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 : Rambus (RMBS) - Eagle or Penguin -- Ignore unavailable to you. Want to Upgrade?


To: richard surckla who wrote (40520)4/20/2000 8:03:00 PM
From: Barry Grossman  Respond to of 93625
 
Vitesse to acquire network-processor company Sitera for $750 million
Semiconductor Business News
(04/20/00, 09:15:05 AM EDT)

semibiznews.com

CAMARILLO, Calif. -- Vitesse Semiconductor Corp. here has agreed to acquire network processor startup Sitera Inc. for $750 million in common stock.

Sitera, a three-and-a-half-year-old company based in San Jose, released the first Rambus-enabled network processor earlier this month (see April 3 story). It is aimed at Internet switch/routers for traffic management, traffic classification, policy enforcement and guaranteed delivery.

"The powerful capabilities of Sitera's network processors extend Vitesse's capabilities from OC-192 down to OC-3 and DS3 and provides significant flexibility to implement emerging services for Intelligent edge switches and routers, ISP/ASP service platforms and core optical routers," said Lou Tomasetta, president and CEO of Vitesse. "Sitera has assembled a very talented team, bringing additional networking and development expertise with innovative products to Vitesse, insuring our dominance in the emerging network processor markets," he added.

The transaction is expected to be completed in the quarter ending June 30. Vitesse expects the transaction to be accounted for as a pooling, but said that pooling is not a condition to closing.

For Vitesse, this is the second acquisition in a month. Earlier it purchased Orologic Inc., a LAN/WAN-chip supplier for switch-fabric applications, for $450 million (see March 27 story)



To: richard surckla who wrote (40520)4/20/2000 8:06:00 PM
From: Barry Grossman  Read Replies (1) | Respond to of 93625
 
Intel loosens timing spec to spur Rambus usage
By Anthony Cataldo
EE Times
(04/20/00, 02:57:59 PM EDT)

semibiznews.com

TAIPEI, Taiwan ( ChipWire) -- Intel Corp. said it has relaxed a timing specification for Rambus-in-line memory modules (RIMMs) as part of an ongoing effort to reduce the modules' cost and to qualify more companies to make Rambus DRAMs and RIMMs. The move is one of several cost-cutting moves that Intel, Rambus and DRAM vendors are using to drive down RIMM costs, which today are as much as triple the price of SDRAM modules.

But the still-high costs of Rambus memories won't impact the launch of Willamette, Intel's next-generation 32-bit Pentium processor, slated for rollout this fall. But Rambus prices could determine how rapidly Intel moves Willamette -- which will support Rambus DRAMs and RIMMs -- into mainstream markets, one Intel executive said. Technical problems with Rambus DRAMs and a lack of Rambus parts forced Intel to delay the launch of systems last year tied to Camino, Intel's first core logic to support Rambus.

To counter the high prices of Rambus parts, Intel said it increased TDP timing specs -- a measure of the maximum time a signal can travel from a memory controller through a system's memory modules -- from 1,500 picoseconds to 1,560 ps. The extra 60 ps should help increase the yield of RIMMs as they are tested, and will allow several module makers vexed by the tight timing parameters to get qualified "overnight," said Pete Mueller, engineering manager for the platform components group at Intel.

"Basically, we realized we had some picoseconds left over," Mueller said, following a presentation at the Intel Developer Forum here.

The timing spec has proved a formidable barrier for RIMM makers, affecting the design of the RIMM's pc-board substrate itself through to the final testing of the board populated with RDRAMs.

While no work is being done to allow for systems with three RIMM slots, the revised timing spec will be able to accommodate such a configuration if Intel decides to add the third slot in the future. "We just didn't want to give that option away," Mueller said.

In Taiwan, where most of the computer industry's PC-boards originate, many board manufacturers are ill informed about the tight impedance characteristics of the RIMMs, sources said. Often they test the "coupons" attached to the periphery of a board in the production process, rather than testing the actual traces on the PC-board. That's one reason why RIMM manufacturers are asking that every pc-board they receive be tested.

"Right now it's 100% test because the spec is tight," said Mike Resso, product manager with Agilent Technologies Inc., a large supplier of test equipment.

Module makers aren't always getting the message either. In one case, an engineer realized he had been using the wrong test parameters after he finished testing a large quantity of finished RIMMs, Resso said.

PC-board traces for RIMMs must meet a 28-ohm impedance specification with a 10% margin of error, versus the 50-ohm characteristic impedance for today's more common dual-in-line memory modules. If test engineers aren't careful, the cabling, probes and sockets of the test equipment itself can throw off the results, Resso said. In many cases, test equipment must be recalibrated to take these errors into account. "Nobody had to worry about this at 50-ohms," Resso said.

Motherboard testing can also be a headache. Only the newest testers have the bandwidth needed to assess the flight time of a signal with the added complexity of the 28-ohm traces and differential signaling. Otherwise test engineers must move the probes manually from point to point, a process that can take from 30 minutes to an hour per point, said Roger Lo, district sales manager for the local branch of Tektronix Inc.

"The basic problem is that the signal is so fast that the test procedure has to be hard," Lo said.

