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: Jim McMannis who wrote (26683)	12/12/1997 3:26:00 PM
From: Petz	Read Replies (1) \| Respond to of 1577893

Jim, RE:<What really is AMD yield problem> In general I agree with what Paul Engel said, specifically that AMD made a decision to use difficult process technologies on the K6 (0.35): "First, AMD uses a local interconnect at the gate/source-drain level that Intel does not use, even on their 0.25 and 0.18 micron processes - possibly because they investigated the difficulty associated with these steps. ...on the assumption that the long term advantages (smaller die size= more potential K6's per wafer, also less capacitance=higher speed potential) would outweigh the lower initial yield and potentially longer process optimization time. Its also clear that there were TWO yield problems -- speed bin yield and functional yield. As for speed bin yield, this problem is totally solved, since >50% K6-233's are being produced and the market for 200's and 166's is still strong. As for functional yield, AMD stated that this problem was a little over 50% solved in early November. I interpret this to mean that they got halfway to the yield they wanted (say, 70%) from the yield they were at (say, 25%). What is very significant is that the more difficult of the two yield problems is speed bin yield, according to Yousef, and that this problem is totally solved. Looking forward, I believe that because of the experience gained in the difficult local interconnect technology for the 0.35 process, the slow ramp up of production will not be repeated in the 0.25 generation. Furthermore, the benefits of local interconnect (reduction in die size) are apparently much greater for the 0.25 micron generation than for the 0.35. 0.35 generation............die size.....0.25 generation...die size AMD K6......................162mm....Mobile K6...........68mm Intel MMX...................120mm....Tillamook............80mm (The die size of the K6-3D will also be 20% smaller than the next generation Pentium II, the Deschutes.) In fact, a major contributor of the slow ramp-up for the K6 may have been its large die size which exacerbates defect problems. Yes, in retrospect, the design tradeoff for the original K6 made it a money loser in '97, but '98 looks like a different story. Petz

To: Jim McMannis who wrote (26683)	12/12/1997 3:35:00 PM
From: Kevin K. Spurway	Read Replies (1) \| Respond to of 1577893

Its extremely unlikely that IBM is making K6s. If they were, especially if it were without the consent of AMD, AMD probably would have told us about it by now, because it's material. In fact, there has been something in each of AMD's recent 10-Q's that specifically says that AMD believes the K6 is an AMD chip, not a NexGen chip, and therefore isn't covered by NexGen's agreements with IBM. If IBM were to "steal" the K6, as you suggest, AMD would have a duty to correct this information so it wasn't misleading. This is just my opinion of course, no guarantees. You never know with AMD. A year ago, nobody would have guessed that the K6 would see the market acceptance that it has seen. Of course, we also wouldn't have guessed that AMD's production problems would be so bad. If AMD can solve those problems, it has a bright future. For the first time in a long time, it has a product that competes with Intel's best, and if it can make its roadmap happen, this will be true for the foreseeable future. Can AMD build the chips it designs? I'm cautiously optimistic given the recent loosening in supply of 233 mhz chips, and AMD's historic manufacturing performance. Kevin

To: Jim McMannis who wrote (26683)	12/12/1997 11:36:00 PM
From: Paul Engel	Read Replies (1) \| Respond to of 1577893

Jim - Re: "Still with IBM pumping out K6's as well as AMD they will muddy the water..." I don't know what you are referring to. In August, AMD and IBM announced that AMD would sell partially completed wafers to IBM (all processing but C4 bumps) and IBM would then own the wafers, finish the bump process, and complete the package and testing and selling. BUT AMD STILL MANUFACTURES (95%) the wafers - NOT IBM. AMD's yield problems become IBM's yield problems! Paul