I pretty much agree with your evaluation of the performance of CPUs. But the situation does not appear boring or unusual to me. Probably the anticipation and extended waiting for the A64 made it seem like it was going to be spectacular. That is the nature of extended anticipation. But it is usual for CPUs to be introduced at modest improvements, if any. If they did it any other way, they would be wasting the resources of the corporation. A bigger jump always involves higher development costs.
And it is definitely not boring to get a great performing CPU for round about $100. We've had a great ride, as consumers, while the economy and prices have been depressed. That is beginning to turn around. Therefore the new generation of CPUs will seem expensive for what they do. But that is just returning to the norm back when there was a good economy.
Expect it to be a drawn out process, but SOI will become increasingly better adapated to mass-produced CPUs, and the clock speeds available are going to escalate. Remember the speed of the first Athlons compared to the present, and expect a similar run for A64s. (When Athlons hit 1GHz, it seemed like they were pushing it.) Just assume the AMD engineers planned the A64 for that kind of extensive future. Why? Not because they are extra super-duper engineers, or because they are more idealistic at AMD than Intel. Because it is the cheapest plan that garantees AMD a continued future. Intel can afford to soak up development costs on processors that are ultimately a dead end, and switch to an alternative when it tops out too quickly to be competitive. That is too expensive for AMD, with one fifth the sales of CPUs. Even the A64 is an said to be an extension and elaboration of the previous Athlon conceptually. That is designing with a long future in mind.
Forward looking designs have a hard time starting to roll. Anybody remember the original 60MHz Pentium plain? Everybody knew that Intel intended for them to be 66MHz, but initially it realistically didn't quite make it, not in quantity. Pentiums were over $1000 when introduced, and didn't drop fast. A generally better performing 486DX4 100MHz cost $120, AMD version. However the 66 Mhz Pentium had near double the performance in Intel's rating system, so the pricing wasn't quite as outrageous as it seems. Intel's benchmarks were very heavy on things people didn't do much at that time, and that was mostly because previous CPUs were so inadequate to the task.
But if you don't have a forward-looking plan, you ultimately lose. Intel steadily filled in the pieces of the plan and got the Pentium on track. People now take for granted doing the tasks that once were so specialized, simply because CPUs easily do them. You automatically get the capability tossed in for free with CPUs that cost under $50. There never was any huge demand for these capabilities. Nothing that would justify a large price tag for very many people. Yet there is a demand for them in the sense that people use them. They are available now practially for free because initially there were a small segment of buyers that WOULD pay extra for them.
I see the fallacious "no demand" arguments presented endlessly. My first HD was 60 megabytes. (It cost about about $250, and the controller card that was required cost $100.) That is mega, not giga. Believe me, there was "no demand" for 60 Gig hard drives then. There was a small demand for drives that did actually exist, maybe 200 megabytes at the time, although not a lot. So how did we end up with 40G drives for $40, if there was "no demand?"
My second computer had 32 kilobytes. (about $500.) That's kilo, not mega. The total address space was 64 kilobytes. So there was "no demand" for 512 megabytes. Believe me. It would have cost $8,000,000. Evidently, even though there is "no demand" for a product, it comes into being somehow, and I wonder how that may be?
Although it is true that there is no demand for things far beyond anything that exists, there is a small demand for things that push the limits of things which do exist. There is a progression that moves the things which exist to a lower cost, and simultaneously brings into reality things former beyond possibilty. By arguing that bleeding edge technology is useless and pointless (which is nearly true), and therefore should not be put on the market (which is false), people would stop the progression.
IMO, 64 bit addressing is past due. Although we could get by with just a couple more bits of address in the next few years, it is easy to go all the way to 64 bits given present technology, and by doing it that way, we have a consistent system that will last into the indefinite future. That makes the progression to ever larger memory easy. In order for 64 bit style programs to be around when over 4G memory becomes so cheap it would be common, there has to be a real 64 bit processor in production yesterday. The lack of that 64 bit processor for the last couple of years has been holding back the progression.
Facebook
Twitter