What happened from the FOOL:
The noise (FUD) is reaching a crescendo. I think this is a further indication that the end is near.
My scenario for all if the noise leading up to the lawsuits this last week goes something like this:
1. In 1991 Rambus wrote an article, published in the trade press, about the memory technology of the future. It didn't reveal the details of its patents applied for, but it laid out the basic requirements for future high speed memory designs in sufficient detail to get the interest of Intel and the DRAMURAI who are looking for a match to the demands of processors on the horizon.
2. Intel began looking very seriously at Rambus inventions under NDA. In 1992 Intel, signed on to the Rambus project. Intel received full disclosure of the Rambus patents applied for and pending.
4. Rambus had already developed and investigated the basic architectures (on paper) for what was later called SDRAM and DDR, but their patent disclosures didn't use those names, rather they patented the features the principles that made SDR and DDR work, as well as the bus features of RDRAM.
They had already compared theoretical performance features of all three types of memory and determined to focus on a bus based architecture which became RDRAM 5. During the next several months most JEDEC members signed RDRAM licenses. Rambus disclosed the details of the RDRAM architecture to
the DRAMURAI.
6. In 1992 the European patents applied for 18 months previously become (by law) public and are available for anyone who does a proper search. The DRAMURAI, being interested in what Rambus did, got details of the RDRAM architecture with their licensing agreements and by searching the European patent offices, knowing full well that Rambus is an IP company and that they will have protected their ideas as soon as they were developed.
7. In 1992 Jedec, began planning a new synchronous standard. They invited Rambus to participate. Their intention is to get what ever they can from RAMBUS and incorporate it into a new non-proprietary standard. They already have a good start with the license data and the European data, but they hope to get more.
8. JEDEC has difficulty completing the standardization process, so in 1994 they adopt the SDRAM spec developed by Intel. They do not know and are not informed, and probably don't care, that Intel has developed the spec based on RAMBUS IP revealed as a result of Intel's special
relationship with RAMBUS.
9. Rambus looked at the spec, knew that it had come from Intel and was based on their IP and that it infringed.
They informed JEDEC and its members, in writing, that RAMBUS is an IP company, that it has IP that the new standard appears to infringe, and that Rambus intends to seek royalties which do not meet JEDEC guidelines. They withdraw from JEDEC. They also informed Jedec at that
time that they had applied for additional patents to protect their IP.
10. JEDEC completed the SDRAM standard based on the Intel spec, and the DRAMURAI began development and, in 1996, production of SDRAM.
11. Development of the DDR standard had begun at the same time but it didn't progress as well (because Intel hadn't done a DDR spec for them).
11. In 1996, the DRAMURAI, lead by Micron, began a campaign to stall RDRAM acceptance in the market. Their main purpose was to buy time to develop and introduce DDR as a competitive alternative.
DDR itself, was based on data from their RDRAM license disclosures as well as the European patents.
12. The stalling campaign was effective for several reasons
a. RDRAM is more complex than SDRAM so it requires more silicon. This increases cost, particularly in initial low volumes.
b. RDRAM requires a new package design. This increases cost, particularly at initial low volumes.
c. RDRAM is merely a piece of the system and Intel chip set is inadequate to utilize the full throughput of the RDRAM. The 820 has problems and is not well matched to RAMBUS performance. The 840 is better but goes only into workstations.
d. Most of the licensees didn't build RDRAM.
e. Of the licensees, only NEC, Samsung and Toshiba spent the money to ramp up production. Each had (has) a specific target customer and each gained nearly 100% market share in their respective target markets, NEC for the Nintendo; Toshiba for the Playstation, and Samsung in the PC workstation market.
f. Volumes picked up but prices stayed high because there was no competition.
14.The stall worked!
Intel became unhappy with Micron and began to dump MU stock.
The P4 release was put at risk as was the Intel Rambus relationship.
Intels may fail to meet its RDRAM contract obligations and to lose millions in warrants.
Intel was forced to drop RDRAM exclusivity because the prices failed to come down.
15. Rambus sued Hitachi and Toshiba for non-performance and patent infringement. Both settle. Now SDRAM and DDR are on the table. The DRAMURAI will have to pay even if they don't produce RDRAM. The plot is unraveling.
16. As the P4 launch neared, the RDRAM licensees decided they couldn't pass up the potential volume opportunity. NEC, Toshiba, Infineon, and yes even Micron get qualified and announced plans to ramp RDRAM.
17. In August 2000 partial benchmarks on prototype P4s begin to leak out showing potential blow out performance in the combination of RDRAM and P4. The ramp for RDRAM begins in earnest. RDRAM OEM prices begin to drop rapidly. Even aftermarket RIMMS drop significantly in the face of rapidly growing demand for RDRAM, driven by the success of the PS 2 and the PC workstation market (Intel 840)and the P4 look to put Rambus over the top unless P4 can be stopped or blunted. Timna is then possible which will undermine the rest.
18. Micron and Hyundai sue RAmbus in the hopes of eliminating the royalty requirement, and more importantly, regaining power over the DRAM market.
19. The FUD campaign grows to a thunderous level as AMD, Micron and other Intel/Rambus enemies realize that time is running out.
20. This final campaign is tied to a planned rollout of DDR based Athalon PCs planned for Q4 2000 or Q2 2001. They just need another quarter or two of delay on the P4 and RDRAM in order to field the promised competitive solution.
21. My predictions:
a. P4 performance will be even better than the hints we have already gotten. The best clue is that the P4 price at launch is far lower than any previous Intel high end processor <$800 Q1.
Two possible explanations:
i. The P4 is dog slow and they can't sell it at the normal price. (Why introduce a dog? It would destroy Intel to do so.)
ii. The P4 is fast and manufacturable and they want a very fast ramp so as to destroy the competition and end AMD's recent string of successes.
b. No memory maker will allow themselves to be locked out of the P4 market. They will build them and the RDRAM prices will drop like a stone.
c. The Timna will ship in Q1 2001 and low cost PCs will be built with RDRAM(possibly in 32 MB configurations) The price/performance will be phenomenal.
d. The DDR lawsuits will drag on until the courts dispose of them, but DDR will be dead in the PC market as the public sees the price/performance possible with a PC designed to use RDRAM.
JMHO |
