It’s always interesting (though sometimes depressing) to see how people interpret what you write.
Let me summarize the key points of the article:
1) The AMD/Via platform is not as reliable as it should be, and needs to be to significantly penetrate the business world.
2) We’re going to take a closer look at Intel products in the future to see if their future products are more reliable.
Why did we say such things?
Between Joe and I, we’ve worked with about fifteen Via mobos, a few for Intel chips, more for Athlons. Just for one example, of those fifteen, five have had poorly or
non-functioning multiplier adjustments (and the CPUs used have had no problems in other machines). We’ve had a couple Via Apollo Pros
that were OK, and maybe three KT-series mobos that proved to be at least OK from a stability standpoint out of the bunch. Even worse, we’ve
fairly often had two or more samples of certain boards, and one proved to be good, and one didn’t.
Far more importantly, we’ve been monitoring how people have been doing with these boards in emails and forums, and people are having problems out there. Sure, many, probably most of them
are due to user error, but many are not. It’s pretty hard to blame users outside of redoing a pencil job when the multiplier adjustment doesn’t work right.
Per AMD CPUs, Intel has set a very high standard. I’ve almost never heard of an Intel chip just dying after a fairly short period of time, and usually, those I heard about were owned by Dr. Frankensteins.
I asked a while back about people’s experiences with dead AMD chips, and got well over a hundred responses. Sure, some of them were due to crunched ceramics cores. Some were due to improper or poor heatsink placement. There were a few using insane
voltages that eventually died. I’d even say that constituted a majority.
But after removing those cases, there still were quite a few cases where the processor functioned for a week, or a month, or a couple months, and then just died. I’ve had tech support people in computer stores report significant numbers of dead CPUs in OEM machines running stock.
Some of these went rather spectacularly, with blown resistors strewn about. Occasionally, even those that had been functioning a while went that way.
I’m sure the vast majority of AMD processor with reasonable care will work perfectly fine, but first, you have to take a good deal more greater care in installation, and second, there’s no doubt in my mind that if you took 1,000 AMD and 1,000 Intel processors and tested them, more AMD chips would die. Not most, probably not even that many out of the thousand. Just more.
Overall, about half of you who emailed me generally said, “Thank God somebody said it” and the other half said, “You biased, lying, incompetent, maybe corrupt idiot; I’ve never had these problems.”
Let’s look at “bias” first. Yes, I am biased. I am horribly biased in favor of high-quality, reliable machines. That’s whose “side” I’m on.
Yes, Intel has had its problems with the 820 chipset; we condemned them for it back then. We didn’t say much about the 1.13Ghz chip because unlike the 820, Intel did the right thing early on, and nobody in our audience was interested in buying the damn thing at that time anyway.
We didn’t say, “You must buy Intel.” We know full well the PIII is on its way out, and Tualatin has a questionable life ahead of it. We haven’t looked at Willies much, and we’re kind of loathe to look at products that we know are going to be extinct shortly.
The problem we have is that people are looking to upgrade, but most want an optimal solution, and we see no optimal solution out there. Any selection right now has significant tradeoffs. Any selection.
What we will not, and never do, is seek safety in numbers by being on “one” side or the another, and shush any unpleasant facts about one side or the other for the sake of the “cause.” That would be a great disservice to you. If we see either company doing something bad, we’re going to say it.
This site started off as virtually an Intel-only site. For almost a year, we’ve been almost an AMD-only site. Right now, we’re inclined to say “a plague on all your houses.” So much for any claims of being on Intel’s payrolls directly or indirectly.
At What Point Do You Have A Problem?
The impression we’ve gotten is that most Athlon/Via boards perform well enough, but some don’t. Even after taking out user error and any and all fixes and workarounds, there are some that just don’t do what they are supposed to.
Do I know the exact percentage? No, but I can look in any forum and see lots of people with problems, including some that simply should not happen with properly functioning equipment.
A few have said, “If you don’t have a number, your comments are meaningless.”
OK, I know there are a lot of gay men and women in New York City. I don’t know exactly how many, nobody does. You came to New York once and no one sharing your gender propositioned you. Does that mean there are no gay people in New York?
Taking it a bit further, if you asked someone who lives in New York just how many gay people there are, that person will probably come up with a much higher number if he or she lives in Greenwich Village or Chelsea than if he or she lived in Bensonhurst, and neither estimate would probably be an accurate one for New York City as a whole.
It is a great fallacy to presume that if you can’t exactly measure something, it doesn’t exist. I may not be able to tell you if 10 or 20% of Via boards are defective, but I feel very safe in saying some people are having some problems, and it’s neither a freakish accident nor commonplace.
Let’s say as a really rough guesstimate that 90% of Via boards are good, and 10% aren’t. I suspect the percentage of “bad” boards is somewhat higher, but let’s say 10%.
A defective rate of 10% is far higher than it should be, but most people won’t have a problem, and will think you’re crazy or stupid or corrupt for reporting problems they don’t have.
We even had a few report putting together quite a few systems and reporting no problems at all with any of them. I don’t doubt the experience of these people, but if you have a 90-10% split, you’d expect to see that.
Just for argument’s sake, let’s assume that 90%-10% split for AMD systems, and 95%-5% for, say, Intel 815 boards. Under those circumstances, you’d have to call the Intel boards more reliable overall, but most individual owners won’t be aware of the difference.
Let’s say there’s two slot machines. One pays off 10% of the time, another pays off 5% of the time. If you pull the slot one or twice, can you tell the difference between the two? No. Even if you pull it a dozen or two times, you might still not come up with a single winner, or maybe you’d get five.
It’s only if you pulled it hundreds of times or watched hundreds of people pull it that you’d get a true idea of your odds on one particular machine, never mind two.
It may well be do to what programs you run. I have a fairly set routine of programs that I use. Nothing that you’d call horribly stressful, but my routine has a habit of bringing Via machines down (some more than others), but not Intel systems.
It could very well be my typical program set “likes” Intel better, and yours “likes” AMD.
I wouldn’t be shocked if I took some of your machines, installed my stuff, and your rock-stable machine turned to mush. On the other hand, you could take my machine, run your stuff, and it would turn at least the less bad machines into Rocks of Gibraltar.
The sad reality is, unless you run hundreds of systems under a wide variety of circumstances, neither you nor I can count on individual experience as a general predictor.
We’ve had experience with more boards than most of you, but again, we wouldn’t try to come up with an exact percentage based on our experiences, we don’t have a big enough sample. What we can say is that we know that these problems exists, and they’re legitimate for at least some.
Our point is that Persons A, B, C, D, E, F, G, and H can have no problem, and Person I can. Just because Persons A-H have no problem doesn’t mean Person I doesn’t have a legitimate problem. It doesn’t mean Person I must be doing something wrong. It also doesn’t mean 90% good is good enough.