SI
SI
discoversearch

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)
INTC 35.10+2.3%Nov 19 3:59 PM EST

 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext  
To: Jules B. Garfunkel who wrote (37743)10/27/1997 3:14:00 PM
From: Jeff Fox  Read Replies (1) of 186894
 
Jules/Paul, re: Low end tech moves

Thank you for your post detailing the likely reasoning on Intel's capacity shifts. Your points are very likely to be substantially accurate.

On a couple of points on chip integration and fabs. (Point on fabs in a later post).

On chip integration for the low end, this does not necessary mean one chip for all system functions. Chipset integration is driven to minimize cost. Lowest cost is not necessary one chip due to I/O pins, package cost, technology interface and feature flexibility.

First I/O. The system must interface memory, I/O buses and Graphics, all which dictate a certain number of chip leads. The chipsets north bridge part serves as a lead "expander" for the processor providing more pins at a cheaper price than can be had on the "expensive" CPU package.

Then there are technology legacy voltages. ISA requires five volts, PCI requires 3.3volts and must tolerate 5 volt add-in cards. New processors operate at a moving target of about 2.5 volts and moving rapidly lower. Again a separate chipset component serves to translate from the low CPU voltages to the older, higher system voltages.

Certain functions are so cheap that they must be manufactured on old processes to make a profit. There are very cheap sources for a bundle called "super I/O", a part that has the floppy, timers, etc. It is never clear whether to swallow these functions or leave them as a separate component. If Intel does integrate these then they are getting virtually zero return on their engineering.

Lastly some new system functions are not yet stable, such as 1394 (superfast serial bus). These new functions need to be on high tech chips for performance reasons and on chips with a short life as the function must evolve.

All these reasons work against single chip I/O or sigle chip PC. Most attempts to date have gone quickly to the scrap heap of obsolesence, most never making a profit.

Now what will happen? There looks to be movement to combine the graphics and north bridge functions for low end machines. This may make sense.

DO NOT expect I/O functions in the CPU. This does not make economic sense. Unfortunately it often looks like it makes sense causing Intel, AMD (i.e. Elan) and Cyrix (GX) to try it. They have and will find that the market moves are too hard to track.

DO expect Intel to get it right. Look carefully at their low end offerings and I think you will find that they are damn near optimal for the market segment.

Jeff
Report TOU ViolationShare This Post
 Public ReplyPrvt ReplyMark as Last ReadFilePrevious 10Next 10PreviousNext