Ok, deze info staat al twee dagen op JC's site, maar is zo goed dat ik het toch nog ff post. Ik heb geen idee of onderstaande info betrouwbaar is, het klinkt in ieder geval wel lekker:
Oooohhh.... Got a really neat email just a few minutes ago. Contains info from Prudential Securities, an analyst, regarding AMD's near term goals. I'm reading it as I report on it so I'll just bullet the interesting stuff:
-at this early stage in the new K7 processor ramp things look very good. K7 units are good, yields appear to be very good, and for the first time AMD may actually lead Intel in the 32-bit x86 arena from an architectural standpoint.
-We believe AMD has shipped already between 1K-2K K7s and is set to ship in excess of 40K for the second-quarter as a whole. This is much better than our conservatively modeled K7 5K units for the quarter.
-Assuming a ramp similar to the K6, the K7 could hit 350K-400K units in Q3 and 1 million in Q4. yields are very good at the targeted 500 MHz, 550 MHz, and 600 MHz speeds at 0.25 micron.
-we believe that some K7 samples much higher than 600 MHz are available, possibly up to 700 MHz (likely using 0.18 micron). We feel this boosts the confidence of PC OEMs that engage early on with AMD as it demonstrates a "real" product roadmap.
-[K7] performance should exceed that of an equivalently clocked Intel PIII. It is not clear by how much at this time, however the company appears to be very confident with the "significantly faster" qualifier.
- If this holds, K7 ASPs would be well above $200 in the second half of the year and longer term enhances AMD's ability to meet its overall $100 processor ASP target, which is $10 above our modeled $90 ASP in 2000.
- AMD is positioned to offer the PC market four processors for three general segments: the K6-2 for the basic PC, the K6-3 for low-end desktop, the K7 for mainstream desktop, and the code named K7-Viper for the high end desktop or workstations.
- We believe Compaq, IBM and Gateway will be early adopters of the K7.
- We believe that Flash memories are coming back and this clearly helps AMD given that the company has nearly a 20% exposure in this area. High density Flash pricing is going up and we expect visibility here to increase substantially.
JC heeft ook nog een posting over de K7 preview bij Firingsquad. De benchmarks in die preview waren erg teleurstellend, maar waarschijnlijk ging het om een oude revisie van de K7 waar nog veel bugs in zaten, waardoor nogal wat functies noodgedwongen gedisabled zouden zijn geweest.