We now more or less have three DDR chipsets out there: the Ali, the AMD, and the Via.
Don’t know which one to pick? Don’t worry, no matter which one it is, you’ll find some website that agrees with you.
You can go here for a laudatory review of the MSI Via DDR board, and if you want to find out more, just click on the MSI ad on top of every page of the article. Or go there to see the MSI ad before you read the article, and if you need a few bits of creative reasoning as to why the Via is better than some of the other alternatives (you’re all going to be running DDR at 200Mhz, aren’t you?), go here.
If you prefer an Ali board, you can go here to read all the reasons why a particular Ali board is a great choice, and then buy it from the author.
If you prefer an AMD board, you can go here and find out that pre-release motherboards from one company are now better than those from the reviewer’s previous favorite company. BTW: If you want more information about that company, there’s a new ad on the front page to click through.
Same tests, different results
Different sites, different configurations, different benchmarks. Sometimes, though, a few do the same thing.
Cachemem, for instance, you can have your choice, here, too!
If you want to read that the AMD does better than the Via in read performance, go here. If you want to read that the Via does better than the AMD, go there.
Sadly for pro-choicers, both agree that the AMD’s write performance is better than Via’s, but if you want to believe the difference is only 24%, go here, and if you want to believe it’s 37%, go there.
If you want to see Ali get trounced by AMD in write performance, go here. If you’d rather have an Ali chipset representative beat AMD in the same benchmark instead, no problem, go there.
If you prefer SiSoft Sandra’s memory benchmarks (which are based on Linpack), if you want to believe the Via is slightly better, go here. If you want to believe the AMD is about 7-10% better, go there. If you want to buy an AMD board, and 7-10% better isn’t good enough for you, have no fear.
If you want to read the AMD is as much as 17% better, this is the place for you!
Let us charitably ignore the small impact differences in memory speed have upon performance in real applications.
I could go on and on and on.
Sure, there’s explanations other than evil for this. Somewhat different configurations, different BIOSes, different tweaks and settings. But if the results vary so much due to these, just how useful is the benchmarking?
The Winner Is . . .
People run tests, and Guy A finds Product A a little bit better than Product B, and Guy B finds Product B a little bit better than Product A. Only thing Guys A and B agree on is that Product C is no good, but don’t worry, C lovers, we have Guy C to defend the Faith.
Centuries ago, people pulled out their Bibles to back their position. Now, they just pull up websites.
If you want Via, here’s your chapter and verse ace-in-the-hole, “Overall, the VIA KT266 is the fastest chipset currently available for the Athlon.”
If you want AMD, here’s your Holy Writ from Not Doubting Thomas, “go for Athlon plus AMD760 and NOTHING ELSE”
If you wish to heed the words of the Prophet Ali, “the ALi . . . appears to have the best all around performance”
If you wish to hear me, “The smaller the win, the louder the cry”
Making Mountains Out of Molehills
Bold statements for margins of victory that are usually 3% or less, and usually not all going the victor’s way, either. Can you tell the difference between a twenty-nine and a thirty second commercial?
You can say, “a win is a win,” but this race isn’t over. This is more like a snapshot taken half-way through a marathon, and the judges can’t even agree on who’s ahead now, though a few seem awfully sure about not only who’s ahead at the moment, but who’ll win.
After all, we must have a winner, even if we don’t have one. Don’t we?
We Don’t Need More Artificial Intelligence, We Need Some Natural Intelligence.
Tools are tools, not rules. If you blindly use the tool as a rule, you’re a fool.
Tools are very valuable, used properly. But they have limits, and can be dangerous when used improperly or blindly.
Tools can help you think; they do not do it for you. The tool tells you what; it doesn’t tell you why. The tools can say A does more than B; they do not tell you whether that matters or not. They tell you what was; they do not tell you what it could be.
The Courage to Say, “I Don’t Know”
A fool is always certain. He knows even when he doesn’t. A wise man knows when he doesn’t know.
But fools scorn the wise as fools for not knowing, knowing not that “I don’t know” can be wisdom.
Pride goes before a fall, and this is how the wise often fall before fools. To stay considered wise, they must become the fool for the fools.
Well, if I am wise here, this is no great accomplishment, but since I’d rather be called a fool than become one, let me
say what I know and don’t.
What I See and Don’t See
Whenever you hear loud discord about which is best, it’s a pretty safe bet there’s not much difference among them.
Overall, right now, the AMD board looks to have a slight performance advantage, an insignificant one in most cases. Right now is not forever. That could change two days from now. It might never change. We don’t know.
Motherboards are not stone, they morph. Even the same piece of silicon and epoxy can be made to do better, and as time goes on, newer, better versions of that silicon and epoxy can evolve.
The AMD and Ali chipsets have been around a while. They’ve matured and changed over time. The AMD chipset looks to be exiting the stage early, which may well mean improvements will slow or stop. The Via boards are just getting to the stage. They may develop quite a bit, but where and by how much? How much in BIOS and driver upgrades you can download? How much in new versions of silicon that you can’t? We don’t know.
Cost is a big factor to many. The AMD boards tend to be more expensive. There are reasons for that, is the extra cost worth it? We don’t know. Different people can come up with different answers.
Stability is a factor. The Ali boards seem to have more problems than the AMD boards. The initial fragmentary results look promising for the Via board, but it’s too early to call. Again, we don’t know.
Every board has plusses and minuses. Depending on what you value and how much you value them, you could rationally choose any one of these boards at the moment. That choice could look foolish a month from now, or it may not.
The AMD boards have a slight edge in realized performance. The Via boards have a longer future, but even if that future turns out to be better, you may not get some or most of the benefits of that future if you buy the firstborn. We don’t know.
Even more importantly, the most rational choice for many who have recently upgraded is no choice at all.
There is no clear-cut winner now. There may not ever be a clear-cut winner. To say loud and long that there is or will be is like political hacks looking at hanging chads.
Is This Whole Argument Moot?
Who will provide the processor for most computers? The answer is Intel, and will remain so for at least the next few years, even in AMD’s wildest fantasies, simply because no one else can supply the number of CPUs required.
Who will provide the chipset for most DDR boards? The answer is Via, simply because no one else can supply the number of chipsets required. AMD won’t, Ali can’t, at least not for quite a while.
This could well turn out to be a Betamax argument. If you can only find VHS, it doesn’t matter if Beta is a little better.
What We Will Do
We will look at Via boards. We will look at Ali boards. We will look at AMD boards if we see them having any kind of future in the marketplace.
We’ll look at them differently, though. We did not criticize because we think we can the same old job better. If we do the same job, some of the criticisms would apply just as much to us. What we have been saying is that the old job now only doesn’t cut it, it may be doing more harm than good, and that we should all be looking for a new job.
We are in a world where it seems like where 80-90% of the products are fine, and the rest have something wrong with them. For the average person reading reviews then buying something, that’s like playing Russian Roulette with one chamber loaded after you’ve been told the gun is empty. You’ll probably survive the experience, but you or your widow might not find ignorance bliss.
The sad reality at this stage is that we are all guinea pigs running on our own treadmills, and it’s too much a matter of chance right now whether you or I get a good or broken one. So long as there is a significant number of either out there, any review by anybody hurts anyone on the other end of the good/broken scale.
But at least we can find out what the guinea pigs are finding out, run around the treadmill ourselves with the equipment we get to try to help out, and finally relay what’s happening out there. That’s not an easy task, a lot harder and more time-consuming than putting a computer together and marking down some benchmarks, but until the situation changes, I don’t see what else we can responsibly do.
Be the first to comment