Max, unfortunately you didn't answer my question.
Jim and I seem to agree that demand in the low-end of the market is pretty high. If you recall, low-end demand is the reason why Intel underclocks its processors from time to time. AMD has established a name for themselves by selling CPU's for $100 and helping to create the current sub-$1000 craze. Because of this, they seem to be unwilling to sacrifice that in order to move on to higher margins.
Why? Capacity. That K6-3 can't come for free, you know. Its larger die size and lower yields means that for every K6-3 they sell, they'll be sacrificing two K6-2's that could have been sold. This is great for ASP's and profit, but it also gives their new OEM customers the shaft if too many K6-2's are sacrificed for the K6-3's. It's different from pushing the K6-2 to 400 MHz, since that's just an improvement in bin splits and tweaked processes. If a K6-2 can't run at 400, AMD will just set it at 380, or 366, or whatever. If a K6-3 can't run at 400 MHz, that either means it's marked as a dead part, since AMD wouldn't dare sell a K6-3 that runs under 400 MHz, or a K6-2 with on-die L2 cache disabled, which means a much lower margin because of all that wasted silicon.
All of the above also applies to the K7 ramp-up.
Jim mentioned that Motorola can help with capacity, but I'm still wondering why Motorola would want to if it means AMD will take market share away from Apple's iMac. IBM could also help, but AMD already gave them the boot earlier this year. What do you think, Max?
Tenchusatsu |
