There’s a review over at Tom’s Hardware, and the results illustrate something we’re seeing.
We had the expectation that the second generation of KT133 motherboards would build on the lessons learned in the first generation, and be the major source of improvement, not the addition of ATA100 support.
We have not found this to be the case with the Soyo and MSI boards, and neither did Tom’s. They do OK, but you don’t see the level of improvement you’d might expect. Not extraordinary or even significant performance improvement, but something
that tells you, “This makes the first generation boards obsolete.”
After dealing with the MSI board, I can’t say you should buy this instead of the A7V. It’s not a bad board, but it’s cranky. Make a change and it throws a fit for a couple boots, then goes back to work grumbling.
Can I live with it and tolerate it? Yes. Could you? Probably, especially if you leave things alone.
While working with this board, I also put together that Christmas system with the A7V. Not saying installing the K7T was the worst experience of my life, but putting together the A7V was much smoother sailing. I’ll grant you that in the case of the A7V, I used jumpers rather than BIOS, and
continually setting jumpers in a box is a big pain; but playing multiplier roulette isn’t a whole lot better.
What’s the difference? Maturity.
The A7V’s been out for a while. There’s been more time to fix problems. There’s been more time to tweak performance here and there.
Maybe the MSI board will look better than the A7V in three or four months. Maybe. Will MSI keep working on tuning the product, or will they go on to the next project? What will any of these companies do with current and future products?
A New Criterion for Selection
Motherboards come out half-baked. That is unfortunately a fact of life, and very unlikely to change; there’s too much pressure to get the product out the door.
So if you decide to be a pioneer and buy the latest, you’re going to get arrows in the back, no matter what you buy. The question then becomes, “Does the company at least provide medics?”
I don’t even think you should even take anybody’s word in a product review as a reason to buy a product, not even ours. Use them to get rid of the obvious losers, and look at features sets, but beyond that, they’re proving to be useless because they don’t catch problems like quality control or many day-to-day problems.
Just for an example: I’ve been upclose and personal with three socket A boards that get very favorable reviews: the A7V, the KT7 and now the K7TPro2A.
The A7V eventually became a good, not great board. Give them that one by default.
The KT7 is a loser: resellers keep telling me the recent Abit boards are awful for quality control. They get a high RMA rate, and they can’t even get the things to work. I get told they only stock them because people insist on buying them. You shouldn’t have to play crapshoots hoping you get a good one. Nonetheless, you read the review sites, and it’s the most wonderful thing on earth.
The K7TPro2A isn’t quite as bad, but only because you can eventually bully it into working. Think it doesn’t crash? Like hell it doesn’t.
So my personal scorecard gives these choices a 1 out of 3 (and that “one” reluctantly). One out of three ain’t good.
In all fairness, that’s probably as good if not better a score than anyone else’s. In all fairness, a big chunk of the reason why I find two out of the three less than wonderful are outside the purview of a review of a single sample. But in all fairness, if current practice doesn’t get the job done, then don’t hide behind the practice, change it.
Don’t Blindly Follow Numbers
Much, if not most of the time, only a couple percentage points separate all but the outright performance losers. What the hell difference does it make if you get 129 versus 126fps in Quake? Can anyone out there honestly tell they can tell the difference between 129 and 126? Can anyone honestly say, “I was fragmeat when I got just 126fps, but at 129, I rule?”
On the other hand, you’re not fragging anybody while you spend time trying to get a balky system to work; you’re benchmarking 0 fps then.
But figuring out quality is hard. Right now, you have to do a lot of reading in newsgroups or forums to see what the pioneers are saying, and whether or not they know what they’re talking about.
In comparison, counting fps is easy, but counting fps isn’t good enough.
Does Your Motherboard Have A Mother, Or Is It An Orphan?
What is a much bigger factor in your long-term satisfaction is not what a mobo does when it goes out the door, but what the engineers do with it after it goes out the door.
The A7V had some real problems coming out the shoot. So did the KT7. Asus kept flinging BIOS revisions out left and right; got it more or less right after about the fourth or fifth try, and has improved a bit since then. In contrast, Abit’s done little, just came out with a BIOS revision after three months, and one that didn’t promise a ton.
(Not so sure MSI following Asus’ lead is necessarily a good idea. Downloaded and updated BIOS found on German site to 2.01. MSI Everywhere Else came up with version 1.9. Tried installing that, it won’t. Not fun having to reset three times during the attempt at a BIOS flash; nor am I the only one getting this.)
The next time you buy a motherboard (and DDR’s going to represent a much bigger change than going from a 686A to B Southbridge), you’re going to have to make some decisions.
If you don’t want to go through pioneer days (and you can’t count on reviews, based on past history), you’re going to have to sit on the sidelines for a while, probably at least a couple months.
If you don’t mind being a pioneer, you’d better scout ahead a bit and figure out which companies are most likely to clean up after they rush a product out too quickly.
We need to pay less attention to what’s new, and more attention to what’s good. We’ve got some ideas on this, which we’ll be implementing fairly shortly.