||||
| Operating System | Windows XP Professional Service Pack 1 and DirectX 9.0b |
||||||||
It took a while for some of the KT600 boards in this comparison to get stable, and some aren’t even readily available for sale yet. However, each of the boards we’re looking at today is rock solid with its latest BIOS revision. Unfortunately, each board seems to prefer a different set of BIOS-level latency and performance settings, despite the fact that our Corsair XMS3500 DDR400 DIMM is rated to work with 2-3-7-3 latencies and a 1T command rate.
In addition to running our eight KT600 boards against each other, I’ve also thrown in an nForce2 Ultra 400-based motherboard from DFI to illustrate how the KT600’s performance compares with the best NVIDIA has to offer for the Socket A platform.
We used the following versions of our test applications:
- SiSoft Sandra Standard 2003
- ZD Media Business Winstone 2002 1.0.1
- ZD Media Content Creation Winstone 2002 1.0.1
- TCD Labs HD Tach v2.61
- Futuremark 3DMark03 Patch 330
- Quake III Arena v1.31
- Unreal Tournament 2003 demo
- RightMark Audio Analyzer 5.1
- RightMark 3D Sound 1.0
- Cinebench 2003
- Sphinx 3.3
The test systems’ Windows desktop was set at 1024×768 in 32-bit color at a 75Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests. Most of the 3D gaming tests used the high 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.

Our nForce2 platform is smack in the middle of a stack of KT600 boards in Sandra’s memory bandwidth tests. Among the KT600 boards, SOYO, MSI, EPoX, and SOLTEK all offer impressive memory bandwidth from a single channel of DDR400 memory. AZZA, FIC, ASUS, and ABIT don’t fare so well.
What’s particularly impressive about the KT600’s performance in Sandra is how well its single-channel memory controller is able to keep up with NVIDIA’s dual-channel design.

In Cachemem, MSI leads the pack when it comes to read memory bandwidth, and EPoX beats out the rest of the KT600 pack in terms of write bandwidth. None of the KT600 boards offer more write bandwidth than the nForce2 Ultra 400, but the KT6 Delta-FIS2R and EP-8KRA2+ both manage more read bandwidth.

MSI leads in memory latency, easily outdistancing the closest competition. The nForce2 ends up in the middle of the pack again, and ABIT and ASUS bring up the rear.
Unfortunately, I was unable to get the Promise controller on MSI’s KT6 Delta-FIS2R to produce reliable scores in HD Tach. The controller’s read burst speed scores were physically impossible given the bandwidth limitations of the PCI bus and even the 8X V-Link interconnect, so I’ve left the PDC20378 out of our disk controller results. However, even without the PDC20378, there are still a staggering number of scores to scroll through.
“Parallel” ATA controllers were tested with a Maxtor 740X-6L 7200 RPM hard drive, while Serial ATA controllers were tested with a Western Digital Raptor WD360 10K RPM drive.

In HD Tach’s read burst speed test, SOLTEK’s SL-KT600-R is the only KT600 board to differentiate itself from the pack, and not in a good way. Drives connected to the board’s south bridge disk controllers hit a wall around 92MB/sec, which is only a small drop in performance for Serial ATA drives, but a huge hit for “parallel” ATA drives.

The SL-KT600-R redeems itself in HD Tach’s average read speed test, where scores are more dependent on hard drive performance than anything else.

In HD Tach’s write speed tests, ASUS, FIC, and SOLTEK’s boards have a few performance problems. The A7V600’s performance with both Serial ATA and ATA/133 drives puts the board in last place.

There’s a slight variance in disk access time scores between our assembled KT600 boards, with SOLTEK, FIC, and ABIT leading the way with a little help from Western Digital’s Raptor. The CPU utilization scores get even more interesting.

Yikes. It looks like VIA’s VT8237 has some serious CPU utilization issues with “parallel” ATA drives. However, ASUS’ A7V600 has no problem with CPU utilization at all, which leaves me scratching my head. The CPU utilization results for each board were consistent across three test runs, and I’m at a loss to explain why ASUS’ board has such lower CPU utilization using the exact same IDE drivers as the other boards.


