Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : Intel Corporation (INTC) -- Ignore unavailable to you. Want to Upgrade?

To: semiconeng who wrote (128208)	2/23/2001 5:23:22 PM
From: Tenchusatsu	Respond to of 186894

Semi, it's useless to argue with the AMDroids on die size. AMD has always enjoyed a die size advantage over Intel, even during the times of K6 and K6-2. But die size is only a small part of the overall picture. I know of chipset components which will have larger die sizes than Athlon. Big deal. Intel's die sizes are larger mainly because we can afford to have larger die sizes (except during the supply-side crunch of 2000). AMD needs to keep the die size small because they only have two fabs and won't have another one up for several years. Tenchusatsu

To: semiconeng who wrote (128208)	2/23/2001 7:57:31 PM
From: Dan3	Read Replies (1) \| Respond to of 186894

Re: So, these things that you say intel is going to "put back in", who said we're going to do that? If additional cache and the removed execution units aren't put back in to P4 for Northwood/Foster the game is over for Intel. But I'm sure they will be. P4 is in particularly dire need of a much larger cache. The present chip is limited to caching 2K locations (no victim cache, and 128 byte cache lines in a 256K cache) while its competition can cache 6K locations (a 128K L1 and a 256K L2 victim cache both using 64 byte cache lines). A 256K cache Athlon can cache more unique locations than a 512K cache Foster will be able to. Considering the more effective structure of its 16way L2 (against P4's 8 way cache), it will probably take a 1 meg cache Foster to equal the workstation/server performance of a 256K cache Athlon. As clock speeds increase, P4 performance won't scale without a much larger cache. And they're going to have to fix the problems that make it so slow on the mainstream code base. They'll do it, just as Pentium Pro's limitations were corrected by PII, but the result will be a larger die. Even on .13 Northwood and Foster will be pretty big, and a lot of wafers will have to be run through to get enough to allow Intel to hold onto a large share of the market. Since Intel is evidently building new FABs to run .13/copper, it looks like billions of dollars of recently built FABs will have to be written off. Tualatin will be competing with VIA and Duron - even if they price Tualatin at $25 a piece and drive VIA and AMD out of the bottom feeder market, it won't increase sales much. Total sales of $500 PCs will always be limited, and if someone is buying a performance motherboard, DRAM, hard disk, CDRW, DVD, and monitor, they won't settle for a VIA/Duron/Tualatin CPU even if it's given away free. Intel's going to have to make P4s, and they're going to have to make them big, and they're going to have to make a lot of them. Tualatin without .13/Copper can't compete with Aluminum Durons from Austin, while .13/Copper Tualatins should dominate it. But that's another part that has to come from this new FAB space - and that can't be manufactured by any of the existing plant. Tualatin can't compete with Palamino/Hammer so a lot of those big Northwood/Fosters are still mandatory. Maybe they can take those 6? 8? FABS they finished converting to .18 a few months ago and use them for flash - but flash needs quite a different process, doesn't it? No problem, a few billion more capex, a one time write off on the value of the FABs in their present state. But those expenses are starting to add up, aren't they? Dan