Home TR reviews Abit’s KT7A-RAID motherboard

TR reviews Abit’s KT7A-RAID motherboard

Andy Brown
In our content, we occasionally include affiliate links. Should you click on these links, we may earn a commission, though this incurs no additional cost to you. Your use of this website signifies your acceptance of our terms and conditions as well as our privacy policy.
Manufacturer Abit
Availability Now

In this context, the KT7A-RAID (and indeed, any motherboard based on the KT133A chipset) was behind the curve, a half-baked solution in the eyes of hard-core geeks. After all, why just crank your system bus up to 133MHz when you could do that and double your memory bandwidth? Duh! At Comdex, nobody took the KT133A chipset seriously.

But it’s certainly being taken seriously now, isn’t it? So what happened? That actually falls into two categories: what did happen, and what didn’t happen. What did happen was that people testing KT133A boards found them to be within spitting distance of AMD 760 boards, DDR be damned. Few if any would’ve predicted that such a board was capable of such a feat with half the memory bandwidth of its competitor, but there it was.

And what didn’t happen was DDR. There were no piles of DDR DIMMs selling for a piddling ten percent premium. Heck, for months it was nearly impossible to find a motherboard on which to put the stuff. That’s starting to change now. As I write this, Micron has started selling PC2100 DDR DIMMs through its Crucial website for a mere three percent premium over CAS2 PC133 memory. But in the meantime, the KT133A chipset has had a chance to prove itself, and that DDR upgrade isn’t as pressing as it used to be. Into this context steps the KT7A-RAID.

Show me what ya got
So let’s take a look at what the KT7A-RAID brings to the table. We’ll use this nifty little chart that I blatantly ripped off from Damage’s AV32 review. Heh.

CPU support Socket 462-based CPUs, including AMD Duron and Athlon processors
Form factor ATX
Chipset Via KT133A (VT8363A North Bridge,
VT82C686B South Bridge)
ISA slots 1 (shared with PCI)
PCI slots 6 (one shared with ISA)
AGP slots 1, 2X/4X AGP w/sidebanding and fast writes
Memory 3 168-pin DIMM sockets for PC100/133 SDRAM (1.5GB max)
Storage I/O Floppy disk
2 channels ATA-100
Highpoint HPT370 RAID controller
Ports 1 PS/2 keyboard, 1 PS/2 mouse,
2 serial, 1 parallel, 2 USB,
2 additional USB ports via expansion header
BIOS Award PnP with Abit Soft Menu III
Bus speeds 100MHz-183MHz
(official support for 100 & 133MHz)
Monitoring Voltage, fan status, and temperature monitoring

As you can see, the KT7A-RAID is just packed with goodness. Notable features include a whopping half-dozen PCI slots, support for 512MB DIMMs (and thus a capacity of 1.5 gigs of RAM), and Highpoint’s IDE RAID controller in addition to the standard ATA-100 controller, for a total of up to eight IDE devices. They even throw in a token ISA slot in case you still have need of a legacy card.

The selection of bus speeds is also worth a second look, not only because the board allows you to set speeds up to an eye-popping 183MHz, but because of the official support for a 133MHz bus. The 133MHz bus support, of course, is what puts the A in both the KT7A-RAID and its KT133A chipset. It’s there to support the newer “C” Athlons that are sold to run on a 133MHz bus, but that doesn’t mean we can’t whip out the pencil and have some 133MHz fun overclocking the older “B” (100MHz) chips.

And with the KT7A-RAID, fun is only a delete key away. Pop into the BIOS, and Abit’s Soft Menu III allows you to set the bus speed as well as core and I/O voltages. Assuming you have an unlocked CPU (or have unlocked it yourself) you can also manipulate the multiplier. With so many things to play with and no jumpers to worry about, finding the ideal multiplier/bus speed combination is, in theory, easy. We’ll get to actual practice later on.