Lead by MSI’s KT6 Delta-FIS2R, the KT600 rips through the Business Winstone test. All the KT600 boards are pretty closely matched here; barely more than a single point separates the fastest and slowest boards. For the first time in this comparison, the nForce2 falls to last place and doesn’t look all that hot against the KT600.
The nForce2 Ultra 400 redeems itself in the Content Creation Winstone test, but it can’t quite catch SOYO’s SY-KT600 Dragon Ultra. FIC’s KT-600 PRO brings up the rear this time around.
Gaming





In all of our gaming tests, the KT600 boards lose out to the nForce2 Ultra 400. Among the KT600s, SOYO and MSI continue to lead the field.




The KT600s can’t catch the nForce2 Ultra 400 in Cinebench, either. MSI’s KT6 Delta-FIS2R again leads the KT600 boards, followed closely by SOYO’s SY-KT600 Dragon Ultra and EPoX’s EP-8KRA2+.
Sphinx speech recognition

The nForce2 Ultra 400 is still out ahead in Sphinx’s speech recognition tests, and as usual, MSI leads the KT600 pack. The performance of the ASUS and ABIT KT600 boards continues to be disappointing, and I’m beginning to suspect that both boards have more relaxed internal timings. Less aggressive internal timings may make the boards more stable at overclocked speeds, though.


Thus far, MSI’s KT6 Delta-FIS2R has outperformed the rest of the KT600 competition, but the board’s CPU utilization is alarmingly high in our 2D and 3D DirectSound tests. On the flip side, ABIT and ASUS have been dogging our performance tests so far, but both boards turn in low CPU utilization in our audio tests.
SOYO’s SY-KT600 Dragon Ultra is the only KT600 board to bypass VIA’s VT8237 integrated audio in favor of C-Media’s CMI8738, which pays off when it comes to CPU utilization. The Dragon Ultra’s CPU utilization is nearly as low as the nForce2, which is pretty impressive.
Audio quality
In RightMark’s audio quality tests, I used an M-Audio Revolution 7.1 for recording. Analog output ports were used on all systems.
To keep things simple, I 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 LAN Party NFII Ultra’s Soundstorm audio offers better dynamic range than any of the KT600 boards, but MSI’s KT6 Delta-FIS2R isn’t far behind. All told, the integrated audio solutions available on most of the KT600 boards we’re looking at offer only average dynamic range.
Many of the KT600 boards falter in RightMark’s frequency response tests, but ABIT’s KV7 excels. The board’s integrated Vinyl Audio does well against even NVIDIA’s vaunted nForce2 audio. FIC’s KT-600 PRO doesn’t sound that bad, either.
Our USB 2.0 and Firewire transfer speed tests involve transferring a mix of files totaling 1.89GB from a USB 2.0/Firewire external hard drive enclosure to the test system. The hard drive enclosure houses a 7200RPM Maxtor DiamondMax D740X hard drive.

With the exception of AZZA’s KT600 ALX, all the KT600 boards have impressive USB 2.0 transfer speeds. ABIT’s KV7 leads the pack, but MSI’s KT6 Delta-FIS2R isn’t far behind. Unfortunately, AZZA’s KT600 ALX didn’t recognize our hard drive enclosure, despite the fact that the board’s USB ports worked just fine with other devices.
Note that the KT600s are roughly 50% faster than our nForce2 Ultra 400 board when it comes to USB 2.0 transfer speeds. The nForce2 isn’t necessarily faster at everything, folks.

EPoX’s EP-8KRA2+ just edges out SOYO’s SY-KT600 Dragon Ultra for the lead in our Firewire transfer tests. The nForce2 board doesn’t fare too poorly here, but it actually uses the same VT6307 Firewire chip as the rest of the field. Since the boards from ABIT, ASUS, AZZA, and SOLTEK don’t have Firewire support, they don’t have a score in this test.
Our Ethernet speed tests involve downloading the same 1.89GB batch of files used in our USB and Firewire tests from a file server based on Intel’s Springdale platform. The Springdale board’s CSA-attached Gigabit Ethernet port was used to serve the files, and both the server and test systems were connected to a 100Mbit Ethernet switch.

