Rambus bestrides memory world like an arrogant colossus

IF THERE'S ONE thing I'm tired of, it's watching Rambus strut across the world stage, alternatively crowing about some supposedly-amazing technological advance that fails to materialise or sobbing about how its been so mis-treated the last seven years or so.

See Doctrine of "unclean hands" affects Rambus-FTC case and Rambus alleges DRAM conspirators forced Intel to choose DDR.

Now, according to the latest information available, we're supposed to believe that poor little Rambus was deliberately attacked by the Dramurai, its product sabotaged, and its future world dominance shattered by those evil, evil, companies.

Right. Now that we've heard the Rambus-side of the story in general, we'll examine its claims specifically:

#1: Manufacturers joined together to convince Intel Rambus DRAMs would be too difficult to build and too expensive to buy
Anyone ever thought this might be true? Back in 1996-1997, SDRAM was only beginning to have serious market share, and typically ran at 66 MHz. RDRAM is a very different technology than SDRAM, runs far faster, and uses a completely different bus design. It would (and did) mean considerable expense for the Dramurai to re-tool their fabrication plants and re-learn how to manufacture an entirely new standard of memory, especially after having just begun production of SDRAM.

#2: Manufacturers claimed [to Intel] that there were alternatives that offered cheaper and better performance
Gee. Now THAT'S surprising. Anyone remember when RDRAM cost eight times what SDRAM did, but delivered only a minimal speed increase? How about in 2000 when Intel themselves published benchmarks showing an SDRAM-equipped P3 system beating an RDRAM-built machine. Even if we grant Rambus' tenuous claim that the Dramurai deliberately charged huge amounts for RDRAM, the second point is incontestably true. There were alternatives to RDRAM that offered better performance.

And even the accusation that the RAM manufacturers deliberately colluded to raise prices on RDRAM is odd. The bulk of evidence from the time suggests otherwise, in fact. For starters, both Tom's Hardware and Anandtech reported at the time that RDRAM was not yielding well at high speeds. Zen-News reported in April of 1999 that RDRAM was not yielding well, as did Ace's Hardware and our own Mike Magee. Other news agencies carrying this information included EBOnline, TechWeb, and a variety of independent sources. Aces even has information posted directly from a source confirming that around August of 2000, RDRAM yields were only running about 12-17% per day. Here.

Obviously, there were yield problems with RDRAM. The PC600 and PC700 standards, in fact, were created to help manufacturers improve yields on difficult-to-manufacture PC800 parts, as discussed in a further Techweb article. Extremely low yields mean extremely high prices, and extremely high prices mean extremely low sales. This isn't rocket science, its Econ 101, but evidently the folk at Rambus Inc failed to graduate.

SDRAM manufacturers promoted DDR and SDRAM because they were royalty-free open-source standards
Hmmmm, another toughie. The plain fact is, most proprietary standards have failed on the PC, especially those with a high cost. Anyone out there remember Micro-Channel Architecture? How about Apple, the biggest proprietary computer manufacturer left….with 4% of the market? It's worth noting that Apple computers have become far more interoperable with their PC counterparts over the last twenty years, but it's a much-agreed upon fact that Apple's decision to retain a proprietary hardware control model cost them what could've been a dominant position in the PC market. IBM's original VGA standard and OS/2 were both proprietary. VGA obviously went mainstream, and OS/2….supposedly exists.

The simple fact is, the major successes in the computer world have almost all been open-standards. JEDEC, as a group, was created to continue to design DRAM along such lines, and it's ridiculous to think that an organization dedicated to creating and preserving a successful open standard would rush to embrace a proprietary model.

Manufacturers were wary of allowing a single customer (e.g. Intel) to control the market
Another (obvious) point. Monopolies, whether a monopoly of buyers or sellers, are never things businesses or customers walk into care-free. The example of history, in fact, is that monopolistic control slows innovation and tends to offer less to consumers than a competitive market. Only a fool ties himself to a single source of income or a single customer.

It's odd that Rambus is slinging mud about improper business practices considering this is the company that offered Intel a mammoth stake in itself if Intel would only ship a few hundred-thousand chipsets with its product. That didn't look suspicious or anything….

And frankly, if the Dramurai were truly price-fixing back in 2000 and 2001, I'd say they did a piss-poor job of it. The general goal of price-fixing is to enhance your income and skim profits off the top. In the 3rd quarter of 2001, Micron posted a net loss of almost two hundred sixty-six million dollars. Now, I can't speak for a company, but it seems that if I was losing money at the rate of a billion dollars a year, I certainly wouldn't work to destroy a RAM standard that was one of my few profitable ventures.

If you're looking for an explanation for why Intel dropped you, Rambus, look no farther than the mirror. The failure of RDRAM technology in 1999 was bad, but that's not what put the icing on the cake. It was your attempt to force the entire DRAM industry to dance to your tune, the threatening tactics you took against anyone who dared disagree with your interpretation of patent law, and the fury of a technical community who saw a corporation attempting to sneakily litigate into existence what their own product had never been able to accomplish.

You were economical with the verite all throughout 1999 about the performance and real-world capabilities of your product, and if there's one thing technical people hate, its being rolled by marketing. Rather than waiting for the introduction of the Pentium 4 and the successful emergence of RDRAM as an actually viable technology, you chose to go after the entire DRAM industry. Big mistake.

By the time the P4 was introduced, RDRAM prices had fallen to within 10-15% of SDRAM prices, and remained there for quite some time, even approaching price parity at some points. Yet throughout this entire time period, people rejected an RDRAM P4 solution. It wasn't price, in the end, Rambus.

It was the conscious refusal of consumers to pay money to a company that flat-out didn't deserve it. You made your bed. Lie in it. µ