William - Re :"1) existing .35 micron facilities can be converted to MMX technology in a two to three month time frame - still would not give INTEL the volume they need at present ?"
In fact, this is just the OPPOSITE case - the new MMX chips REDUCE Intel's output (at 0.35 microns) because of the large dies sizes:
Pentium (Classic) - 90 sq. mm.
Pentium MMX - 140 sq. mm
Pentium II - 203 sq. mm.
So, switching to the Pentium MMX cuts available die by about 30 to 50%, and switching to Pentium II from Pentium reduces available die by a factor of 2 to 2.5. Bad news.
Re: " you had stated previously that three to four facilities will be converted to .25 micron by the fourth qtr for two reasons - faster microprocessors ( 300 mhz and above ) and mmx technology . Would not the new die allow for doubling of the volume ouput of the new microprocessor thus abating volume allocation for mmx in the fourth or first qtr 98?"
A. These Fabs are being converted/phased in from 0.35 micron to 0.25 micron. I did not state that this will be completed by the fourth quarter of this year. It will most likely be early to mid-next year for full conversion.
The enhancement to stepper/scanners (purchase of additional ones), extra metal layer and CMP equipment, as well as the C4 technology - which requires new, wet plating processes - all these add up to new equipment, floor space constraints, and new processes to be adopted and perfected. It's not like going to Safeway and buying New and improved Tide instead of the old Tide.
B. Die size reduction - it could easliy be a 50% reduction - condisering the extra metal layer (5 vs. 4), .25 vs. 0.35 minimum feature size on about 6 layers (out of about 20 mask steps), and C4 technology which eliminates pad ring constraints.
Re: "PAUL if the MMX product imbalance is corrected could INTEL not
lower it price for MMX tecnology and basically put AMD & CYRIX in a tight corner in the first qtr for consumer MMX chips while capturing the high end with the new .25 micron chip ?"
Be careful here - Many AMD investors and analysts make this mistake about Intel. AMD and Cyrix/IBM are ALSO trying to get to 0.25 micron processes. These companies do not stand still - they know to do so will cause them to "get run over" (I think Lee Iacocca said that!).
Also bear in mind - AMD's K6 is 20% smaller than Intel's Pentium II TODAY because of their more advanced 0.35 micron process - 5 metal layers, C4 technology, local interconnect, shallow trench isolation, etc.
Intel will actually be playing "catch-up" to AMD as far as die size is concerned - although Intel's performance exceeds AMD's for the same process feature size.
Intel will have the advantage of a huge manufacturing capacity over AMD - AMD has but one Fab which is still not at full capacity - and is apparently having difficulty YIELDING K6 devices on their new process.
However, these yield problems will most likely get solved in due time - AMD folks have a lot at stake here - and therefore a lot of motivation.
Paul |
