Electronic News on the Rambus Trail
Jonathan Cassell, Editor-in-Chief
electronicnews.com sp
Being a pioneer can be a good thing. It can mean you are first in line for the land grab. It can even mean that you have a good chance of getting a town or a mountain named after you, preserving your legacy for history. However, being a pioneer can also mean that you are the one that gets hit with the most arrows.
Over the past year-and-a-half, Electronic News has been a pioneer in covering the most important issue facing the memory market: Intel Corp.'s support of direct Rambus DRAM (RDRAM). In this area we've done trailblazing coverage that has brought important issues to the attention of our readers and has garnered recognition and an award from our peers. However, we've also taken more than our fair share of arrows in the back.
With Craig Barrett, Intel's president and chief executive officer, admitting last week that his company's support of Rambus may have been a mistake (see story), we feel vindicated in our skeptical, rigorous coverage of RDRAM. Let's look at some of the highlights of our Rambus coverage:
It was 16 months ago that Electronic News became the first publication to take a detailed look at the technical challenges and the cost and performance issues associated with RDRAM in our series Prying the Lid Off of Rambus Designs (See issues dated May 3, 10 and 17, 1999.). The series examined the cost of RDRAMs, associated system-level costs, heat-dissipation concerns and packaging issues.
Some of RDRAM's technical problems would later play a role in Intel's repeated delays in its introduction of its 820 Rambus-enabled chipset, its motherboard recall (Intel to Replace One Million Bad Boards, May 15, 2000) and its continued difficulties with its MTH chip (Intel Can't Whack MTH Bugs: Timna Delayed, June 12, 2000; Intel's Memory Translator Hub Implodes with Timna, Oct. 9, 2000). In our Page 1 story this week, there's evidence that Rambus' MTH difficulties may be related to packaging problems.
The series presented some of the first intial evidence that RDRAM might not represent the pinnacle of memory performance, as RDRAM supporters had asserted early on. More than a year later this would be borne out by Intel's own benchmark figures (Intel's Tests Give SDRAM Edge Over Rambus, July 17, 2000).
Electronic News ' coverage of RDRAM heated up considerably with our exclusive story IBM Says No to Rambus (May 31, 1999). That story was the first to report that IBM had dropped RDRAM from its future PC product plans due to performance limitations of the technology. The news was significant, as IBM was the first PC OEM to pull support for RDRAM. It was also significant because IBM performs internally a much greater percentage of its product engineering than other x86 PC OEMs do, and thus was the first computer maker to notice the performance deficiencies of RDRAM.
This story generated quite a bit of controversy in the marketplace and Electronic News took a lot of heat from Rambus investors, many of whom sent abusive e-mails to our staff accusing us of inaccurate or biased reporting. The attacks intensified when IBM appeared to issue a denial of our article, but IBM's statement was actually the company's response to mangled versions of our story brought to its attention by our competitors.
Our story was supported by IBM's decision to roll out double data rate DRAM (DDR DRAM) modules for servers, a development that hammered Rambus' stock (The Antenna, March 27, 2000). IBM since has introduced some workstation and server models and commercial systems using RDRAM, but almost a year-and-a-half after our story, the company still has not introduced a real PC that uses it.
I still haven't received an apology from the investors and competitors who attacked our credibility, and I'm not holding my breath.
The first strong indication that Intel's support for RDRAM was wavering came a few months after our IBM story, when Intel said it would support an alternative to RDRAM for next-generation memory (Microprocessor Giant will Develop PC-133 SDRAM Chipset, Sept. 6, 1999).
In the same issue, I wrote that our reporting on Rambus had been validated by Intel's announcement and I predicted that PC-133 would be the next volume leader in the DRAM market, to be followed by DDR (Told You So, Sept. 6, 1999). The following week, I received yet another piece of hate mail from one of the same investors who had attacked us for the supposed inaccuracy and bias of our previous coverage, this time criticizing me for gloating. With some people, you just can't win.
Semico Research Corp. now reports that PC-133 will be the predominant memory type in 2000 and probably through 2004, with RDRAM representing less than 1 percent of the market this year and peaking at 1.3 percent in 2001.
Our Rambus coverage hit a high point two months ago with our exclusive story showing that Intel had no plans to use Rambus in mainstream PCs until 2002 (Intel Scales Back Rambus Plans, Aug. 14, 2000). Once again, we came under abusive attack for our coverage. Inexplicably, most of the criticism followed the same theme: that our story was "old." This, despite the fact that our story was based on a confidential Intel roadmap that had never been seen publicly and that our article mainly concerned unannounced Intel products whose names had never before appeared in the press.
Apparently, the sources of this piece of misinformation were Rambus and Intel themselves, even though they must have known differently.
This week we see that Barrett is acknowledging that his company's support for RDRAM was a "mistake," amid reports of internal strife at Intel over Rambus. We are pleased to see that Barrett has the guts and integrity to admit when he has made a mistake and we congratulate him on his courage.
Where the path leads from here is uncertain. The Intel/Rambus relationship may not have much of a future.
Despite our many arrow wounds, we are proud to be pioneers, and proud of the scars. Our refusal to turn back in the face of criticism and harassment was a result of keeping clearly in mind our mission: to serve our readers. Thank you for sticking with us.
Talk to Us
Electronic News encourages its readers to send in their comments. Please e-mail comments to jcassell@cahners.com, or mail them to Jonathan Cassell at 1101 S. Winchester Blvd., Building N., San Jose, CA 95128 |
