I’m calling this a first look because the initial prognosis is so disappointing I want a second opinion.
The Abit KT7 is supposed to be the chief competitor to the Asus A7V for socket A chips.
We got the RAID-less board, since we know what RAID and the Highpoint 370 controller can and cannot do.
If you want a list of features, go here. I’d rather tell you whether or not they do you any good. 🙂
Six PCI slots and one ISA slot, a handy configuration.
The Softmenu III controls let you overclock without setting jumpers on the mobo. However, other “features” largely negate this convenience.
There are far more tweaking controls than are available for the A7V.
There’s a temperature sensor built into the board (though you do have to gently bend it down in order to get the processor in).
The “active cooling” for the Northbridge looks to be a 486-style tiny fan on the chip. I’m sure there’s a heatsink under that fan, but this is not big-time cooling.
The particular board I tested didn’t have RAID, but unless you’re using very large files, RAID doesn’t do you much good. Additionally, the Highpoint 370 controller, while OK, is less flexible than
Promise’s controllers. For instance, it will not let you plug just one drive into the conroller and span it, which is probably something many of you would want to do.
For those of you used to BX boards, the Award BIOS has annoying quirks (this is also true for many other socket A boards). The biggest one in my book is the less-than-seamless recognition of IDE drives. The BIOS will often not recognize a drive, even though you’ve configured it to “Auto” unless you also hit “IDE HDD Auto Detection” and have the machine recognize the drive right then and there.
The BIOS does not let you easily boot from specific drives, but rather from particular hard disks. This could prove very aggravating when you have multiple OSs on a single drive.
The FSB settings were disappointing. While the KT7 does have settings that provide for a PCI divisor of /4, the first one doesn’t come at 124Mhz or even 133Mhz, but rather at 136Mhz. This means that unless you could manage 136Mhz, you have to overclock your PCI bus quite a bit to get a decent FSB overclock.
Unfortunately, at least my experience shows that thinking of such speeds was fantasy.
This won’t be a surprise to veterans, but don’t expect any real instruction on what those zillion tweaking controls mean from the manual. The manual is usually cryptic on this and on more important points. For instance, it gives you five options for memory, but no intelligible description of what each means.
Performance? It was middling. Nothing bad, nothing great.
It is not a stellar overclocker
The Abit board promises improved circuitry to handle power-hungry chips and provide increased stability at high speeds. Some other reviews have claimed that the Abit does extremely well overclocking FSB.
Not this one.
The chip used was a Duron 600. This chip has also been used with an AZ11 and A7V. It’s not the best overclocking chip in the world, but it’s usually been pretty stable at 850Mhz in both water and air-cooled environments.
For my testing, a modified Alpha was used, along with silver thermal grease, so cooling was pretty good. The brand new UL BIOS was used.
At 850Mhz, the Duron was racking up 45C (or a bit over 110F) doing nothing. Do something, and temperatures went up to about 120F pretty quickly. Then programs started crashing.
The primary victim of these crashes was a JPEG viewer. Now running a copy or two of this program and tossing up some pictures is not exactly deSade territory, but the Duron would jump up 5C, and suddenly one or more of the programs died. This does not happen at lower speeds (nor happens with the water-cooled TBird at 1Ghz), so I can only blame what is really not a ton of heat by Duron standards.
The point of this is not that the KT7 did badly, but that it didn’t do appreciably better than the other boards, beefed-up circuitry notwithstanding.
The real problem I found was trying to overclock the FSB. There’s no point in having a million settings if they don’t work, and this thing wouldn’t boot at 110Mhz no matter what I did to it. It would do 105Mhz, though, big whoop-de-do.
Just so you know, the supporting cast was a Matrox G400 card than never had a problem running at 88Mhz on a BX board, 256Mb of Micron CAS2 PC133 RAM, and an IBM 75GXP hard drive. None of these components should have minded a 10% overclock, and actually, it doesn’t look like they did; they never got a chance to object.
Athlon chips don’t die gracefully when pushed too far. They either work, or they don’t, at all. Push it past what it wants to do, and you get a blank screen, not even a sign from the video card of any kind of life. This takes a little getting used to.
It Recovers From Overclocking Efforts Badly
What takes even more getting used to is how badly the Abit boards responds to abuse. The FIC and Asus boards we’ve tested would die, but recover on the next boot once we became more reasonable in our demands.
Not the Abit. Granted, Abit boards have usually taken a while to reset themselves, but this one was worse than the usual. Hitting the reset button did no good. It took a minimum of two and usually three cold boots for the board to reset itself.
This in and of itself is not a reason not to buy the board, but people overclocking tend to get a little tense at the moment of truth, and if you don’t know this is “normal,” it’s not exactly reassuring.
Problems with the platform
I’ve used or worked closely with the Abit BX6, BH6 and the AOpen AX6BC BX boards. I’ve also used the Via AX63Pro.
Over the last couple months, I’ve gotten familiar with the Asus K7M and A7V, the FIC AZ11 and now this board.
In all honesty, I find the two sets in two different leagues. Only the A7V approaches the first. It’s not that the Athlon boards are bad boards, they work, they’re just not as good as the others.
The first set just seems to operate more smoothly, and gets less perturbed by operations.
Let me give you an example:
Remember the JPEG viewer? Well, that program slows to a crawl during a download. By crawl, I mean move a picture or resize it, and it takes 5-10 seconds. I’m using an ISA US Robotics full-fledged modem, no Winmodem here. The machine had 256Mb of RAM, so it had plenty of room to breathe.
There always been some of that with the other machines, but not as bad. In the case of the TBird, it was pretty bad with only 128Mb of RAM running Windows 2000, but giving it 256Mb pretty much fixed that. This was happening with 256Mb in Win98, and nothing else open.
We’re Getting A Second Opinion
It’s conceivable my equipment just doesn’t like the KT7 for some reason, so in a few days, I’m going to bring this up to Joe Thursday, and we’ll try it using different equipment. We’ll give it a second chance.
But for now, don’t run out and buy one expecting an overclocking wunderchild. It’s not bad, it’s just not great. Not even too sure that it’s as good as the A7V. Before you buy, wait for that second opinion.
I think at this point, given AMD’s efforts to shut down overclocking, you have to ask yourself if that is the straw that breaks the camel’s back. Is it worth a bit better performance to pay for a $150 mediocre motherboard running a very hot chip that you now have to check for missing pins to determine whether you can overclock at all?