Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: Daniel Schuh who wrote (111738)	5/21/2000 6:02:00 PM
From: Elmer	Read Replies (1) \| Respond to of 1572033

Re: "Elmer, isn't .35u what the original Pentium Pro was fabbed in? Did Intel convert those lines to the BX instead of upgrading them when the PPro was phased out? The BX must have went into production at about the time the processor lines were going to .25u, not? It makes sense in a way, getting a couple more years production out of old tooling. This somewhat confirms a hunch I had, the chipset business is sort of a loss leader. Regardless of how much silicon area is actually in there, producing those 500 pin packages can't be particularly cheap." Their chipset business model is to follow behind the processors as they move to the newer generation processes and other manufacturing equipment which has been fully depreciated yet is still serviceable. It just makes sense. Why scrap what is still productive? The problem they ran into, IMO, was that the BX was so wildly successful and without a viable successor they had to keep an old process running at multiple fabs after it's useful life should have ended. It just lasted longer than the business model was intended to and the fab space had to be reserved for it. I think the BX will be approaching 200 million units shipped before long and that's a lot of wafers that could have been producing newer generation products. Furthermore test and burnin has to accommodate them as well. As for producing 500 pin packages, it's one of the arguments for going to RamBus to greatly reduce the pincount and all the associated design and test issues that result from so many pins. You can match RamBus bandwidth by just adding pins if you want to but the headaches that result may not leave you that much better off. With increased on-die L2 is memory bandwidth as all important as it used to be? EP