Lifting the lid
Breaking open the KT7A-RAID’s box reveals a well-packed motherboard sitting on a cardboard shelf. Remove the board and flip the shelf up and you get to the rest of the goodies. These include a nice beefy user manual (typical Abit), a drivers CD, all the necessary cables including two 80-wire IDE cables, a floppy cable, and a USB cable. The latter features two USB ports on a backplane, with a connector that plugs into the USB header on the motherboard. This complements the two USB ports located next to the mouse and keyboard ports, bumping the total number of USB ports to four.

Setting the other contents aside for a moment, let’s take a closer look at the motherboard itself. One extra that immediately jumps out at you is the north bridge chip. Not only is it turned at a 45 degree angle (Abit claims this shortens traces and enhances stability), it’s actively cooled—a nice touch for overclockers.

The KT133A north bridge chip is mounted 45 degrees off square

Another component that jumps out at you is that rightmost fan header just below the ATX power connector. Whaddaya mean it looks just like the one next to it? See, that header is actually very important. Without a fan attached to it, the KT7A-RAID won’t even power up. As some of you are undoubtedly aware, AMD CPUs run hot. No, seriously, I mean damn hot. Hot enough that if the CPU fan suddenly decided to stop spinning, it’s entirely possible your CPU could get fried before you figured out what the problem was.

The board’s dual fan headers

Abit’s solution is that magical little fan header. Plug your CPU fan into it, and if the RPMs on the sense line drop to zero, it instantly powers things down. Pretty slick idea, in my opinion. If you’re feelin’ lucky, the feature can be disabled in the latest BIOS.

Finally, there’s the obvious chip between the IDE ports and the battery that has “KT7A-RAID” printed on it in extremely large lettering. This is actually the Highpoint HPT370 chip, the controller that puts the RAID in KT7A-RAID.

That’s probably enough time spent just looking at the board, so let’s talk about actually installing it. With so many features, the KT7A-RAID is a fairly large board, measuring approximately 12″ by 9.1″. Most cases should fit the board without a problem, but if you’re looking to shoehorn it into a small case, check the measurements before purchasing. In my experience, when you’re measuring a case, it’s a good idea to allow for some extra room between the top of the board and the power supply. Socket A heatsinks can be quite challenging, and every bit of space helps.

Speaking of heatsinks, if you’re looking to build a system around the KT7A-RAID, make sure that your choice in heatsink will fit. As you can see, the board has some substantial capacitors just south of the socket, and these can interfere with the fitting of some heatsinks. Your best bet is to order your heatsink and board from the same place, and make sure at the time of your order that fit between the two won’t be an issue.

The DIMM clip’s placement is a bit inconvenient

I discovered another minor fit issue during installation: the AGP slot potentially interferes with the DIMM sockets. The scope of this problem will likely depend on your video card; the GeForce 2 GTS pictured here keeps the clip on the third DIMM socket from opening completely. If you have a longer card like Voodoo 5, however, it could very well interfere with all of them, though I haven’t verified that. The bottom line is, this is an annoyance but not a very big deal. If you’re careful, you can still put a DIMM in the third slot without removing the video card.

Other than the above issues, installation went smoothly. Be careful when installing the heatsink on your CPU; AMD’s Duron and Thunderbird cores are known for being fragile, and if you rush this step, you may end up with a cracked core and a dead chip. This isn’t the KT7A-RAID’s fault, of course, but it bears mentioning in any Socket A review.

Fire it up
Abit boards have long been known for having a large number of tweakable options compared to the competition, and the KT7A-RAID is no exception. At the heart of the tweaking power is Soft Menu III, Abit’s BIOS-driven utility for manipulating clock multiplier, bus speed, and voltage.

The Soft Menu III system allows for a lot of flexibility on this front. Multipliers from 5 to 12.5 are selectable in .5 increments, while bus speeds are selectable in 3-5MHz increments from 100MHz all the way up to 155MHz. Additionally, a selection called “CPU FSB Plus” allows for finer, 1MHz adjustments.

