I agree that Intel has more money and seems to have the advantage in process technology. But, it doesn't seem to be doing better anyplace else. For example, AMD used to be a full generation behind Intel. The AMD K5 (in spite of its name) was not really a Pentium class processor. I guess AMD's first shot in the arm was its acquisition of Nextgen and it seems to have kept up the pace ever since. The result has been that with less money, resources, and engineers, it has caught up with Intel.
Intel is clearly slipping. This was clearly evident when the Pentium 4 turned up slower at the same clock than the older Pentium III. I realize that Intel is pushing the Pentium 4 very hard right now, trying desperately to get a little more speed out of it. Sort of like a jockey whipping his tired mount in the home stretch. The important difference is that the Athlon 64 will still be able to run a few more laps while the Pentium 4 will be exhausted and fall behind. I'll say again that Intel will not be able to pull another generation out of the Pentium 4 architecture.
I agree about marchitecture but I'm puzzled that you didn't mention the obvious ploy with the 800mhz front side bus. For example, a 3.5Ghz P4 should be happy with dual DDR 266 and and a 533mhz frontside bus. A 4.5Ghz P4 should run well with dual DDR 333 and a 666mhz frontside bus and only a 5.4Ghz or faster P4 would need dual DDR400 and an 800mhz frontside bus. I don't think that anything faster than 4Ghz will be available before Athlon 64 begins shipping. So, this seems more like window dressing instead of necessity or any real performance gain. The only way this makes sense to me is if Intel plans to continue producing the P4 past 5Ghz in competition with Athlon64. This just doesn't seem very likely.
On consumer satisfisfaction with system power.
I also keep hearing about how everyone is happy with the current level of computer power and that is why sales are down. I don't think this is true. Sales are down in the U.S. because sales of all major purchases (like cars, homes, and large appliances) are down. Computers are still relatively expensive and something that families can put off if money is tight. It is much easier to spend $150 for a low end DVD player or high end game machine than spend $600 for a new, low end computer system (and a midrange $1000 system is completely out of the question). The demand is still there; the economy isn't.
This argument is easily bolstered by considering the common level of consumer purchase. The mainstream level used to be a full generation behind. For example, few people bought Pentium computers when they first came out. But that is no longer true, Pentium 4 systems were bought as soon they were available. Contrary to the notion of consumer satisfaction, consumer purchase has actually shifted toward more power, not less. In short, consumer desire for more powerful systems has grown faster than computer system power.
Anyone who is older than 25 should remember this. Ten years ago, a top of the line system cost $2,500. Factoring in inflation, a top of the line system today should cost about $3350. Except they don't. A really nice system is more in the $1500 - $2000 range. I paid $670 for a Commodore 64, 5 1/4" floppy drive, and a 13" composite video monitor. I paid another $120 for an external dma memory unit. That was the low end around 1984/85. That would put a low end system today at about $1400. So what has really happened? The consumer market has essentially dropped the low end. What used to be the low end is now the ultra cheap systems for $600 or less. The midrange has moved down to the low end and the high end has moved into the midrange position.
Typical consumers today purchase systems almost at the cutting edge of performance. One would think that if it were true about consumer satisfaction then most systems sold would be the very cheapest systems, in other words, the lowest possible power system. I have seen no evidence of this.
Brent Rehmel
Email address supplied but no spamsters here
----> Hello
The AMD K5 _was_ a Pentium-Class Processor. It was a superscalar design (more than one execution unit), used advanced branch prediction, speculative execution etc.
The K5, one of the first x86 CPUs that decoded CISC into RISC instructions, was clock for clock faster than the Cyrix 6x86 and Intel´s Pentium. Unfortunatly the K5 failed to achieve high enough clock speeds to remain competitive with the Pentium.
In many ways, the K5 is sixth-generation, but its performance level is held back to fifth-generation levels due its low clock speeds.
See this and this, etc. Regards
Horst (email address supplied)