ABIT’s KV7 nails our Ethernet transfer test, while SOYO, MSI, EPoX, and our nForce2 board fight it out for second place. Had we used a Gigabit Ethernet switch for testing, the GigE chips on ASUS and MSI’s KT600 boards would have had an opportunity to stretch their legs. However, Gigabit networking gear is still on the expensive side for many businesses and home networks.
Since SOLTEK’s SL-KT600-R doesn’t offer integrated Ethernet support, it can’t participate in this test.
Overclocking
For our front-side bus overclocking tests, I used an unlocked Athlon XP 2600+. To isolate the motherboard’s stability at out-of-spec front-side bus speeds, I lowered the processor’s multiplier to keep the processor running below its stock clock speed of 2075MHz.
Just because some of the board samples we tested took to front-side bus overclocking better than others doesn’t guarantee that other samples won’t be more or less comfortable at overclocked speeds. Overclocking success is never guaranteed, and it can sometimes depend as much on the individual characteristics of on-board chips as on karma and the proper alignment of the planets.


EPoX’s EP-8KRA2+ managed the highest stable front-side speed of all the KT600 boards, but ASUS’s A7V600 and MSI’s KT6 Delta-FIS2R were both comfortable at 210MHz. Surprisingly, neither the ABIT nor SOYO KT600 boards were stable with overclocked front-side bus speeds, even with relaxed memory timings.
A lack of multiplier control prevented me from isolating the front-side bus of AZZA’s KT600 ALX and FIC’s KT-600 PRO, so they don’t appear in our overclocking tests.
The big question on the minds of many PC enthusiasts is whether VIA’s KT600 has a chance at toppling the well-entrenched nForce2 as the best chipset choice for Socket A. Unfortunately, the answer isn’t an easy yes or no. Let me explain.
The results of our testing clearly show that the nForce2 platform is faster for content creation, rendering, speech recognition, and gaming. If you’re looking for top performance, it’s still the way to go. However, the KT600 does highlight a number of nForce2 deficiencies, including weak USB performance and a lack of south bridge Serial ATA. From a features perspective, the nForce2 looks a little dated next to the KT600, whose performance really isn’t all that far off in many of our tests.
While power-hungry users are still better off with nForce2 boards, the KT600 is quite an attractive platform for feature-driven enthusiasts who don’t mind losing a couple of frames per second here and there. The fact that KT600 boards don’t need two DIMMs for optimal performance is just icing on the cake.
So if the KT600 is a more appropriate platform for feature- rather than performance-driven enthusiasts, which KT600 boards are worth checking out? Here are my favorites:
- ABIT KV7 – I know what you’re going to say: “But the KV7 is one of the slowest of the lot.” Yes, it is, but not across the board. The KV7 is actually the best KT600 board in terms of audio quality, and the fastest when it comes to USB and Ethernet transfer rates. At only $74 online, the KV7 is also the cheapest board in this round-up, and it comes with the most feature-rich BIOS and smallest footprint.
- EPoX EP-8KRA2+ – If you’re willing to spend a little more, EPoX’s $88 EP-8KRA2+ adds IDE RAID and Firewire to the KT600’s already rich feature set. The EP-8KRA2+ also has a useful post code display, great performance, and enough BIOS options to keep more casual enthusiasts happy.
- MSI KT6 Delta-FIS2R – At $126 online, the KT6 Delta-FIS2R isn’t a cheap KT600 option by any stretch, but the board’s performance and features deserve distinction. With Firewire, Gigabit Ethernet, IDE and Serial ATA RAID, and the fastest overall performance of any KT600 board, the Delta-FIS2R looks pretty sweet for enthusiasts who want it all.
What about the rest? For starters, AZZA, FIC, and SOLTEK’s boards don’t offer better performance or more features than the competition. ASUS’s A7V600 has Gigabit Ethernet, which is nice, but that’s about the extent of the board’s extra features. The A7V600 fails to shine in the performance department, too. SOYO’s SY-KT600 Dragon Ultra is an interesting board, and I certainly like the Sigma box as a bundled extra. However, the Dragon’s multiplier dip switches are a big turnoff. The board also has no answer to the KT6 Delta-FIS2R’s Gigabit Ethernet and more flexible IDE/SATA RAID support. Well, there you have it. With the KT600, VIA has succeeded in creating a viable alternative to NVIDIA’s nForce2 Ultra 400. The KT600 may not win every benchmark battle, but the chipset’s Serial ATA support and peripheral performance make it an attractive low-cost option for enthusiasts looking for a competitive and feature-rich Socket A platform.














