Carl, Re: "I don't think that "traditionally", Intel has been inferior in performance, at least if you include reliability in performance."
I didn't mean that. I think Intel has always lead in performance, especially against AMD's K5 and K6 processors. The K7 was the first time AMD showed they could make a processor that was faster than Intel's flagship. People are still wondering if this is AMD's one-trick pony, or if it's the start of a new kind of leadership from the company.
What I had meant was that Intel could demand a premium, even if it's chips were inferior in every way. Since they weren't this has put AMD between a rock and a hard place for many years. On the other hand, if you look at the introduction of the Pentium 4, especially the 1.3GHz introduction in January, one could say that Intel's chips were inferior to AMD's in every way, but they still sold for a premium.
"Intel's better overclockability was related to the higher margins that Intel built into their chips.... That changed with the Athlon.... People were hot clocking it 40% in 1999."
I think it's tempting for people to equate overclocking with headroom in a processor core, but I don't think there is as much of a relation. AMD seems to have labeled their cores according to batches run at different areas of their fab. Batches that have the label AXIA have always been highly sought after by overclockers, because of the perceived ability for them to hit high clock speeds.
After a few months of their introduction, however, you'd see BB messages of members who have lost the ability to overclock their machine. I used to hand around the Sharky Forums, which is an interesting place to watch the experienced and youthful overclockers share their first hand experiences. On most occasions, I've heard of AXIA components losing their ability to overclock, and this is usually a sign that the core has been pushed past its limits, and as a result, it has been partially broken.
There have been suggestions that Intel puts lower specs on their parts, simply because they are more strict, and they have a better idea of where their chips can still be reliable. Meanwhile, AMD's upper bounds tend to destroy the chip over time. While their relaxed specs seem to allow them to have temporary amounts of headroom, over all they have to cherry pick their parts to ensure this.
Many people have wondered why AMD has not introduced a 1.5GHz chip, since overclockers have been able to easily hit that speed in the past with normal air cooling and voltage. I used to answer that it was a bin split problem, and although AMD may be getting a fair percentage of chips able to clock at 1.5GHz, they are still unable to get enough to satisfy their current distribution of parts. That is, AMD tries to sell mostly to the upper bins, and a 1.5GHz part wouldn't satisfy demand unless it had excellent bin splits.
But I also think that their hesitation is based on the fact that other parts of their manufacturing process are already stressed high enough, and a 1.5GHz speed grade wouldn't have the highest reliability, despite being able to easily clock that high temporarily. My point is that there are so many tricks in manufacturing that it can be easy to confuse some people into thinking that they are far ahead due to overclocking ability. But if other factors are involved that can shorten the life span of the processor, then those high speeds won't do any good. The fact that AMD has delayed their 1.5GHz processor for so long suggests to me that they may be having trouble getting reliable yields at that speed.
"Consumers have a strong tendency to keep buying from what has worked for them in the past. Heck, I know people who buy Ford trucks because their daddy always did. But over the long run, the population as a whole does adjust their perceptions."
This is undoubtably true, but if it weren't for Ford continuing to advertise great products, they probably would have left long ago. Intel has a great advertising engine. People see them everywhere, and it puts AMD at a disadvantage, since they don't advertise at all. This is still going to be a large factor in the future. Some people may turn around and try the Athlon right now, but if Intel comes out with a superior part in the future (or even the near term future), then their marketing will eventually bring these people back.
"On the other hand, I see AMD as an investment for a fairly long time."
I only see AMD as a good investment if they can both increase market share *and* increase margins. Whether it be by cost cutting, or rising ASPs, I think AMD has a long way to go before I have any confidence in their upper management. The people who run AMD are much more likely to take this battle personally, and go for their own gains, then to do what's best for the company. I know this is a serious thing I am accusing them of, but boy is there a lot of evidence to suggest it's true.
"Right now AMD has almost no sales to business. That percentage has lots of room to grow."
That's not true. Recently, this thread presented marketing data that showed AMD in 38% of small businesses. I think there is quite a lot of market share for Intel to gain back.
"Stuff like PCWorld selecting 8 of 10 of their best PowerPCs as being AMD."
Yes, things like this would make excellent marketing fodder for AMD. However, I wouldn't count on publications like this always favoring AMD in the future. PCWorld's WorldMark 2001 is a benchmark that seriously favors AMD due to including programs that are extremely out of date, and weighing their results highly towards office productivity. Since PCWorld ranks their results according to this test alone, I expect AMD to have the consistent advantage. On the other hand, this benchmark will probably change over time to include newer software, at which point AMD may lose their advantage. Just MOH.
wanna_bmw |
