Comms chips

Network Processors: Easy Come, Easy Go

As fast as network processor developments get put on hold, other ones take their place, judging by the program for the Network Processors West conference, to be held in San Jose, Calif., next October.

The program lists no fewer than eight new startups:

  • Astute Networks Inc. "will provide the first look at its transport protocol processor, which can be used for TCP termination as well as upper-layer protocol processing," according to the conference blurb.

  • Seaway Networks Inc. will describe how its "Streamwise" technology for content processing works. A content processor can look deep into packets at Layers 4 through 7 for applications such as Web servers and firewalls. Seaway was founded by a bunch of former Nortel Networks Corp. (NYSE/Toronto: NT) engineers (see Nortel Folk Float to Seaway).

  • Propulsion Networks (no Website) will be unveiling its first product, which is expected to be a supercharged data-path processor suitable for OC768 (40 Gbit/s) speeds. According to information on an investor's Website, the startup has developed techniques to address the memory bandwidth issues -- how to get information in and out of memory quickly and efficiently. Investors include Athena Technology Ventures and Woodside Fund.

  • Israeli startup HyWire will introduce its first product, a search engine that enables searches at lower power than CAM-based (content addressable memory) search engines. Most network processor solutions require separate search engine chips from an outside vendor.

  • A mystery startup will also make its debut. Who could it be? Stealth-mode startup Dune Networks is a possible candidate. Another startup that has yet to reveal anything about its product plans is Teradiant Networks Inc. (see Teradiant Turns Up).

  • Two switch fabric startups -- Tau Networks Inc. and Zagros Networks Inc. -- will also unveil their first products. Packet-based switch fabrics provide policing and shaping of traffic flows and are an essential part of a complete router solution.

These new players are filling the shoes of a bunch of network processor startups that have quietly disappeared in recent months. These include:

  • Entridia, whose Website now boasts a sign saying “For all Entridia’s product lines, sales, licensing and other inquiries, please contact Stratigos Networks, LLC.” (See Entridia Offers 'Rhapsody'.)

  • Clearwater Networks, which has disappeared without a trace. (See Clearwater Unveils Services Processor.)

  • Terago Communications, which doesn’t appear to have responded to resuscitation efforts, judging by the nonexistent pulse of its Website. (See Terago Soldiers On.)

At least two heavyweight vendors -- PMC-Sierra Inc. (Nasdaq: PMCS) and Vitesse Semiconductor Corp. (Nasdaq: VTSS) -- have also canceled future network processor products (see PMC-Sierra Pulls Packet Silicon and Vitesse Drops Some Packets).

The bottom line is that there are still plenty of players in the network processor market. Investment probably tops $500 million -- a significant sum, considering the fact that the market itself is thought to be worth about $1 billion in total, according to Linley Gwennap, principal analyst of the Linley Group, a consultancy specializing in network processor technology.

So how many network processor vendors are there? A new report on network processors -- written by Simon Stanley, president of Earlswood Marketing Ltd., a U.K. consultancy -- published yesterday on Light Reading lists 17 established vendors and advanced startups (see Network Processors).

— Pauline Rigby, Senior Editor, Light Reading
http://www.lightreading.com Want to know more? The big cheeses of the optical networking industry will be discussing this very topic at Opticon 2002, Light Reading’s annual conference, being held in San Jose, California, August 19-22. Check it out at Opticon 2002.

Register now and save $500 off the registration fee. Just use the VIP Code C2PT1LHT on your registration form, and deduct $500 from the published conference fee. It's that simple!

docsisdude 12/4/2012 | 9:58:39 PM
re: Network Processors: Easy Come, Easy Go here i thought Propulsion Networks was going under ...
edgecore 12/4/2012 | 9:58:32 PM
re: Network Processors: Easy Come, Easy Go
Nice to see startups in this booming market...where do I invest?

Lastly, this market needs a longer runway and a few more engines to take off...last year the predicted market was roughly 1B...the actual total was 90M...Interesting!