Of course, one still has to be careful of PCI and AGP bus speeds, which are based on the FSB clock. For example, one of the settings results in the PCI clock going as high as 44 MHz. It’s extremely doubtful your PCI peripherals (not to mention your AGP graphics card) would run at 33% over spec. Still, Abit tries to stick as close to the stock PCI speed as possible, and even the 155 MHz setting runs the PCI bus at 38MHz—not necessarily a sure thing, but at least worth a try.

In terms of voltage selection, the KT7A-RAID lets you alter both the core voltage and the I/O voltage independently, giving you more flexibility when attempting a dicey overclock. Other Soft Menu III options you can play with include Fast CPU Command Decode, CPU Drive Strength, and Force 4-Way Interleave.

I wanted to do some benchmarks that would highlight the advantages of the KT133A chipset and its support for the faster 133MHz bus speed. I chose SiSoft Sandra’s memory benchmark, Q3TA timedemos, and Content Creation Winstone 2001 to see how the newer hardware would affect synthetic, gaming, and business benchmark scores. Unfortunately, CC Winstone wasn’t cooperating, crashing during the tests. I’m hope to get to the bottom of it and add those results later.

A brief talk about methodology: All scores were run at least twice and the results averaged. The SiSoft SANDRA benchmarks were run using version 2001.3.7.50, and Quake III Team Arena benches used version 1.27g. Quake 3 settings were obtained using “Fastest” configuration, with 16-bit color depth explicitly set. Six combinations were tested: 100MHz bus with 100MHz RAM, 100MHz bus with 133MHz RAM, and 133MHz bus with 133MHz RAM. Each of these was tested with and without 4-way memory interleaving. With that out of the way, on to the results.

Memory performance
First up is SiSoft Sandra’s modified STREAM test. This is more or less a full-bore memory bandwidth test, so it should show off the 133MHz memory nicely.

As expected, we see a large jump in performance between the 100MHz and 133MHz memory speeds. Increasing the front side bus to 133MHz results in a comparatively small increase, highlighting the fact that memory speed is everything on this test. Note, however, the substantial improvement offered by 4-way interleaving, which manages 70-75% of the 133MHz RAM’s improvement at 100MHz. Of course, the interleave helps the 133MHz memory as much as the 100MHz. The moral? Interleaving is your friend.

Gaming performance
Sandra is good for a synthetic benchmark, but let’s look at the Quake 3 Team Arena scores and see how things look in a real-world test.

Once again, interleaving proves itself, giving an approximate 3 frames per second increase at the 100/100 and 133/133 level, and a 2.5 frames per second increase at 100/133. Though not as important as in the Sandra tests, memory speed is apparently more important to Quake 3 Team Arena than front side bus speed. Bumping the RAM to 133MHz resulted in roughly a 6 frames per second increase, though the increased front side bus still helped out to the tune of 4.5 frames per second or so.

For those interested, most of the KT7A-RAID’s overclocking functions worked well. I had no difficulty manipulating multipliers at will, and racheting down the multiplier on a “B” spec Athlon then cranking the bus to 133MHz went off without a hitch. Voltage manipulation also worked fine, though this particular KT7A-RAID had a tendency to feed approximately .07V more to the core than it was set for, so a 1.75V core voltage became 1.82V or thereabouts. Some research online confirmed that many of the boards do this. Abit apparently chalks it up to tolerances in voltage regulation and/or voltage monitoring.

One area that gave me trouble was nonstandard bus speeds. I’m still investigating why, but any attempt to boot above 133MHz resulted in the board not POST’ing. I can understand not hitting 155MHz or more, but it seems odd to me that the board would hit a wall attempting even a bus speed of 136MHz (with a PCI clock of 34MHz). I’m going to be doing more testing on this; perhaps I have an exceedingly picky PCI card, or perhaps I just need to play with some BIOS settings.

