Dan, Re: "It's been posted here by Intel folks, generally as a part of "we will buy you" message. 2 are 10K wspw (or more), at least 2 are 8k wspw."
Link please? Although I don't entirely doubt your comment, I'm not sure if you are placing the right WSPW numbers with the right facilities in question. Since I have no idea, I am certainly not taking your word for it.
Re: "Yes, and the others of the 7 are .13 or .18 moving to .13. Not to mention Intel's 10 or so other FABs that are used for chipsets, com, flash, etc."
You are talking about 3/7 fabs that I mentioned, and like I said, one of them is brand new, and the other two are old development fabs (hence the names D2 and D1B - the former name for Fab 20). I don't know if the've ever run .18u wafer starts, but I'll check for any press releases.
About the 10 or so other fabs, I don't think there are that many more, but what's the point - it doesn't prove your argument. You seem to put it in as an after-thought, or maybe just to nit-pick.
Re: "AMD makes other parts too: with two families of embedded processors and a networking products division - and while they are more focused on CPUs, they don't have 10 other FABs, either. AMD is also doing development work at its production facilities, as well as being in the middle of transitioning Dresden to .13, Dresden to SOI, and Austin to flash."
Regarding AMD's commodity chips, the first thing is how do you know these are produced in Austin or Dresden? Didn't AMD have a third fab that they recently shut down? Did that production shift to their other fabs? If it did, the results will show up this quarter, not in previous quarters, which we need to use as a basis for comparison. In other words, knowing what AMD can produce in Q3 helps us to compare to Intel, but current changes, process ramps, flash transitions, and possibly a transition in commodity chips are all recent things that will show up at the end of this quarter - Q4.
Re: "Available for sale output from Dresden won't be fully ramped until Q2 of next year, either. AMD hasn't been spending nearly $2 Billion per quarter to ramp its FABs, either (closer to a tenth of that). And production at Austin is winding down. Probably something like 1.5 million mobile Durons and 2.5 million desktop Durons from a facility capable of producing 6 million parts."
I'm not so much interested in how much money has been invested to keep production going as I am in getting to the bottom of how many units can be produced versus how many wafer starts are made (which will give an approximation of yield). Intel obviously has many more plans that they are investing in, besides just transitioning their current fabs to .13u. Fab22 is a brand new fab, and it seems like that was money well spent. Fab11X will come online next year, and that will also be another brand new fab. Intel is refitting Fab17 in Massachusetts with the money, and that will be producing Intel CPUs for the first time. Intel's spending is obviously going into new facilities and the complete renovations of others, so of course they are going to spend more than AMD.
As for Durons out of Austin, I believe you are still using Q3 for a reference, and like I just mentioned, Q4 may be very different.
Re: "Let's face it, that's never stopped any of us"
No, Dan, it only seems to not stop you or pgerassi. The rest of us like to point out when we don't have enough information to pass off a reasonable result, but you pass off your guesses as facts, and expect people to believe them. I suppose I should reference this comment in the future. It might be quite handy in reminding you that you often try to come up with phoney numbers out of limited data.
Re: "Are ramping, one up, the other down."
Which only proves my point that Q4 may be very different for AMD. Stick to Q3 for comparisons.
Re: "FAB25 is winding down, but available in case CPU demand suddenly recovers."
You are still going off-topic. Note that "winding down" will result in changes in Q4 (depending on how fast it's actually "winding down"). The fact that it may be available if AMD changes their mind is irrelevant to this conversation.
Re: "I'm trying to show that the two companies have roughly similar yields - Elmers "AMD yields are disastrous" comments notwithstanding - and that if either company has an significant edge, it's probably AMD, which is pretty much inevitable since AMD's average die size is quite a bit smaller."
You have no cohesive connection between the data that you provide and the conclusion you are trying to make. If you stick to Q3 numbers, like I've been trying to say, you won't have any argument for Dresden transitioning to .13u and SOI or Austin transitioning to flash. While these things may have started in Q3, they hadn't been announced as progressing very far. If AMD sets records this quarter, with everything that you claim as going on in the fabs, that may be a huge accomplishment. We'll see if they can hit that target.
wbmw |
