Do we think RMBS patents are strong enough
electronicnews.com
DRAM's Fate in Committee
Group may lead Intel away from RDRAM
By Peter Brown
The real impact of the formation of a DRAM developer's group last week may not be apparent for many years, but one thing is certain right now: Intel Corp. has learned a lesson in cooperation when it comes to implementing new memory technology.
The Santa Clara, Calif.-based microprocessor giant is taking a group approach to developing next-generation DRAM. This is a far cry from the previous strategy it used in implementing a PC main memory, which involved dictating to DRAM vendors and to PC OEMs which memory would be used, namely direct Rambus DRAM (RDRAM). The DRAM developer's group -- consisting of Samsung Semiconductor, Hyundai Electronics, Infineon Technologies, Micron Technology, NEC Electronics and Intel -- will develop a cost-effective, high-performance DRAM technology for mainstream PCs targeted for delivery in the year 2003 and beyond.
An Intel spokesman said the company will have an influence only as one of the core members of the group and will not lead the group in any way. This leaves many of the DRAM vendors to choose their own course of action for the new technology.
"All existing memory technologies will be considered for what will be next," the spokesman said. "It could be RDRAM, it could be DDR (Double Data Rate), it could SDRAM or it could be something completely different, or a combination of the three."
Interestingly, Rambus Inc., Mountain View, Calif., has been left off the dance card and it may not ever make it onto the floor.
"You can see two extreme outcomes from this," said Steve Cullen, principal analyst for DRAM for Cahners In-Stat Group, Scottsdale, Ariz. "One is the group will come up with something that has no relation to RDRAM whatsoever, leaving Rambus to find new businesses outside of the PC market. The other is that it takes quite a bit from the RDRAM specification, and Rambus will continue to prosper."
One of the problems that Intel faced when it anointed RDRAM as the next-generation main memory was that it never gave DRAM companies a choice. In the past, memory makers decided the next type of technology as a group. This was the first time that memory outsiders, namely Intel and Rambus, were in charge of DRAM technology developments.
Because of this, DRAM vendors have been less than enthusiastic about supporting and developing RDRAM, which has contributed to the current stall in the proliferation of the technology.
Jeff Mailloux of Micron Technologies
"We are working toward a group that's somewhere between the all-open standard of JEDEC extreme and having Intel and Rambus dictate what everything is going to be and having to deal with the problems of that extreme," said Jeff Mailloux, marketing manager for Micron Technology Inc.'s DRAM group.
Mailloux said in Micron's case, this would have little impact on its roadmap, which consists of SDRAM, DDR SDRAM, RDRAM and the defunct SLDRAM. Micron will continue to develop multiple DRAM technologies, believing the market going forward will be segmented with three or four different memory types each occupying a roughly equal portion of the market.
"The DRAM guys need Intel because it is 80 percent of the microprocessor market, but by the same token Intel needs these DRAM guys because they represent 80 percent of the DRAM market," said In-Stat's Cullen. "(The formation of the developer's group) is a positive move for everyone because it will now silence those DRAM guys who have complained because they had no role in the Rambus initiative and it will allow Intel to get what it wants in terms of DRAM technology."
Does this mean the technologies that the group adopts may replace RDRAM or the other DRAMs that are now on the market? Some say maybe.
"We have to be more cost-effective in the future, while being high performance," said Jan Du Preez, vice president at Infineon Technologies. "There may very well be some technologies that exist after what comes from the group, but I don't know how long they will survive. You can guarantee that if 80 percent of the DRAM market determines this is the direction we are going, then I think this is what it's going to be."
Du Preez said the market should see what direction the group may start making in about six months.
"No one can make a prediction if any of today's technologies will be used in the new DRAM technology," said Du Preez.
However, not everyone thinks Rambus, or for that matter any of the current DRAM technologies, may be running out of steam. Since the committee is looking at a number of alternative technologies, it could adopt one involving a Rambus patent, according to Nathan Brookwood, memory analyst at InSight 64.
"If that's the case, then Rambus will definitely play a part in the future. But it's so far off that we may not know what will happen for a number of years," Brookwood said.
Intel agrees that it is far too early to tell. However, the company is firm in its stance that it will support whichever DRAM technology the group decides on. At least that's its stance for now.
"Whatever the group comes up with is what the group will embrace, and we are included in that group," the Intel spokesman said.
If the formation of the DRAM developer's group and Intel's wavering support weren't enough bad news for Rambus, start-up Transmeta Corp. last week said it had no plans to utilize RDRAM in its Crusoe processor and maybe not for the life of the company. The company said it would use SDRAM and DDR as its main memory.
Things aren't looking up in the test arena either. According to a Goldman Sachs report, Teradyne testers continue to be bought for SDRAM and DDR memories. Teradyne's management expects that 80 percent of its equipment this year will be for SDRAM and DDR memories, with much less activity in RDRAM due to what Goldman Sachs analyst Gunnar T. Miller described as a debacle for Rambus technology. |