Why no a test with AF1 Deluxe from ECS? It´s should be better than MSI!! What not???
You people all suck. AMD needs your help – buy their processors, fer chrissakes, and offer to pay *more* than the asking price!
“Oh, wah!”, you say. “Opteron is too expensive!” It’s pricey because AMD needs the money. Here’s what you must do: Buy two Opterons, and then write AMD a check for $500 (in addition to whatever you paid retail).
Otherwise, the terrorists will have won. If you don’t overpay for an Opteron, you love Osama bin-Laden.
Nice write up. Makes me wanna upgrade my mobo/case this month.
q[
KT600 is DOA with the KT800 right around the corner.
but then isn’t everything?
I don’t follow. The kt800 is a athlon 64 board, and its already came around the corner. It will be a long time before the athlon 64 is a common find in households. I welcome more XP compliant motherboards, I wish another Nforce board would come out. Via has already done 2 Nvidia should do 2 (not counting the 266fsb era).
Just out of curiousity when exactly did ‘enthusiast’ turn into ‘cheapskate’?
Back in my day ‘enthusiast’ meant totally tricked out, hopped up, and loaded pc’s, ie the guys who WOULD be buying the FX-51’s and raided 10K drives and 9800Pro XT’s.
I always thought enthusiast was “build your own.” Kinda taken from car circles, and sewing granniesg{<.<}g "Enthisiast Knitters 0wnz0r5 J0000g{
enthusiast knitters have the market sewn up… 🙂 Yes I have a promising career in newspapers.
Yeh enthusiast to me was always a person with a higher than average knowledge of the hardware and who built his own machines, not necessarily one who spent craploads on high-end hardware.
q[
Duuhhh! Completely forgot about that.
Now I’m curious as to why I have the digital out like that. I guess it is just more choice.
I would think using a cable from the CD drive to the soundcard for digital playback would be more efficient than doing it without the cable (which is basically equivalent to ripping and playing it back in real time), since it shouldn’t utilize your bus or CPU at all.
hmm, hard to say. Regardless, I doubt anyone could tell the difference.
I’ve always had digital cd playback simply because I’m lazy to connect the analogue cable, and because a lot of mainboards only have 1 CD in connector.
I had felt the CPU utilization for the playback was pretty negligible… not to mention you’d be getting one less hassle of a cable, too.
No CD digital in?
My DVD-ROM has digital out and it plugs into my Live!’s cd digital in. I noticed that mainboards with integrated sound don’t have this. What’s up with that?
q[
yes, I think so. Abit are using Rubycon capacitors throughout on many of their mainboards nowadays.
My personal experience with cap-busting boards has been 2 Abit, 2 MSI, and one FIC. One of the MSIs managed to fry a power MOSFET and vaporize a couple of PCB traces when the caps blew. The caps on the FIC didn’t actually burst, they were just bulging slightly (so I replaced them proactively).
I’ve lost an Abit board to bad caps as well as a GeForce 4200. I’m running two more older Abit boards and I fear that I’m going to loose them as well.
Good write Diss…
I’m afraid. I think this is one of the signs of the End: LicketySplit gets his KT600 review.
Great job, Geoff. It proves a point that some people refuse to acknowledge; VIA has learned what needs to be done.
(Are you listening Klopsik206?)
VIA has gotten better and better over the last few years.
I am writing this on a A64 3200+/MSI K8T combo, and it has been one of the most solid boards I have ever used.
(My other system is a MSI 875 board with a 2.6C, so I am comparing it to that.)
VIA is a perfect example of watching, listening, and learning to find out where your product needs to improve. This is what ATI did the last two years, and Nvidia needs to learn this as well.
Hell, I still keep my fingers crossed the Matrox can learn this, if they want the enthusiast back. 🙂
I am glad the MSI board performed well and got a good rating.
I started off using Abit and Asus boards, but my last 4 or 5 systems have had MSI boards and they have kicked ass.
Well, bear in mind that A64 chipsets are vastly less complicated than athlon or P4 chipsets to design. The memory controller has been taken out of the loop which is a BIG chunk of complexity,…
and keep in mind VIA has been at this game for many, many years, so you’d think they’d ‘learn’ sooner or later. The real test will be after the chipsets have been on the market for a few more months. Early impressions mean little.
I’m sure Diss thinks the EE is a steal at intel’s price. cheesh what a fanboy.