While test equipment makers are trying to get the word out about proper testing procedures, Intel, Rambus and some memory suppliers are considering other ways to reduce the cost of the PC-board itself. Currently, RIMM pc-boards must use either six or eight layers and cost about $8 apiece, but observers here said they want to introduce four-layer boards.

Other cost-cutting measures are in the works. At least four companies have announced new testers for high-speed DRAMs that can test 32 to 64 chips in parallel, which would pare down some costs to the manufacturers, said Chris Slocum, business development manager for Rambus Inc. in Mountain View, Calif. Others said that Intel, Rambus and chip vendors are also discussing ways to reduce the size of the RDRAM memory array, which Toshiba Corp. indicated it would do for its next 288-Mbit Rambus part.

As to countering the higher cost of the chip-scale packages used for RDRAMs, Slocum said it will depend largely on how quickly the industry ramps up CSPs. He noted that many suppliers are switching from 35- to 48-mm-wide tape to boost production.

For all that is being done to lower test, packaging and manufacturing costs, Intel and Rambus said the most significant cost reduction will come when more companies produces RDRAMs in volume and price competition ensues. To date, Intel has qualified RDRAMs made by Hyundai, Infineon, NEC, Samsung and Toshiba. As a result, these chip makers are still able to sell the devices with a heavy premium that, Intel contends, does not reflect the cost of manufacturing the devices. "We know the cost has been very inflated," Intel's Mueller said.

Mueller said he's pushing DRAM makers to get the RDRAM premium to 20% about SDRAM prices by the end of the year. That could be possible if more DRAM makers come into the market this year, and if SDRAM prices continue to rise throughout the year, which would help close the price gap between the DRAM types, he said.

But one of the current RDRAM vendors, Samsung, denied that prices are artificially inflated, and said there is no great difference between the RDRAM selling price and its cost of production. "I've heard that 20% figure from Intel about 10 times," said Jay Hoon Chung, manager of DRAM marketing for Samsung Electronics Co. Ltd. in Seoul, currently the largest supplier of RDRAMs. "But 20% is not probable by our point of view. We expect the price gap will be 1.5x by the fourth quarter."

For its part, Rambus is looking to spur more competition by getting more companies to sign up as licensees of its technology. Winbond is the only DRAM company in Taiwan to license the technology, but Rambus' Slocum said he will pitch Nanya and Powerchip in an effort to sign them as licensees when he visits Taiwan.

In addition, Slocum said that Micron Technology, a big supporter of double-data rate (DDR) DRAM and one of the few major suppliers not to qualify an RDRAM, is expected to ramp its RDRAM production by late second quarter or early third quarter this year.

Whatever the price of RDRAM over the next few quarters, it will have no bearing on the launch of Willamette, which will feature a 3.2-Gbyte/second interface between the processor, north bridge and DRAM. Intel's prime concern is not how RDRAM will affect the price of Willamette-based systems later this year, but whether high RDRAM prices will prevent Intel from quickly driving down the cost of the Willamette platform.

"We can launch it today, even with the high [RDRAM] prices," said Pat Gelsinger, vice president and general manager of Intel's desktop products group. "Whether you can get it down to the sub-$1,500 price points is where it gets difficult."




To: richard surckla who wrote (40520)4/20/2000 8:14:00 PM
From: Don Green  Read Replies (1) | Respond to of 93625
 
Richard>... if that's OK with you?

Fine with me, I have always felt that not posting everything, "the good, the bad and Bilow" was doing an injustice to this thread.

Don



To: richard surckla who wrote (40520)4/20/2000 8:15:00 PM
From: Barry Grossman  Read Replies (1) | Respond to of 93625
 
ITC to hear Rambus' patent-infringement suit against Hitachi
By Jack Robertson
Semiconductor Business News
(04/20/00, 12:06:15 PM EDT)

semibiznews.com

WASHINGTON -- The International Trade Commission, as expected, agreed Wednesday to investigate the complaint of Rambus Inc. that Hitachi Ltd. and Sega Enterprises Ltd. have violated the U.S. firm's patents on synchronous memory and high performance bus interfaces (see March 23 story).

The ITC has appointed Debra Morriss, an administrative law judge, to handle the case. She will make a preliminary decision on whether any Rambus patents have been infringed. The full commission will then make a final determination. The process usually takes about a year.

Mountain View, Calif.-based Rambus is seeking to have Hitachi DRAMs and microprocessors barred from import into the United States due to the alleged patent violations. Sega was named in the ITC case because the electronic game maker uses Hitachi microprocessors in its latest Dreamcast players.

Rambus has also filed a patent infringement suit against Hitachi in the Wilmington, Del., district federal court. So far, only Hitachi has been sued, although industry observers said virtually all memory and microprocessor chip makers use the synchronous timing and interface technology that Rambus claims it has patented.

Hitachi last month filed a counteraction in the federal court charging that the synchronous technology was invented and widely used long before the Rambus patents were filed. The Japanese firm also claimed that Rambus got its synchronous technology by participating in the industry Joint Electron Device Engineering Council (Jedec) standards deliberations in the mid-1990s. Hitachi said Rambus then left the Jedec discussions and used the data to form the basis of its patent applications.