The previous version of the japanese roadmap (dated May 17)was very similar too. It showed 1.5 at the same time as 1.4, and Duron 800 in Q1 01, so there's been a little fine tuning but no major push-out of chips (except that 1133 seems to have been lost).
My opinion about the whole speed grade thing is that they had a contingency plan to bring out 1.3 and 1.4 this year, maybe even 1.5, but with the "non-participation" of the Pentium 4 this year, they reverted to the original plan as detailed on those pc-watch roadmaps. Also, as has been investigated here, I don't think anything was really promised except 1.5 Ghz in early 01.
The excitement seems to have died down about this but I already wrote it so here you go :)
There are only two relevant questions (IMHO):
1. Which way makes more money? I'm for that one!
2. Are they being less aggressive because of manufacturing problems that will bite them next year?
Short answer: I think stopping at 1.2 for 2000 makes them the most money, and I don't think there are any manufacturing problems (with the CPUs) that will cause problems next year.
Long, even more opinionated answer (skip if you don't care):
I think the answer to (1) can be found by looking at what I want for Christmas (being a somewhat representative computer geek). I would love a 1 gigahertz computer. 900 would be nice, but not SEXY. 1100 would be awesome, as would 1200, 1300... but the 1 gig barrier is magical. They can define it as the new standard, and will make the most money possible by leaving the 1 gig price as high as is feasible.
Two factors influence this 1 gig price - what chips they have at higher speeds, and what Intel charges. If they were selling 1.3 and 1.4 Ghz chips in December, you can bet the price on 1 gig chips have to be fairly low. Intel can't afford to drop their 1 gig prices by much, since it's the top of their line and would devastate their ASPs.
Ok, Chris, you may ask, why even come out with 1.2 Ghz? Well, Intel may make another attempt at 1.133, so 1.2 keeps the speed crown at AMD.
So it's my feeling that, in absence of Pentium 4, they're doing the right thing to stop at 1.2 for this year.
For (2), the issue is a little more cloudy. Clearly there are some issues that would benefit from some delay. The power (and cooling) requirements for the 1.3 to 1.5 gig Thunderbirds are high, making it convenient to stop at 1.2 for now (awaiting either Palamino or better motherboards). Also, it looks more and more like PC2100 is late to the party, so 1266 Mhz on AMD760 (and 1133) may be a non-starter for this year. But these aren't chip manufacturing problems. If they don't PREVENT the sales of the 0.9 gig to 1.2 gig chips this fall at good ASPs (and they shouldn't, since most computer buyers have not heard of DDR or any supposed plans for 1.3+ Ghz chips), they are not a big deal for Q4.
I wish I knew if the 760 was ready and just waiting on DDR memory. That's my biggest concern, but I think they can make a healthy amount of money without it.
My gut feeling, reinforced by people here and reports that 900 Mhz will be the lowest Athlon soon, is that the chip manufacturing end is in fine shape.
Well, I guess we'll see what they say Wednesday.
Chris |