SATA is kinda neato, but I work with 40+ computer folks, and I don’t know a soul who has it or even is thinking about doing it. Reality check. SATA feature isnt that hot yet. yet. next year this time. maybe.
Well, i guess if you’re buying a motherboard for only the next 11 months, you’ve got a point. However, the KT600 is at least well-equipped for when Serial ATA does get big, which should be an important consideration for anyone looking to keep a board for more than the next six months.
Perhaps you should read my coverage of the P4 EE. I’ve repeatedly stated that the chip’s price is horrendous. On the other hand, native Serial ATA support in a mainboard for only $75 is a pretty sweet deal.
SATA rocks. Your friends are missing out. It’s not fantastic from a performance standpoint, but the new cabling is most welcome.
Think you meant number 9 there Forge… 😉
I guess he’s just somebody that have to read up a bit better, we wouldn’t want to consider him a troll.. right.
Yeah! No! Ummm… You! AG over there!! You!
Stoopid auto-reply.
Serial ATA gives me faith that somebody hates ribbon cables, master/slave and bent pins as much as I do. But….where are the SATA optical drives already.
We need more SATA ports to hook them to, first. As is, only half of my hard disks are on SATA, since only the chipset native ports perform correctly in all situations. The SiI ports are mostly right, the Promise ones are often broken, and HPT is taking their sweet time makign mobo-ready SATA gear.
Once we see new mobos having 6 or 8 or even 10/12 SATA ports, on good controllers (perferably chipset ones), then SATA optical will be due.
Most people only want to hook up 1 hard drive and 1 optical drive. Two opticals if they want their DVD and burner separate. Most people /[
I agree. The main thing I looked at wast the serial ATA preformance. You my friend have just saved me a load of troubles.
I’m sure Diss thinks the EE is a steal at intel’s price. cheesh what a fanboy.
Did he say anything about the EE? Did he say it was faster or better than the a64?
No, but for every AMD post, he’s like wait till intel releases X. And every intel post has a How is AMD going to compete with this line in it.
I always thought they were pro AMD. I’m pretty sure for the most part they are or at least not far from it.
But when it comes to ATI or Nvidia it gets disgustingly clear.
I agree TR is a ATI hoe. Even I love ATI but I can’t stand tainted reivews.
It’s because, the NV3X series really does suck at DX9 apps it doesn’t matter what way you look at it. The NV3x series design was too overambious for Nvidia not following the DX9 spec to the letter. I don’t know whoever is at fault Nvidia or MS. The result is that NV3x design is more like a Qutero then a actual gaming card and I won’t be suprised that the NV3x series were better at content creation in terms of ether performance or IQ then R3xx series. The fact is that Nvidia the company’s managment rather accepting defeat with huminity and try to catch up with the NV40. Instead Nvidia resorted to deceitful PR tactics and dysfunctional drivers that sacfitce image quality for performance. I have no personal product loyalty I get whatever is the best bang for the buck at the time of purchuse. Getting the bleeding edge is never worth it unless, you get a new system every year or two.
You act like it’s the difference between 0fps and 100000fps. There isn’t a big gap between the fx and the 9xxx.
The only time you see something dramatic is when HL2 is thrown into the mix. Which should be no surprise HL2 is coded with vendor specific code it isn’t technically PURE dx9 all this shows us is that any game with the HL2 engine will play on nvidia hardware. That means what 5 games max?
Then you start blubbering about Nvidia is going to fail Nvidia screwed me Nvidia blah blah Nvidia blah. It’s not that big of a deal. You ATI FANATICS should know tolerance h*ll WTF was going on when your drivers sucked beyond tolerance. You just sat there like it was no big deal and faithfully stuck by them. You are NOTHING more than hypocrites.
I am sick of this attitude and your ignorance.
Dude, the difference is that Radeon 9700 PRO can run TR:AOD a directx 9 app not HL2 I mind around 40-50 fps were the Geforce FX 5900U barely pulls 30 fps at same res and settings. Geforce FX 5900U was release nine months later and has considerible faster specs not mention the Geforce FX 5900U does it at lower IQ then the Radeon 9700 PRO. I believe you are one of those Nvidiots who was boned when you invested into Geforce FX 5900U now DirectX 9 apps are coming onto the horizon they will run and look like crap compaired to ATi line. The only thing the Geforce FX 5900U and it’s sibings are good for is ether OpenGL or DirectX 8.1 or eailer. Where Nvidia didn’t have to “optimized” their drivers since the Nv3x series does well if not better then R3xx in these areas. What upsets me the most of this is the whole ordeal is arrogence of Nvidia’s management resorting to PR and FUD tactics ever since the Radeon 9700 PRO was released.
