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 : Altera -- Ignore unavailable to you. Want to Upgrade?


To: Stu who wrote (1411)12/3/1997 8:20:00 AM
From: Charles Halley  Respond to of 2389
 
The following is article about ALTR from recent Forbes...

Chips that change their spots

By Srikumar S. Rao

WHAT DO YOU WANT in your microprocessor-versatility or speed? It used to
be a law of semiconductors that you could have one or the other but not
both. A versatile Pentium is not specifically designed to run Microsoft
Word or to paint pictures on your personal computer screen, but it can
do both. A cheap, specialized chip could be designed to do either task
10 to 100 times as fast, but it couldn't do both.

So it is that the world of silicon chips divides itself into two camps:
the all-purpose ones that power desktop computers, and the fast,
specialized, cheap ones that run cellular phones, answering machines or
car engines.

Now comes the tantalizing possibility of combining speed and versatility
on a single chip. The chip acts as if it were hard-wired to do just one
task extremely well, but it can also be rewired on the fly to do some
other task just as fast.

These adaptive chips go officially by the name of field programmable
gate arrays. What do you do with one? One thing you can do is use it to
run a complicated electronic appliance like a printer or a telephone,
reconfiguring it as you go to keep up with changing technology or
consumer preferences. Hewlett-Packard did that cleverly in designing a
gadget that gooses the performance of desktop printers (see box).

Adaptive chips have been around since the 1980s. (In an earlier, simpler
incarnation, they were called programmable logic devices. These are
still being used in many applications.) Why are they catching on now?
Several reasons, say John Villasenor and William Mangione-Smith,
professors of electrical engineering at UCLA who do pioneering research
in the field.

"Two years ago," says Mangione-Smith, "we had chips with 13,000 gate
equivalents; now we have chips with 85,000 gate equivalents. The number
will continue to increase as manufacturing improves."

A gate is the fundamental unit of logic in computing-for example, the
command to feed back a 1 if either input A is a 1 or input B is a 1, but
not if both are 1s. From such elemental "if" commands are built complex
arithmetic calculations, violent videogames and programs that do your
income tax returns. "A gate is roughly analogous to a molecule in
physics," explains Villasenor. "You put large numbers together to handle
complex tasks like running Microsoft Word."

Adaptive chips are getting cheaper. The cost of a 10,000-gate chip that
was $1,000 five years ago is $30 today, according to Xilinx, a midsize
outfit that leads in this business.

What goes on inside an adaptive chip? Think of it as a stage whose sets
are constantly being changed. The sets we're talking about are groups of
logic circuits that handle some specialized task, like adding numbers or
comparing them. The pieces of scenery are activated by switching on and
off so-called passthrough transistors that control the flow of
electronic signals from one part of the chip to another. In turn, the
passthrough transistors obey the commands of a section of memory on the
chip that acts as a stage director.

There's a cost in energy, time and money to all this set-changing. If
you wanted to do only one task and do it the same way every time, you
would do better with a custom chip design-what the engineers call an
application specific integrated circuit. These ASICs use less power than
adaptive chips, are faster and, if you make large numbers of them,
cheaper. So whenever possible, manufacturers prefer to put ASICs in
their laser printers or cellular telephones.

But there are situations where the flexibility offered by adaptive chips
more than compensates for the disadvantages. Jean Calvignac, an IBM
fellow with IBM's Networking Hardware Division at Research Triangle Park
in Raleigh, N.C., gives an example. Four years ago his group was
designing switches for a computer network. The engineers selected a hot
new technology called asynchronous transfer mode. The problem with a hot
new technology is that standards and formats are up in the air. Using
ASICs would have left the group with a product failure if they guessed
wrong on the standards.

Using adaptive chips allowed IBM to confidently sell the switches to
customers. "We could always go back to the customer, even after the
switch was installed, and make changes," says Calvignac. For similar
reasons, adaptive chips are going to be big in the telecom industry.
Cellular phones are a rat's nest of conflicting protocols and vendor-by-
vendor variations.

"We encourage our customers to do their first production batch with
FPGAs," says Willem Roelandts, chief executive of San Jose-based Xilinx.
"This way they can start production straight away. When they are fully
satisfied with their design they can use us, or another vendor, to
convert that design into an ASIC. If they are manufacturing small
numbers of products, they may find it cost-effective to simply stay with
FPGAs."

Electronic Buyers' News, a trade publication, estimates that Xilinx has
29% of the market for programmable logic chips, and Altera, also based
in San Jose, has about 25%. AMD's Vantis and Lattice Semiconductor are
next in size, while larger companies like Lucent Technologies and
Cypress Semiconductor are beginning to grapple for market share.

Will adaptive chips change the world in ways that you and I notice? You
bet.

"Say you buy a $1,200 high-definition television," says Roelandts of
Xilinx. "Two months later they come up with a new compression algorithm
that makes the picture much better. You could plug your TV into the
Internet and download reprogramming instructions and bingo! You are
state-of-the-art again."

"Take a cellular phone," speculates UCLA's Mangione-Smith. "Suppose you
wanted to include a miniature camera and screen so you could see who you
were talking to and she could do likewise. The chip it now has can
process only audio signals. Putting in another chip and a power supply
for it would make the phone too bulky and expensive. A continuously
configurable FPGA could flip back and forth between audio and video
processing to do the job."

Talk about fast scene changes. The fastest of today's adaptive chips can
switch configurations 1,000 times in a second. Two years from now, say
the professors, chips could swap tasks 10 times as fast as that. They
foresee such applications as encryption, precise navigation and fast
object recognition.

Beams Villasenor: "You cannot even imagine the types of products and
applications that will arise."



To: Stu who wrote (1411)12/4/1997 12:53:00 AM
From: Robin  Read Replies (3) | Respond to of 2389
 
RE: Dips in the past have never
been this big and have never been linked to the earnings performance we're seeing.

my friend during july 96 this stock dipped from 80 to 26 then back to 130 pre split! This dip is noting in comparison to historical dips!