InvestorsHub Logo
icon url

SemiconEng

09/21/03 12:28 PM

#13652 RE: Elmer Phud #13607

5500 WSPW /10 = 550 WSPW.

550 WSPW X 13 weeks = 7150 wafers.

99,999 / 7150 wafers = 14 die per wafer.



500 WSPW in a 5000+ WSPW Fab, for a new product launch, sounds about right..... initially. If your wafer start and yield assumtion is correct, what ramp rate, would you expect for the next few quarters?

Semi

icon url

dougSF30

09/21/03 11:29 PM

#13728 RE: Elmer Phud #13607

Elmer,

So if it's 300K in those 13 weeks, your reasoning gives 42 die per wafer.

What's the maximum # of usable die per wafer, assuming 100% yield? I realize the exact shape of the die will impact edge effects, but roughly speaking, what is this #?

Doug

icon url

yourbankruptcy

09/22/03 11:25 AM

#13761 RE: Elmer Phud #13607

Elmer, maybe there is another way to examine the situation, to check the phrase "AMD says they hope to ship 10s of thousands of A64s in Q4". Maybe they can ship more, but doubt that market needs that stuff in Q4? Maybe they are not counting mobile?

I wonder how many 3200 chips the market is consuming? If 2%, that's only 100K. The A64 should not be much more popular...

But in general, I agree, that sounds quite bearish.

icon url

yourbankruptcy

09/22/03 3:15 PM

#13794 RE: Elmer Phud #13607

Elmer, but where did you actually get that? I see thins story:
http://theinquirer.net/?article=11678

to mention 400K A64 planned for this year. Quite a difference. Pretty in line with normal yields and die size. With ASP $200, it's 80 million. Such a small number I think easy to be absorbed by murket.

icon url

ChrisC_R

12/27/03 9:10 PM

#21604 RE: Elmer Phud #13607

Elmer, re: "Ok, I did a little figuring. AMD says they hope to ship 10s of thousands of A64s in Q4. Let's give AMD the extreme benefit of the doubt and assume they really mean 99,999 ........
99,999 / 7150 wafers = 14 die per wafer."


And you accuse people of lying because they say you opine that AMD has yield problems. (LOL)

Did it occur to you that their prediction might be based on the market being demand limited, and they wanted to be conservative in their estimates?
icon url

kpf

12/28/03 5:44 AM

#21621 RE: Elmer Phud #13607

elmer

What else could it possibly be???

Inventories. Takes a couple of months to realign its structure while migrating to a new architecture. Requires a tough grip on prices, which requires a tough control of channel-inventories as well.

All that is happening right now. The latter things are obvious, the inventory structure alignments invisible- although there are hints how things are going there; introduction of Model64-3000+ before Christmas carries a message thisrespectively. Next message to be expected with the pricing-round early January.


K.





icon url

yourbankruptcy

12/28/03 6:01 AM

#21623 RE: Elmer Phud #13607

Elmer, did they explicitly said 10s of thousands? I recall Mageek published some under-carpet figures :

http://www.theinquirer.net/?article=11758

where he expects 450,000 A64's in Q4 and 1,700,000 in Q1.

If you take your calculations and replace 100,000 with 1,500,000 you will get 210 dices per wafer.