To: THE WATSONYOUTH who wrote (100886 ) 3/30/2000 1:35:00 AM From: chic_hearne Read Replies (1) | Respond to of 1571035
THE WATSONYOUTH Re: Yes, it would appear that Intel had to do this just to achieve 1GHz. I think this trick (combination of 25C lower junction temp plus 9% reduced BEOL resistance plus .05V Vcc increase due to better gate oxide reliability at 60C) only bought them about 75MHz. (IF) so, this means that without it, they only ship 933MHz Coppermines. Actually, assume the 1GHz shippable parts average say 1175MHz raw Fmax which allows for a minimum 12% guard band and some reasonable headroom past that. If you simply pro rate that 1175MHz (1.7V/60C/notch process) back to (1.5V/normal 85C/pre notch process) you would predict about 905 MHz raw Fmax which would ship as a 800MHz part. This is what I predicted for Coppermine prior to all these changes. (IF) this is correct, we will see no Coppermines (at least at the same ground rules) shipping above 1GHz (like 1067MHz). That would also mean that all the talk (SaturnV/Process Boy/etc.) about careful speed path optimizations / clever process tweaking / Process Boy/Design Girl cubicle all nighters was pure bullsh*t. I agree with Elmer. If Willamette is delayed, we may see a process shrink (maybe 15% - 20% area) from one fab. That would be even more desperate. I guess we'll just have to wait and see. Interesting times ahead. THE WATSONYOUTH Wats, You really know your shit, I'll give you that. In your profile, you don't give much info. Please do one of two things for me: 1. Tell us what your occupation is, etc. 2. Reply that you don't want to tell us. Please reply honestly. BTW, I work for the company "WATSON" started, do you also????? chic