Yousef, your answer to "Why do any of them have to be used
for both the 0.35 and 0.25 um process?" is a lie.
In your response you state "it doesn't make financial sense to have a separate dedicated .25um process line running at low utilization. The smart thing financially is to utilize as much of the .25um equipment as possible making money running the .35um process.
Baloney, if AMD would save money doing it this way then that's the way they would do it, but they don't HAVE to. Why would anyone in their right mind switch masks, chemical composition, temperature and other factors every other day, when you've bought brand new equipment to optimize your new process on. There is NO shortage of equipment in FAB 25 that requires sharing of equipment between the 0.35 and 0.25 assembly lines. This is especially true when the 0.35 process has questionable yield (until November) and, in fact, the 0.25 process as optimized in Sunnyvale development fab gives more K6's per wafer than the 0.35 process.
You then say,
BTW, this is exactly what Intel does at their development Fab D1B...this [sharing of equipment] provides invaluable process learning.
Yet, you stated exactly the opposite for AMD:
This is always difficult because to solve process problems, one must run experiments, maybe modify equipment ... so it is hard to do both jobs (process turn-on and manufacturing) well. Also as chip demands change, new equipment for .25um will need to be brought in and old equipment "retired". This "juggling" of equipment and capacity by process type can lead to missing deliveries on either .35um or .25um. Both of these mentioned issues also make it hard to troubleshoot and fix yield problems.
Since the floor space and quantity of equipment in Fab 25 is sufficient to produce several different microprocessors with both 0.35 and 0.25 process technology, your claim that AMD has to "juggle equipment by process type" is exposed for what it is: the wishful thinking of an AMD-hater.
Petz |
