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: Paul Engel who wrote (46612)	1/19/1999 1:06:00 AM
From: RDM	Read Replies (3) \| Respond to of 1572807

Computer simulation have changed chip development timeframes before and after first silicon. After first silicon the bugs of a simulated chip are mostly reduced to those that resulted from conditions that were not simulated, otherwise they would have been fixed. In some cases simulation errors occur because the models do not depict realitiy, but this is rare. Due to simulation there are fewer errors and fewer turns of silicon prior to product. This is apparent to anyone that has seen it done both ways. In the 70s very few chips were simulated to any large degree. It simply too much time and resources to simulate more than basic building blocks. More RAM and faster CPUs still do not allow anywhere near simulation of who CPUS, but it is possible to simulate large blocks at a time. In the 70s you would have been warranted in calling a designer a two faced liar to claim that any complex chip worked on first silicon. In the late 90s it is not unusual that a complex chip can be mostly functioning on the first pass silicon. Occasionally a large ASIC may even be 100% functional. However, there many be many steps for a CPU to transition from "mostly functional" to production worthy. To say that because most of the simulation is done before the first pass silicon there is not any effect on the development time after first silicon seems to me to be poorly thought out and erroneous. However, to state that "NOBODY starts production of a major CPU until they have firm design wins and orders on the books." is silly. AMD will go into K7 production with or without a prior major OEM full commitment if the part works and provides the performance expected. The stakes are just too high and OEM commitments are not binding anyway. To speculate about the major OEMS is a little silly as well. It is clear from the K6-2 that AMD can attract major OEMS. If AMD delivers superior parts many major OEMS will come. Is it possible that AMD now has a mostly working K7 and is starting to run small production lots? Yes it is possible, though early. It is also possible that they will have to rework the wafers from the first runs with late mask level changes. It may be close enough for a pilot run of 500-1,000 wafers through ($1-3,000,000).