Paul, re:<Where did I get 2M (CXT core)>
First, chips with the CXT core first started trickling out in mid-November and non-CXT production continued after that, so estimating that 36% of production was CXT core is pretty reasonable. There was also a question by an analyst that since 400K K6-2-400's were basically made in the month of December only, was that the expected run rate for 400's from now on. (Answer, no with the speed path fix that number will go up.) But he implied that K6-2-400 production lasted more than one month, but not much more than a month.
The other way to estimate it comes from stuff said at the cc.
Sander's said 800K to 1000K were sold as 300 and 333 MHz and 400K were sold as 400 MHz. Also, the scuttlebutt is that the "reject" chips, the ones where the speedpath "mattered," would not run on 100 MHz bus, but were marked with the designation "AFR 66."
The 380 MHz speed is not as popular as 400, so maybe there were 300K of those.
The 366 MHz speed was only available for a short time and has pretty much disappeared. Maybe 100K of those, plus the non-CXT core could yield at 366 MHz as well.
The 350 MHz speed requires 100 MHz bus. There was an intimation at the CC that some of the "reject" chips would have run this fast, but had to be marked down all the way to 333 for an undisclosed reason. [because they couldn't handle the 100 MHz bus.]
So I came up with the following for the "bin-split" of working CXT core chips:
300 and 333 AFR 66: 900K
350: 400K
366: 100K (some of these sold yielded from non-CXT core, as well)
380: 200K
400: 400K
Why am I telling you all this? Can't you figure it out yourself?
Petz |
