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: Saturn V who wrote (81718)	12/2/1999 1:53:00 AM
From: THE WATSONYOUTH	Respond to of 1572381

Re: "The ISSCC 90 abstracts describes a IA-32 processor at 1GHz at room temperature. Given the fact that Willamette has not taped out,this paper has to describe a Coppermine or its derivative. This means that Intel has had Coppermine running at 1Ghz RT for a few months. Will this affect your guess for the Intel's MHz schedule ?" In my experience, what you can show at a conference can be very different than what you would ship as product to a customer. Since I predict Intel will eventually reach 933MHz as a shipable product with Coppermine, they certainly will have 1GHz parts. A very reasonable 7.5% guardband on a 933MHz part gives a raw Fmax of 1003MHz. If Intel demos this 1GHz part and indicates it includes an appropriate guardband, meets 10yr gate oxide reliability (no jacked supply voltage), meets all hot electron reliability constraints,meets all device parametric constraints, meets any/all power constraints,boots and runs all applications reliably, etc. etc. etc. then I'd say they demoed ONE shipable 1GHz part. For a paper, NONE of this is required except that it run at 1GHz. Then the question would be the binsplit at 1GHz. So, I would fully expect some 1GHz parts in some of the early test runs. One of the things you determine in these early runs is whether the design is functional throughout its modeled range and whether there is any yield loss as a function of reduced channel length within its intended operational range.So, these real fast parts are available very early on. It usually takes some time to refine the process and develop sufficient confidence to run at or near the fastest process limit. So, this does not necessarily change my timetable guess. Then again, maybe I'm wrong and am just underestimating the process. We'll see. Did they say how much beyond 1GHz the part was??? THE WATSONYOUTH