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: Petz who wrote (82631)	12/10/1999 7:03:00 PM
From: THE WATSONYOUTH	Read Replies (2) \| Respond to of 1572746

Re: "If there's a wide variation in Fmax over a wafer, then a 500 MHz wafer may indeed have more than a few chips that will operate at 776 MHz, because the sigma of the normal distribution is large. The overclockers find a few of these chips and report it on the internet. With good process control, a reasonable within wafer variation (chip/chip) is +/-4%, within lot variation (chip/chip + wafer/wafer) of +/- 7.5%, and lot to lot(all data) of perhaps +/- 12.5%. So unless Intel's famous process control is a myth, this 750MHz @1.6V part which was shipped at 500MHz was purposely shot there and for whatever reason got down binned. I don't understand it either given the shortage of faster parts. THE WATSONYOUTH

To: Petz who wrote (82631)	12/10/1999 8:01:00 PM
From: Elmer	Read Replies (2) \| Respond to of 1572746

Re: "I thought that was obvious. If there's a wide variation in Fmax over a wafer, then a 500 MHz wafer may indeed have more than a few chips that will operate at 776 MHz, because the sigma of the normal distribution is large. The overclockers find a few of these chips and report it on the internet. (They don't tell you about the ones that just managed 550 MHz, do they?)" John, I'm confused. There is no such thing as a XXXMHz wafer in x86 processor land. While there may be variation across a wafer, the speed of any individual die is determined at final test and has nothing to do with the wafer. Are you thinking that some wafer level measurement determines the speed of ALL the die on that wafer? Not so. EP

To: Petz who wrote (82631)	12/10/1999 9:47:00 PM
From: Bill Jackson	Read Replies (1) \| Respond to of 1572746

Petz, From what the young boy said there seems to be luttle variation across the wafer. I now wonder if they intentionally start to make 500 Mhz parts and could also make 750 Mhz parts at will. Earlier i had thought that they got a scatter and tested and marked. But at 4% variation a 700 Mhz wafer will make all parts within +/- 28 Mhz of 700? I also wonder how they etch those thin point? Do they implant some sodium at an earlier stage in deposition, laying down a thin layer with a few extra sodium atoms in the mix? ( just a few) since it is well known that sodium in glass makes it dissolve faster in an alkaline etch bath? Probably those sodium atoms would make the SiO2 act funny, unless they deposited it only on the surface in the trenches where they wanted the extra etching by shadow mask type deposition? so it went away at the end. Does anyone know just how they do it within the process methodology? Bill