I will say that experimentation in this regard got a little frustrating, because if my tampering with the overclocking settings resulted in a machine that would not post, I had no recourse but to unplug the machine and pull the battery. The manual recommends powering up the machine while pressing the insert key to reset to defaults, but I couldn’t get this function to work. Abit also documents a CMOS clear jumper, but the jumper on my board didn’t seem to have any effect; only pulling the battery would clear my failed settings and allow me to start over.

Useless Serial Bus?
Now let me recount to you my experiences with the USB controller on the KT7A-RAID, and believe me when I say they were far from pleasant. It all started out well enough. I read along in the manual and installed all the appropriate Windows 2000 drivers, including the Via USB filter driver, and plugged in my plethora of USB devices. Everything worked fine for a while.

Then, suddenly, my USB was just gone. Oh, the controllers were still there, mocking me from Device Manager, but all four ports steadfastly ignored anything I plugged into them. Thinking perhaps some sort of strange conflict had arisen, I tried several devices one at a time, to no avail. Finally I decided to see if anyone else was having this problem.

Whoa, Nellie. Take a look at this Usenet search on Deja.com or Google Beta Groups or whatever the hell they’re calling it this week. That’s over 3,300 hits on “KT7A* USB”.

Houston, we have a problem.

What’s so frustrating about the situation is that there is no easy answer. The KT7-FAQ (which also applies to the KT7A and both models’ RAID incarnations) has a question devoted to this problem, with no fewer than sixteen recommendations to try to fix the problem, and another ten “reported solutions.” One of the latter has ten different criteria that must be fulfilled exactly before USB will work.

Even more frustrating is the fact that the solutions seem to have little relation to the problem at hand, and in some cases, they contradict one another. Enable Enhance Chip Performance. Disable Enhance Chip Performance. Make your memory settings slower. Make your memory settings faster. Install Via’s USB Filter Driver. Don’t install Via’s USB Filter Driver. The list goes on and on.

The bottom line here is that USB on the KT7A-RAID is apparently very flaky. I would imagine that at least some of the myriad of contradictory solutions stems from the fact that USB just randomly stops and starts working. I tried many of the solutions listed in the FAQ with no success. Then once after a reboot, my USB began working again. I had made no changes whatsoever to my configuration. The problem was fixed for exactly one reboot, then it came back again.

I’m not certain whether this problem should be blamed on Abit or Via, but it exists in the KT7A-RAID and you should definitely take it under advisement if you’re considering this board. After spending a lot of time trying to resolve this issue, I’m just going to bite the bullet and buy a $20 PCI USB card. I’d been resisting that just on the principle of how silly it is sticking a USB card on a board with four USB ports, but I’m sick of messing with it.

It appears there are also a number of people who have no problems with the USB on the KT7A-RAID, so I can’t tell you to just order up the board and a USB card as a package deal. But you might want to browse for PCI cards now, so you already have one picked out.

That annoying USB problem notwithstanding, I had a very positive experience with the KT7A-RAID. Granted, my bus speed overclocking experiments were short-lived, but I have serious doubts that the board is to blame for that.

Meanwhile, the board has a laundry list of great features. Abit’s Soft Menu III made fiddling with multipliers, bus speeds and voltage child’s play, but Abit also provides plenty of more detailed settings to help eke every last bit out of your processor. Expansion options are fantastic, with the number of PCI slots bordering on the ridiculous, and an ISA slot thrown in for good measure. RAM capacity should satisfy, well, everyone. If you need more than 1.5GB of RAM, you might want to consider therapy. Other cool Abit touches like the nifty treatment of the North Bridge are just icing on the cake.

Bottom line, if you want a great KT133A board with plenty of tweaking potential, you can’t go wrong with the KT7A-RAID.

The Tech Report - Editorial ProcessOur Editorial Process

The Tech Report editorial policy is centered on providing helpful, accurate content that offers real value to our readers. We only work with experienced writers who have specific knowledge in the topics they cover, including latest developments in technology, online privacy, cryptocurrencies, software, and more. Our editorial policy ensures that each topic is researched and curated by our in-house editors. We maintain rigorous journalistic standards, and every article is 100% written by real authors.