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 : Rambus (RMBS) - Eagle or Penguin -- Ignore unavailable to you. Want to Upgrade?

To: Bilow who wrote (41425)	5/5/2000 10:07:00 PM
From: Bilow	Respond to of 93625

Hi all; Re Rambus and simplicity... One of the bits of fantasy that Rambus seems to spontaneously generate is that Rambus has "solved" the memory interface problem, and therefore there is no more work for memory designers. This is one of those simplistic one-liners believed by people who are a long way from actually having to do the work. Rambus is a lot more work than DDR, not the other way around. Rambus is a lot more complicated than DDR, and is more restrictive as to which chips it can be hooked up to. The one engineering advantage that Rambus has is the reduction in pins, but with new packaging technology, this advantage is not so important as it once was. (Consider the AMD Lightening Data Transport, which uses unidirectional wires with only a single source and sink on each. This is an example of the modern profligacy with pin count. There was no way that the low cost and high reliability of modern packages could have been predicted back in 1990 when Rambus started.) Adding a Rambus interface to a chip involves reading and understanding some reasonable subset of the following: (1) Direct RAC interface. A total of something like 500 wires, just for a single Rambus channel interface. Plenty of stuff for a memory designer to do, 46 pages long: rambus.com PLUS (2) Rambus Memory Controller. If you want to go one layer higher, you can use the RMC from Rambus. You're still going to have deal with something like 350 wires, and in addition to the above manual, another 85 page manual: rambus.com PLUS (3) Rambus Memory Chip. You will also want to understand what some of the various modes that the Rambus chips use, which will require some understanding of the following, 62 page manual: rambus.com By contrast, a DDR SDRAM chip only takes a 66 page manual. DDR interfaces are so easy that they are not generally given as standard cells. Instead, engineers are expected to cook them up themselves. And a full 64-bit wide DDR channel is a lot cheaper (with much higher bandwidth than an RDRAM channel) and is only going to require something around 300 wires between I/O cells and your internal logic: micron.com Maybe PLUS Putting a DDR interface in a Xilinx Virtex FPGA is not totally trivial, as FPGAs are relatively slow, so Xilinx puts out an application note describing a decent interface. The note, which includes everything, is only 15 pages long: xilinx.com By contrast, putting a Rambus interface into an FPGA is impossible. Any rumors that Rambus has somehow simplified the memory design function are just plain wrong. It undoubtedly takes longer to design in a Rambus interface than a DDR SDRAM interface. (I should note that DDR SDRAM interfaces are more complicated than SDR SDRAM interfaces, but only in the data bus area, and not much more complicated there.) -- Carl

To: Bilow who wrote (41425)	5/6/2000 9:24:00 AM
From: blake_paterson	Respond to of 93625

Hi Carl: Good morning from down south. Thank you for your thorough response. I will try and be succint. <<(1) Gelsinger ain't running the desktop group any more. Wonder why? I don't.>> I don't have access to INTC's org charts, but my read of the press is that he is managing the guy/gal who is managing the desktop group. I don't see that as a demotion; quite the contrary. <<(2) ...the market price for RDRAM is still sky high. Where is that link that gives memory prices anyway?>> a. Back flap of May 9, 2000 PC Magazine: Dell Dimension 800 MHz P3 w/ 128MB RDRAM, 30GB HD, SONY Trinitron monitor, NVIDIA GEForce 256 Graphics, DVD, bladi bladi bladi US$2499. Compare with page 211, same issue: Micron 800 MHz P3 w/ 128MB 133 SDRAM, 27GB HD, Micron monitor, NVIDIA GEForce 256 Graphics, DVD, bladi bladi bladi US$2299. Last I checked, Dell and Samsung were in business to make a profit, and neither appeared to be inclined to "buy customers" or markets in 2000. I'll take the former system anyday over the latter, but what do I know. I'm Bubba, and all I want is a SONY monitor and speed. But guess whose money it is? b. I don't have time to find the links for you right now, but I remember seeing your responses to the actual posts (cellhigh was doing some for quite a while there). Also many on the Yeehaww!! thread. I'll look if I have time this weekend. 3. <<The key is what is being done to reduce the price of RDRAM. Intel is doing nothing to reduce the price of RDRAM, instead, their actions are supporting the price.>> INTC bankrolled much of the Capex required to convert to RDRAM production for all of the key memory players. I call that SUPPORT. Again, supply : demand of <whatever> determines price, unless you are in the Industrial Gases industry. (4) a. <<The full lifetime over product royalties for PS2 is mighty small compared to $200 per share.>> Call me crazy, but my calcs value the PS2 business ALONE at US$75/share right now. b. Wow! This is a long one. Where do I start? <<Only a manager would think that to hook up a memory system all you have to do is paste a couple of files from Rambus. The real world is a lot more complex.>> You guys (= memory designers; you are one, aren't you?) are much more important than that. You are CORRECT, you won't lose your jobs. I'm just wondering if the design decision control will be taken away by the RMBS standard, and that is the why of all the emotive response that we are seeing. <<Another group of people want to think of themselves as technologically literate. So they think that they should be able to analyze a company and determine whether that company's technology makes sense, and direct their investments accordingly. They look at Rambus, and it makes sense to them. The problem with these people is basic hubris. They have no idea how complex the many issues are in the industry.>> You are starting to RANT a little bit, Carl. Please be kind. Using your metric, I would sure hope you and 99.99% of the rest of the people out there don't put a penny in Drugs or Biotech, 'cause you guys don't have the SLIGHTEST CLUE of what is going on in those indutries. And the risks of technological failure outweigh your industry by 1,000 fold. 1:50,000 initial drug candidates makes it to the market; 1:100 of those ever makes it to blockbuster. But I don't wish for you or others to stay out. Catch my drift? What would happen to the equities market if we stayed out of these companies for the reasons you have so eloquently advised? <<But the only reasonable reason for sticking with the stock is the hope that Intel can force the technology onto the industry.>> I agree 100%. **OT OT: By the way, don't think for a minute that Buffet understands Pharma. He doesn't. I'm in it and I still don't understand it. Re: trading: involves more luck than anything else, IMHO. All the best, BP