To: AlienTech who wrote (8524 ) 4/26/1998 9:08:00 AM From: LastShadow Read Replies (1) | Respond to of 120523
That's an interesting article AT, but not entirely accurate. I redesigned DEC's 2-layer wafer plant into a 7 layer wafer system for them when I worked for GE years ago, and the facility was only 9 months old at the time (that was in '83, I think). It took 2 days and only cost them about $2 million to build it. There are certainly a lot of R&D costs associated with the development of new architectures, and yield is a primary concern. With Intel chips, the evolutionary design of the 386/486/Pentium class hsipset up to about 266 MHz was relatively simple in comparison to jumps to 300 and then 400 and 450 (in a couple of months). The thing that absolutely befuddles me is why they don't take what they learned about on-chip and on-board cache and leverage that. Even the new 767 MHz Alpha station I am getting only has 3 memory slots - for which the most I can add is (3) 256 Mb chips - woefully inadequate without modifying the system. The one good thing that is going on is that there is now some standardization of components, and that will help the tier one suppliers tremendously. If you look at the 400 MHz HP, CPQ, and other major vendors systems, they are all using the same graphics cards, the same memory, the same system components. The commonality of those components is important in that rather that having some collection ju jour we are seeing leaders ramping their production up to minimize cost and fill the new orders. that will increase their profitiability and grow them. The tech sector will shrink in terms on major players worth following in the next few years, and that will create an interesting trading environment. I take issue with his last sentence - I see lots of traders doing worthwhile and useful things. Especially here. lastshadow