Some reflections regarding NPU's:

-NPU's are still a huge question mark, very few engineers have confidence that they will live up with published performance numbers (especially with all features enabled)

-the developement tools are still very crude

-reference designs for NPU's are huge/bulky, expensive and hard to get a hold of

-if any tier one OEM decides to go ahead and use NPU (which, even though I have a fairly negative slant to my message, some are doing...they will look at NPU's from the key/financially healthy vendors...Intel, IBM, Motorola, Broadcom, Marvell, Motorola....AMCC...etc...taking a chance on a startup for a new technology such as NPU's is not a risk they care to take

Someone once wrote on LR, NPU's are here to stay, how fast/slow they get accepted is hardf to guess and what kind of consolidation will happen is hard to guess...


BobbyMax 12/4/2012 | 9:58:26 PM
re: Network Processors: Easy Come, Easy Go More start-ups spring up whwn the level of expertise need is rather low. A well known example is dotcom companies that were started mostly by H1B visa holders. This is also true of many EDA companies and chip companies. Usually there is no value in buying or selling these companies. It is because of these phenomenon, we as a nation, are not able to develop, any break through technologies.

In many instacers some ignorant VCs even fund their companies. Then they start looking for a company to buy them. The company that buys them hardly gets gets any thing out of it except headaches and cash drainage.

The same trend is visible in the internet, security, networking and semiconductor industries.
drone387 12/4/2012 | 9:58:26 PM
re: Network Processors: Easy Come, Easy Go Any idea when these two companies chips will hit the market? They should be ready to unviel their offerings shortly.
jamesbond 12/4/2012 | 9:58:26 PM
re: Network Processors: Easy Come, Easy Go Any idea when these two companies chips will hit the market? They should be ready to unviel their offerings shortly

Yes, we can't wait, we can't wait. World
will be a much better place once these chips are
Nomoredemo 12/4/2012 | 9:58:15 PM
re: Network Processors: Easy Come, Easy Go Bobbymax said

More start-ups spring up whwn the level of expertise need is rather low. A well known example is dotcom companies that were started mostly by H1B visa holders........ (and a bunch of letters put together after that)


Not much expertise needed to make statements like this..
Unless you do that on purpose hoping to get on the nerves of someone part of one of the group listed in your post (dotcom, chips companies , EDA, H1B holders , VC and also the "Nation"), id say you have a good chance of hitting one of them and get a nasty response (thats what you are looking for, right !) Sometimes i just wander if its not lightreading putting these posts just to keep the hit (in many ways) rate high on this board.
yesteryear 12/4/2012 | 9:58:12 PM
re: Network Processors: Easy Come, Easy Go Guess what guys? You know the real reason why people even began talking about "Network Processors"? I'll tell you why... because the microprocessor design engineers who love doing that stuff couldn't get a VC to look at a business plan involving a processor... they would point to Intel and Sun and state that every generation of processor doubled the size of the team and development cost of the previous generation, and VC's were far too short-sighted to think that a micro development team could make money doing anything other than a full head-on attack of the huge PC processor market... which would be a sure lose with anything less than a half billion or a billion dollar investment. They completely missed the idea that small processors could be profitable and used in many applications, including even networking applications.

How could anyone get their small micro development going in the heydey of the telecom bubble? You got it: Tell the VC's that you were a NETWORK Processor Unit development company. What a joke. These companies are still operating under the assumptions of 1999... how pathetic.

Hello.... the ship has left the harbor... if you enjoy doing processors, then just do processors... milking the "NPU" idea is over...

Maybe this is obvious... maybe not too many people think of it in these terms... but believe me... that is the way it went... I was there.

sigint 12/4/2012 | 9:58:03 PM
re: Network Processors: Easy Come, Easy Go BoddyMax,

Is there anything at all, which in your opinion requires "Much Expertise"? you seem hell bent on trivialising every one else's activities.
sigint 12/4/2012 | 9:58:03 PM
re: Network Processors: Easy Come, Easy Go well said, yesteryear. how depressingly true.
jamesbond 12/4/2012 | 9:57:02 PM
re: Network Processors: Easy Come, Easy Go I don't think there is enough business out there
to justify so many silicon startups. Recovery
ain't gonna happen in next two years.
Sign In