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: grok who wrote (66602)	7/24/1999 7:46:00 PM
From: THE WATSONYOUTH	Read Replies (3) \| Respond to of 1574004

Re. <Yes, they've said that 650 MHz ships this quarter. But they haven't said how many.If most K7s turn out to be 500 MHz then more massive red ink with occur. We really need someone who can report on AMD's speed distribution. Can't anybody call someone they know and get reliable info on this?> Here's my guesstimate. Assuming the initial K7s are from 1.8V .25um process (Is this correct?), and 500mhz represents a "nominal process" (if nominal is below 500mhz AMD is finished / if nominal is above 500mhz, AMD is in good shape) then I would guess there is an approximate 400A delta in channel length between nominal and minimum channel length. For each 75A reduction in nominal channel length, I would expect a 40mhz speed improvement. With a solid design and good process control (good mask/reproducible lithography/good RIE/ et.) AMD should be able to target lots to within 100A of minimum channel length without significant yield loss. Thus, with very good process control, you could expect a good fraction of 650mhz parts if AMD chooses to run there right out of the chute. I would expect a VERY rapid yield loss at 700mhz or above. But they could easily produce enough 700mhz chips for demo purposes to hype the release. However, based on everything I hear, AMD's process control is in fact very poor relative to INTEL. This suggests larger within chip, chip to chip, wafer to wafer, and lot to lot channel length variations. With poor process control the 650mhz sort could be quite small. Alternately, if nominal is say 580mhz (AMD may target there in order to have very little product below 500mhz). This would improve the highest end speed to well beyond 700mhz and spell trouble for INTEL. You are right. Without inside informattion, you can't really say. But as soon as they really start shipping volume quantites, it should become obvious what they can or can not do. However, I have to believe INTEL knows exactly the capabilities of this design and AMD's process and will respond in the best way they can. THE WATSONYOUTH