Scumbria,
RE:At the Embedded Systems Conference, the i815e was all over the Intel booth. No mention of i820e."
----
This could explain it, even Intel employees are distancing themselves from the R word.
electronicnews.com
"NEWS: October 23, 2000
Culture Clash Erupts Inside Intel
Management and engineers do battle over Rambus DRAM
By Steven Fyffe and Paul Kallender
Intel Corp. is pointing the finger of blame at Rambus Inc. over their disintegrating union while struggling to keep a lid on internal bickering among its own family of engineers.
After years of putting on a unified front in public, Intel has admitted that its alliance with Rambus has not been a happy marriage. "We made a big bet on Rambus and it did not work out," said Craig Barrett, Intel president and chief executive officer, to The Financial Times at the company's recent eXCHANGE e-Business Summit in San Francisco. "In retrospect, it was a mistake to be dependent on a third party for a technology that gates your performance."
The alliance has soured Intel's relationships with many of the major memory makers. And when Intel recanted its commitment to only use direct Rambus DRAM (RDRAM) in its Pentium 4 platform, Rambus stepped up its campaign to collect royalties on competing SDRAM and double data rate (DDR) memory technology. Three of the top five memory makers say Rambus has no right to claim royalties on these designs and are now heading to courtrooms across the world to prove it.
Barrett openly denounced Rambus' legal tactics. "We hoped we were partners with a company that would concentrate on technology innovation rather than seeking to collect a toll from other companies," he said, according to the Financial Times article.
If Rambus loses Intel's patronage, its chances of mainstream success are over, analysts said. "Long term, it's pretty much the kiss of death (for Rambus)," said Bill McLean, president of IC Insights.
But just like the breakup of a marriage, it's always the children who suffer the most. The Rambus relationship has taken its toll on Intel's engineers, leaving them divided, disillusioned and distrustful of the hierarchy, according to a former Intel employee who spoke to Electronic News on condition of anonymity.
"There's been an internal backlash, and anybody that can distance themselves from Rambus has done so," the source said. "Now nobody believes the Intel Architecture Group. Intel had been good at generating rules that designers could use to develop boards. They basically broke that trust by forcing Rambus down the throats of the engineers."
Intel's senior management ignored early warnings from its own engineering staff that Rambus technology would be hard to implement, according to the source. "Requests for information went up the chain and there never would be any response. Two years ago we told them that technically it was a poor proposition at best. But they relied on Rambus for all of the engineering expertise in this area," the former employee said.
"We told them there wasn't enough design margin; they were going to be hurt by noise and impedance problems. (One staff member) determined there was negative timing margin. This was repeated to upper management. The reported response from (one executive) was, 'It's not the first time we've shipped devices with negative margin.' "
The lack of reality checking inside Intel became extremely serious, according to the source. "The Intel Architecture Group is an arrogant bunch that all ought to be fired. This is a group that many times has no contact with reality and doesn't know how to manufacture real things."
Engineers who didn't toe the party line were punished with bad reviews, the former employee said. "Inside Intel, the term was 'disagree and commit.' " Intel employees were not allowed to speak badly of Rambus. "Those internal to Intel who failed to 'disagree and commit' to Rambus were summarily given poor reviews because they were not considered team players, as they disagreed with the commitment of Intel's entire chipset resources and architecture to Rambus," he said. "Some extremely talented individuals, and friends of mine, have had their careers shortened or diverted because they did not become swept toward the cliffs with the rest of the lemmings."
The recent failure of Intel's memory translator hub (MTH) project, which was meant to allow Intel's 820 chipset for RDRAM to also work with SDRAM in Pentium III systems, was due to Rambus' highly sensitive technology, the source said.
"The issues were not defects within the MTH. The issues were with the Rambus channel itself and the use of large packages at channel speeds. Technically, the problem has been with microwave-like resonance effects in the component packages, connectors and in the structures formed by these when placed on printed circuit boards."
Rambus' strict design rules left engineers with little elbowroom to be creative, another industry insider said. "Engineers as a whole don't like being dictated to," he said. "With Rambus' design there's no flexibility."
An Intel spokeswoman said it was policy not to comment on statements from ex-employees, but she did confirm that the terms "disagree and commit" and "constructive confrontation" are integral parts of Intel's work culture.
Ignoring dissenting voices was a big mistake for Intel in this case, said Sherry Garber, vice president of Semico Research Corp."Intel just didn't listen to the people who knew. You have to listen to the technical experts that you employ. That's amazing to me that they can shut those people up."
Some observers said Barrett's latest comments showed Intel was eating humble pie and owning up to its mistakes while trying to rebuild battered relationships with memory makers. "It's an olive branch," said Walt Lahti, vice president of market research at Integrated Circuit Engineering Corp."They have to make up to the DRAM industry to get those guys fired up."
Others said that Barrett's statements demonstrated that Intel was wising up to the new market dynamics of the DRAM industry. Intel has learned that the industry has changed since the days when 15 or so DRAM suppliers were clamoring for its business and willing to accept its decrees without question. The DRAM shakeout of 1995 weeded out the weaklings and left only about four or five major players standing, and they have shown they won't be pushed around by anyone, not even Intel.
"I think Intel is ending a long period of denial and is finally coming to its senses with its memory technology strategy," said Nathan Brookwood, principal analyst at Insight 64. "Intel clearly has been moving in this direction for six months."
Intel may have foreshadowed a possible future breakup when it left Rambus off the invitation list to join its Advanced DRAM Technology Alliance (ADT) that is working on next-generation memory, McLean said.
But Intel says it can't abandon Rambus for at least two more years, because of its contract with the Mountain View, Calif.-based company. The sections of the contract that are public suggest that Intel can walk away from the agreement anytime scot-free, said Bert McComas, founder and principal analyst of InQuest Market Research. The real reason it won't leave is buried in the parts of the contract that have been blacked out, he said.
"The (public) contract says all Intel has to do is write Rambus a letter and it is terminated with no penalty, but there is all this stuff stricken from the record. Maybe they've got to pay hefty royalties on their processors. It has to be something of that magnitude or Intel would have bailed out. There is something serious on the table. They just have to hold their breath until 2003."
The former Intel employee said that theory would fit with his own speculation. "I personally thought they had been caught with their pants down and had violated a Rambus patent. Intel has an awful lot invested in DDR signaling technology. If they truly believed that the Rambus intellectual property would hold, they are in deep water."
Many of Intel's next-generation server and PC processors, including the Foster, McKinley and a revamped Merced, reportedly use DDR technology. And Rambus has already sued Hitachi Ltd. and Sega over a controller that interfaces with SDRAM.
--------- |
