Larry: re: The evidence...>
At this point the timing to create a new design is less than the time to revise an old one. The problem is in manufacturing.
The logic and construction of the chip is set in software, not in masks, photo-resist capabilities, etc. The expected capabilities of masks next year and photo-.. machines next year predicates the design to allow the purchase & setup of equipment.
The question is -- Is five layer construction yet more profitable than three layer?(It will be one day) -- Or an alternative question. -- Can we start building 5-layer chips and get there first? -- get the knowledge without passing through 4 layer design.
At the moment I think that the response is "I'll take my profits up front, thank you" from INTC & the market. INTC has made enough money this year to buy the entire graduating classes in Physics, Math, and Electronics.
INTC will have to go to 4, then 5 layer construction - then 10 & 20 over the next 50 years. At the moment, the question seems to be that last year was not the time to add more layer complexity.
IBM made enough money, in the last 50 years to have a near monopoly on brains through the '70s(with ATT). They continue to spend appropriately on research. Do they have the process under control? - Beats me!
Some money will be made - or lost on the K-6, K6-3D, etc. Wars of the near future will hinge on the memory access and video output questions.
For those who like AMD, don't forget CYRX - IBM manufacturing could not keep them. For those who don't, don't forget INTC -- IBM, then AMD manufacturing kept them.
tgptndr |
