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 : LSI Corporation -- Ignore unavailable to you. Want to Upgrade?


To: getgo234 who wrote (9563)2/7/1998 8:00:00 PM
From: shane forbes  Respond to of 25814
 
getgo that makes good sense to me.

---

Something more about that Brocade and Fibre Channel and LSI (which funded part of Brocade):

techweb.com

In the fast-channel category,
Brocade Communications Inc., working with LSI Logic Corp., embedded
Rambus logic in the Silkworm Fibre Channel switch. LSI now offers Rambus
interfaces as a standard element of its Fibre Channel ASICs and ASSPs.


---

If I recall the Red Herring article correctly Brocade is doing quite well thank you in the nascent FC market. And it looks like LSI is the chip supplier. Of course Brocade is peanuts right now - but growing fast.



To: getgo234 who wrote (9563)2/8/1998 6:11:00 AM
From: shane forbes  Respond to of 25814
 
getgo: here's some stuff from the financial times article and how INTC may be going after the STB market:

Even as Intel shifts gears to address the "basic PC" market, the company is
also aiming to ensure that its chips are used in a new generation of television
set-top boxes. These units, which are expected to sell for about $200-$300,
will enable cable TV viewers to receive digital television pictures and internet
services and they represent a potentially huge new market.

While lower-cost PCs and set-top boxes may boost Intel's sales, analysts
predict that the company's profit margins will decline as it moves into these
very cost-sensitive markets.

---

And something about Sun's Java chip

Embedded microprocessor markets are growing quickly in size: companies
such as Motorola and IBM are targeting this market with low cost versions of
their powerful PowerPC microprocessors. There is also a move to add Java
processing capabilities to embedded microprocessors and microcontrollers.
Sun has licensed its PicoJava processor core to several big chipmakers
around the world.

Adding Java processing capability to a microprocessor does not take up much
space on the chip but it allows systems that use such chips to be able to
quickly process programs written in the computer language, without needing
an operating system and large amounts of memory.

----

And about SOC market size

Analysts estimate that the system-on-a-chip market will be worth about
$15bn by 2001 from about $4bn this year. The leading suppliers in this
market include companies such as IBM, LSI Logic, VLSI Technology and
Lucent Technologies.

----




To: getgo234 who wrote (9563)2/8/1998 6:19:00 AM
From: shane forbes  Read Replies (1) | Respond to of 25814
 
getgo: more stuff stolen from the FT article

---

And the difficulties involved with SOC design

But designing and manufacturing a system-on-a-chip is one of the most
challenging projects facing semiconductor companies. The reason is that
combining different functions on to the same chip requires sophisticated
software design tools that can also simulate the operation of the finished chip.


Manufacturing such chips also requires the latest and most expensive chip
manufacturing equipment and then there is the testing of the chip, a huge
challenge in itself.

On top of this, there are the marketing challenges, explains chip design
consultant Ron Collett, president of Collett International. "Producing a
system-on-a-chip is extremely complex, expensive, time-consuming - and
risky. You need to identify a market and a product that can be produced in
very high volumes to make it work. And it is exceedingly difficult to determine
what will be a high-selling product two years into the future.
"

(wow! Could not have said it any better if I tried to! In a couple of sentence the very essence of the problem! And don't we know it! But also note that ramping up volumes takes time and all LSI's current design wins should be going gangbusters in a year to 18 months.)

----

As discussed here earlier the problem is crappy design tools

There has been a design gap for some time," points out senior analyst Bryan
Lewis, head of Dataquest's Asic research program. "Companies would be
able to design system-on-a-chip products with much higher numbers of
transistors if they had the design tools available
."

But creating a system-on-a-chip from an IP library is not as easy at it seems,
warns Wilf Corrigan, chairman of LSI Logic, one of the most successful
system-on a-chip companies. Speaking at the IP Forum conference in Japan,
last October, Mr Corrigan said that LSI has consistently underestimated the
complexity of system-on-a-chip designs despite its extensive experience in this
market. He estimates that high-end designs have taken five times more
engineering man-hours than first estimates.

----

Long run potential

Even though there are significant challenges ahead, system-on-a-chip products
are being produced that push the boundaries of current technology and will
result in lower prices for a wide range of consumer electronics devices ranging
from wireless communications devices to entertainment systems and PC-type
devices.

---




To: getgo234 who wrote (9563)2/8/1998 6:38:00 AM
From: shane forbes  Read Replies (1) | Respond to of 25814
 
And yet more useful stuff

Tomorrow's silicon will merge dozens of today's single function
semiconductors on a single silicon chip - only the most complex systems will
need more than a couple of chips - and many products, such as GSM digital
telephones, will have just one. This move towards integrating memory, logic,
and other circuit types on to one silicon slice to create systems-on-a-chip
represents a key industry trend as the 1990 draw to a close.

Aside from Motorola, SGS-Thomson Microlectronics, LSI Logic, National
Semiconductor, VLSI Technologies and many other companies are already
targeting this market. As a result, system-on-a-chip sales, which currently
amount to only $4bn, or 3 per cent of the world market, are expected to total
$70bn by 2001 - triple the size of today's microprocessor market - and 26
per cent of total chip sales.

(notice that 70 billion is a LOT bigger than the figure of 15 billion in the previous post... Someone's drinking tequilla and eating the worms as well... Or more likely different definitions of SOC... Or to steal a phrase I heard recently "Dataguess" estimates...)

---

And finally:

The real key to effective use of DSPs will be their integration into
system-on-a-chip type products. These will combine DSP cores with other
chip functions, all on the same chip.

"It is very difficult to build such integrated chips," says Nathan Brookwood,
senior chip industry analyst at Dataquest. "The reason is that debugging and
testing such chips is difficult and time-consuming because you have to perform
so many different tests."

Despite these challenges, TI, Lucent, Motorola, National Semiconductor, LSI
Logic and others are pushing ahead with their system-on-a-chip plans.

Those companies that are successful will be able to offer powerful chips which
include DSP, microprocessor, memory and logic functions on a single chip at
relatively low prices. This will spark a boom in the manufacture of a wide
range of low-priced digital consumer electronics products.

I read somewhere that while LSI is definitely a superpower in digital, it might be weaker in analog. I wonder if this will be a bad "developing situation" for LSI at the expense of the stronger analog companies like NSM and ADI. Then again I recall having seen a big article about how the digital LSI presented lots of analog papers at some conference a year or so ago - good anecdotal evidence of developing strength perhaps.

---

If anyone wants to read all the articles (and as getgo attested they are very good) try:
ft.com

Thanks getgo for the "most excellente" suggestion to check this out!

---