TR:AOD hahaha shows how much you know.
I am not a “nvidiot”. The only nvidia card I own is a ti500, and that was given to me. I have always bought ATI products and continue to do so.
Look like crap…? I have seen screen shots I have 20/18 vision. There isn’t a HUGE deal. Again you are running around like the sky is falling. Lol when the nvidia drivers go as far as to do just that 🙂 and produce horrible anomalies then yeah its a big deal, but they fix them.
The whole reason I am even saying anything is the fact that I hate your attitude. I don’t care whose side your on people like you ruin it for everyone. I am disgusted that you represent ATI in such a way. I said hypocrites for a reason. I stuck by ATI in hard times and I fell like a million bucks that they could finally show their full potential but don’t judge so quickly we all went through the same thing. Nvidiots teased us when we had bad drivers lets try to be better than them this time around otherwise we might fall into the same hole.
Since you have such a thick skull:
Then you start blubbering about Nvidia is going to fail Nvidia screwed me Nvidia blah blah Nvidia blah. It’s not that big of a deal. You ATI FANATICS should know tolerance h*ll WTF was going on when your drivers sucked beyond tolerance. You just sat there like it was no big deal and faithfully stuck by them. You are NOTHING more than hypocrites.
I am sick of this attitude and your ignorance.
I never stated that ATi was a perfect angel. I knew ATi had a nasty history of the Rage series which had craptastic performance and drivers. The first generation of Radeons there was misleading information on the actual specs and too many different verisons with various clock and memory speeds. Not to mention the whole Quack ordeal but at least ATi when they were caught they removed the hack in the next driver release shortly after. Nvidia took the whole thing to a new level after the day of Radeon 9700 PRO. Where Nvidia panicked like the sky was falling by giving the FX 5800U exotic 1 Ghz DDR-II RAM, and ramping up with it’s clock speed cooling it with the infamously loud dustbuster. Which still did little against the forminable Radeon 9700 PRO. Nvidia resorted to PR and FUD tactics like the exclusive deals with EA “The way is was meant to be play”. Strong-arming Futuremarks because, the Geforce 5800 FX sucked at 3Dmarks2K3 and Nvidia cheated with their drivers in order to be on par with the Radeon 9700 PRO. Nvidia blackmailed various big-time review sites to give ether favorible reviews or lose privllages. Disnouning the results of any Directx 9 app benchmark such as AOD and HL2. The sheer arrogence that Nvidia displayed on the course of these actions angers me to no end. The amount of resources spend on over-hyping the NV3x series could of been better utillized on designing the NV40 which could be a killer design. It’s quite apperent to me you are the one that is truly ignorant. BTW my previous card that I brought was a Geforce 3 which was great card for it’s time. I decided to skip on the Radeon 8500/Geforce 4 generation and the when the Radeon 9700 PRO came around. The card was a very attractive option both in technological capaiblity and performance. So I decided to take the gamble on getting potentally crappy drivers and support. But, it just got better over time and I’m glad that I got the Radeon 9700 PRO. Which is still running strong and still beats Nvidia’s best at DirectX 9 apps.
XGI Volari is going to be taking the #2 spot quite soon. They are also going to be buying out Nvidia’s graphics market. Nvidia actually wants this so they can focus on motherboards. Due to the failure of the 5800 Nvidia’s attention was taken away from the NF3 hence the poor performance and features compared to a kt800 board.
This will all work out for the best, it’s time Nvidia resigns from the graphics industry.
Well, isn’t that what this business is about? Seeing how all the competition will react to a new and exciting release?
I’m actually pleased with how little bias most TR articles betray. They figure out who’s hot and who’s not, usually with good science, and then they base their recommendations on that. I’d say TR, for the most part, is pro-whatever’s-the-best-at-the-time.
AHAHAHAHAHAHAHAHAHAHA.
That sums it up nicely.
Not really, look at diss’s, win64 delayed(suprised? i’m not), amd64 is a sinking ship post today.
