RE:It is unfair to compare the cost of cutting edge technology to technology that is farther 'down the curve'!
My point is that:
(I don't know the actual numbers, this is very off the cuff but it should illustrate my point and should not be too far from what really happened)
The 386 was 100% faster than a 286 and cost 2X
The 486 33 was 90% faster than a 386 and cost 2X
The 486D266 was 85% faster than a 486 33 and cost 1.8X
The P 75 was 80% faster than a D266 and cost 2X
The P120 was 75% faster than a P75 and cost 2X
The P166 was 40-50% faster than a P120 and cost 1.5X
The PPRO 200 is 100% faster than a P166 and cost 3X
The PII333 is 40% faster than a P233MMX and cost 2-3X
The PII400 is 10% faster than a PII333 and cost 2.1X
I am sure there is some one out there that could put the exact numbers in here. That is not the point.
The point is that the new chips at introduction cost a lot. After a short time, 3-6 months, this premium is reduced, but often a new chip is introduced in the same period.
That trend is consistent. It always has been.
Where there is a down trend is the advantage the new chip has over what was cutting edge 6-12 months prior.
I see this is a major contributor to the decline in desktop pricing. Now with the high-end not really improving, Combines with aggressive pricing from IBM and CPQ, Server pricing is falling in the same way desktop pricing is falling. On the server side, the pricing decline started 9-12 months later. DELL still gets 65-70% of its revenue from the desktop. CPQ and IBM both agree that the days of 40% margins on servers are over. Both expect high 20s to very low 30% margins on server products going forward.
Y'all want to compare everything to a 286 or an XT or a 4004.
Today, a PII300 wastes 35-50% of its CPU cycles either trying to guess what comes next or waiting for the Beef.
What percent of CPU cycles did a 486 Waste / go unused?
What percent of CPU cycles did a 386 Waste / go unused?
(I am guessing the number is lower than 35%, but I don't know so this one could come back at me hard)
Also, for a Person using a big spreadsheet, going from a P90 to a P166-200 made a big difference. Well going from a P200MMX to a PII300 might make some difference, but going from a PII266 to a PII450 won't make a very noticeable difference like the P90 to the P166 did.
Because of this, the replacement cycle is going to lengthen. And when the upgrade comes, it will be to much less than the cutting edge because for half or less of the money you can get 90% of the performance. More importantly, in most computing needs, the difference in performance can't be noticed.
Jim |
