Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Technology Stocks : AMD/INTC/RMBS et ALL -- Ignore unavailable to you. Want to Upgrade?

To: Joan Osland Graffius who wrote (163)	10/7/1999 2:12:00 PM
From: grok	Respond to of 271

RE: <Do you have any thoughts on Larrys question? Message 11475054 Joan> I don't understand how anyone who doesn't actually live in a cave could not realize that there is a perfectly good evolution from PC100 to PC133 to DDR200 to DDR266 to DDR-II which is going to happen inspite of the best efforts of Intel and Rambus and which will maintain the lowest cost per bit at each step while evolving to higher bandwidth than Rambus can supply.

To: Joan Osland Graffius who wrote (163)	10/7/1999 3:33:00 PM
From: Bilow	Respond to of 271

Hi Joan Graffius; Re Larry's question: "Joan One more thing, some one related how every big jump in Intel's processors & Memory has resulted in bugs and delays. If any one knows a system other then Rambus for advaced memory, please let us know. Larry Dudash" The reason that Larry Dudash doesn't know of any other system suitable for advanced memory is because he doesn't know the business. The fact is that if you read the Intel and Rambus literature only, you would think that there is exactly one possiblity for future memory systems, and that this is Rambus. This fact is what has driven a lot of investors into buying Rambus. Intel and Rambus marketing drivel has little effect on the memory industry. What does have an effect is Intel announcing that they will produce ICs that support Rambus and no other memory system. The number of people who post on SI, and who have actually designed a memory system for a commercially successful product is exceedingly small. Those who have, are not saying that Rambus is a winner. I understand that verifying the technical details is way beyond the ability of 99%+ of SI readers, but you can read what the industry thinks about Rambus by reading the articles written in EE-Times, for instance. Here is a link that will provide all the EE-Times articles that mentioned Rambus in 1999. (The total number of hits is 115, at the time I write this.) Read through these articles, and you will have the answer to your question. I could give you links to a few of the more cogent ones, but I know that it is firmly written into the hearts of those who buy RMBS that they should not trust biased reports. So here they are, all of them. Read them and become knowledgable about the various alternatives. Of course, even after reading them, (or just perusing the headlines), you will still not be an electronics engineer with years of experience designing commercial memory systems. techweb.com The problem that Rambus longs have with believing what neutral industry observers have to say about Rambus (and as soon as someone says something other than perfection about Rambus they are no longer considered to be neutral,) is that they can always say, "biased writer." If you want to understand the legal requirements that face Intel, and decide for yourself as to whether they are a biased observer or not, go and read the SEC filings that describe the conditions under which Rambus sold Intel all those stock warrants. Essentially, those conditions require Intel to make best efforts to pump Rambus. None of this really matters. The fact is that Rambus is largely a stock play, not a technology company. They were asked by Intel to design a modification to the old Rambus memory architecture (which had one design win), so as to make it useful for general PC memory. Intel then pushed them with "not good enough" type feedback until Rambus pushed the limits of technology beyond the point where it was manufacturable. That it was pushed this far was obvious to a lot of engineers, but not to all of them. Intel and Rambus then did the traditional thing that management does when faced with divided opinions among engineers - they believed the story that they wanted to believe, and went ahead. The specification was for an 8x improvement over existing memory data rates. In order to reach this high of a number, they had to make extremely tight specifications for the memory, the RIMM, the motherboard, and the controller. The first part to run into trouble meeting those specifications was the memory. The memory makers were unable to test at speed, and had yield problems. This delayed the project until this past summer. The RIMMs also had trouble meeting spec, as is referenced in the PDF file from Hewlett Packard, which shows an out of spec RIMM module as a selling point for the HWP oscilloscope. The memory boards must have also had troubles, and now it is difficult to know whether it is still the memory boards or the controller chip (i.e. Camino) that is currently halting production of RDRAM chips. Intel has shut up completely about the problem. I believe that if there were good news we would be hearing about it. The reason that they cannot publicize the bad news is clear from the restrictions on the RMBS stock warrants that Intel owns. Those warrants are struck at $10 per share, as long as RMBS is above that price, there is no reason to expect Intel to be say anything negative about Rambus. What the general public fails to realize, I have no doubt, is that there is a big difference between what is possible, and what is profitable (i.e. manufacturable at a reasonable cost.) If the box makers had to throw away the same percentage of their boards as NASA loses spacecraft, the scrap costs would eliminate all the profits from almost all of them. Take a look at the profit margins on even highly profitable companies like DELL. Not very much, are they? To avoid scrap costs, engineers design computers to have lots of "margin". This is the same thing as the fact that bridge designers use more concrete than their calculations would suggest. In order to avoid failures, you have to design in a lot of excess strength. How much extra strength you need is a subtle point, one that it is pointless to discuss with people outside the industry. Suffice it to say that RDRAM has demonstrated that it has insufficient margin, and that is why it is not being manufactured. The thing to note is that the RDRAM we are talking about here is not the RDRAM that was used in game machine that was Rambus's great previous success. The two technologies came from the same company, but they are not the same. The newer technology (also refered to as "direct Rambus") is much more difficult than the previous. This difficulty is what Intel demanded in order to make the technology reasonable to use in low end PCs. -- Carl

To: Joan Osland Graffius who wrote (163)	10/7/1999 4:41:00 PM
From: Bilow	Read Replies (2) \| Respond to of 271

Hi Joan Graffius; Are you suggesting that Intel shorted Rambus (or purchased derivatives that would increase in value as RMBS decreased in price)? Using inside information about the Camino release? Man, what a conspiracy theory! (Not that I'm saying it couldn't happen.) Clearly, it was the big box makers that forced Intel to hang the launch, as it was they who found bugs. This would be consistent with the way that box makers test new designs, in my experience. They are a lot more vicious in their tests than Intel would be. But those box makers knew there were troubles at the time when RMBS was trading at $117, I would imagine that a few of their engineers may have got worried and sold a few RMBS shares... My theory is that RMBS trades largely on things other than fundamentals anyway. Did you start reading through the EE-Times articles? Try to ignore the ones that only talk about one product. Those are usually company written drivel, which are given out free to the press to republish. The articles to look for are the ones with a lot of editorial comment, lots of comparisons between different technologies, and without an obvious conclusion. Remember that engineering is always a search between different techniques, and that all "solutions" have drawbacks as well. It really isn't that much to look through. You can read all 115 headlines with just a dozen mouse clicks or so. -- Carl Edit: Intel's RMBS warrants aren't active yet, nor are the other warrants that RMBS gave to other memory companies: techweb.com (This post brought over from the ASL thread, to which it was mistakenly posted.)