I heard that SoundBlaster Audigy got some problems running on previous VIA boards.
Is it the case with KT600 as well?
Anyone knows?
Via fixed those issues with the 8235 southbridge on. these boards use 8237’s so they outta be fine.
If you’re gonna consider a VIA board at least make sure it has the 8235 southbridge as a minimum.
“Via fixed those issues with the 8235 southbridge on”
I hate to admit being, in the past, a VIA beta tester. I can truthfully say that it isn’t going to happen any more.
True: I might miss out on some hardware.
True: I don’t think I’m about to miss out on any truly GREAT hardware.
I think I will leave the VIA beta testing to others.
Ehh, whatever. Beating on VIA is a lot like beating on MSFT or Apple… Fun, but that’s about all.
“Beating on VIA is a lot like beating on MSFT or Apple”
But is it?
I have NEVER bought any Apple computers, but I bought mainboards with VIA chipsets with the promise that all ‘the problems’ had been fixed, when they haden’t.
Didn’t the ‘head guy’ at AMD say ‘real men own fabs’
Your ‘whatever’ not-withstanding, I am going to say “real men don’t buy garbage with VIA chipsets”.
Look your a moron if you can just shrug off via’s problem with the audigy cards.
I spent 2.5k (back in the day) to put my first system together. Via ruined it all for me. It made me sick to think the purchase was a complete waste.
Not only did Via mess up the audigy but they messed up my VIVO card. Only until a year later did I finally figure out the culprit (via denied all of it).
I almost got out of building my own pc’s thanks to VIA. It literally made me sick to think I had wasted 2.5k.
Of course review sites blindly touted the via boards around like a kid eating paste. I learned 2 things 1. never trust via 2. never trust reviewing sites.
After some years have past I have gotten over it but I will NEVER buy a via boar pre a64. I will only buy one of their a64 boards. I have also began to trust some review sites. I am a bitter man, never screw with my money.
Via just doesn’t offer anything worth taking the risk compared to NF2 boards. I know NF2 boards have some problems but at least places like this put down the paste and addressed the problem so we all could know.
Hear, Hear!
“I will only buy one of their a64 boards.”
You sir, are a far more ‘trusting’ fellow than I am. In fact, I will go you one farther……..if AMD isn’t interested in making sure that there is another source (besides VIA) for chipsets to run their newest processors I can ONLY conclude that they do not give a shit about their customers and I can do without!
…except their is another source
Two others, in fact, with a third and fourth alternative coming up soon.
I of course refer to the AMD-8000 series and the nForce3, with ALi and SiS’s chipsets coming soon.
Of course, if you want an affordable chipset with the best performance, it’s all about K8T800 now.
The way I see it, there’s hardly any northbridge to screw up, and Via did a whole new from-scratch PCI bus in 8235 and up, so I’ll go K8T800 in a minute, if that’s the best performing chipset out.
My thoughts exactly!
So are we going to figure out why CPU utilization was so high for all of the Parallel ATA drive tests except for the Asus?
I’m definately not impressed by the IDE CPU usage statistics either, my KT266A board scores about 20% in HD Tach (WD 80gig “JB” model).. Nforce2 boards have the same problem? Looks as though Asus might be the way to go. I’ve held off on Nforce2 boards because of the CPU usage by IDE, and their USB 2.0 performance is sub-par to boot? I must say KT600 looks like a nice economical solution with the SATA on southbridge and all, the slight performance increase in games on Nforce2 hardly seems worth it, especially since you need 2 identical modules, and be picky about choosing them for the dual-channel thing.
Very nice, thorough review! GJ!
Man, Diss, I gotta hand it to you, your photography just keeps getting better and better. I don’t think I’ve seen better component pics anywhere, commercial sites included. I can just imagine what you could do with a panoramascope…
Your writing is squeaky clean this time, too. Top-notch report, Diss!
(And you didn’t even recommend the Soyo board… will miracles never cease? 🙂 )
sunlight always adds the right light for mainboard shots, I find 🙂
hhhmmmmmm
BAH! You should have compared 6 NF2 boards to 6 KT600’s or No NF2 boards at all.
This falsely leads people into thinking that’s a good comparison of NF2 performance. Which is impossible since no one NF2 board can have all the bases rounded like 6 NF2 boards can.
