Tony,
I'm curious about where you got your info on relative yield numbers:
I don't know how good LSI is in the harvest, or yield, area. IBM is still #1, Intel 2, probably Fujitsu next, maybe then Hitachi, TI, Motorola next. I have no idea where LSI would be. Why don't you ask? Well, semi companies guard their yield numbers like gold, unless they are always great. I don't know if this helps.
Is this your feel from experience in the industry or is any of this published somewhere?
Re Shane's question, my feeling for yields is that they are generally in the 80-95% range at wafer sort for most products, depending on the complexity of the design (i.e., size of the die). (And therefore things like Intel's MPUs would be at the lower end and simple things like op amps and A/D converters would be at the higher end.) They don't seem to depend so much on the complexity of the process. If they just switched a chip from 0.35 um to 0.25 um without doing anything else, yes the yields would get lower, but there is optimization done and the yields usually end up about the same, in my experience. I would think that Intel would be able to get higher yields than 75-85% at wafer sort, especially since they are able to bin out the lower performers as slower-speed processors, but I don't have any inside knowledge about their yields.
G.P. |