You supply the 6 NF2 boards and I’m sure Diss would be more then happy to bench them. But you have to HAVE 6 NF2 boards to bench and who here has the $1000 (or more) needed to buy them?
DUH then you leave the nforce2 board out of it you missed my point.
So you’d rather see a straight KT600 shootout with *no* nForce2 reference platform for comparison? Are you nuts? That’s like reviewing Radeon 9600 XTs from half a dozen manufacturers and not throwing in a token GeForce FX 5700 Ultra for comparative reference.
You are the one that is nuts. My first post says it all. A graphics card comparison is different. One mobo platform from the next varies (heck the review shows you that) throwing a random nf2 board into a sea of kt600’s isn’t going to be a fair comparison. Everyone I know is now saying “wow kt600 boards rock NF2 just look at the review”.
It is NOT as strait cut compared to graphics card to graphics card.
/[
…besides, I would find it very odd if the results obtained from the nforce2 Ultra 400-based DFI would be radically different to any other nforce2 boards out there. Things are generally in the same ballpark on the same chipset, especially when it comes to things like USB2 and firewire transfer speeds. And yeh, it wasn’t an nforce2 comparison to start off with, but the comparison adds a lot of juice to the review, so nice touch I say.
First tell me did the DFI board beat every single other NF2 board during the NF2 comparison in all tests?
You’ll notice that we don’t have an nForce2 motherboard comparison.
Though if we ever did one, I’d have included a KT400A reference platform.
I’m not sure what’s annoying you so much here. We compared the performance of 8 KT600 boards with each other, and to give a general idea how they all stack up against the nForce2 platform, we threw in a retail NF2 board. Simple.
It’s not that simple. I am sorry as though I sound so annoyed. I should not so easily dismiss such a great review. The only thing I disagreed with is the addition of the nf2 board.
All I want is a fair review, one that doesn’t accidentally mislead people. That’s not really a “general” idea.
A board will excel in some areas and completely suck at others. 1 board isn’t a “general” idea, if you somehow got a second nforce2 board that excelled at all that the DFI sucked at then I wouldn’t be so flustered.
Otherwise if the DFI board excelled at all areas and beat all other NF2 boards I wouldn’t have a problem at all with it.
Relax, take some blood pressure medicine, and be happy.
I think including 6 NF2 boards in this would have made it way too long and boring, but leaving it out completely would make me go, well, where is this in comparison to other chipsets?
Excellent job, Diss.
Boring? To you perhaps. I happen to like tech. Too long? Fine leave ALL the NF2 boards out.
I have no idea how it could be so freakin hard to understand. 1 board is pointless it only mucks everything up. Remember me saying 4 people are now basing their decisions on this article? Make that 6 now, well they might not be totally sure about purchasing a new board but they have ALL said “looks like VIA takes back the crown” or to some extent.
All good intentions aside it still needed to be carried out fairly. Either go all the way through with it or not at all (comparing the NF2 part not the whole review).
I know you only meant good when throwing that board in there but on accident you caused a problem of misconception. That being people thinking 1 NF2 board reflects how all of them compete with kt600’s.
I agree this wasn’t a good review in THAT respect, all the kt600s was still great. Everyone at my work has been arguing about it (techie people here 🙂 ). 90% are begining to loose faith in TR might aswell start doing stuff like those “other” review sites.
Ok here’s the deal. Lets say you were looking for a board that was the best at a certain feature (take your pick on which). I know of 4 people already that are in the market for a new pre athlon 64 board. All of them have looked at the kt600 review and all have said that the kt600 platform is better than the NF2.
Back to the feature part. The DFI won’t have all the bases rounded (like I said earlier). You can’t throw 1 board into the mix and expect it to compete on all levels. Only if the DFI beats ALL other NF2 boards on every level could that even be considered a smart move.
All this has done is shown people how the DFI board and the DFI alone compares to 8 (sorry for the typos earlier) VIA boards.
They aren’t going to vary by much you say? Take a look at the review people! At one point a board is on top in one test then clear at the bottom on another you have to expect the DFI is going to have weaknesses.
Wow, great roundup I enjoyed every minute of it.