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 (36156)	1/2/2000 1:44:00 PM
From: jim kelley	Read Replies (1) \| Respond to of 93625

....My mistake, I meant to say 128 MB of DRAM. I do not think you will be seeing 128 MB of DRAM embedded on a single chip anytime soon. So your embedded argument is seriously flawed. DELL is expecting to be 50% RDRAM by the middle of this year. I believe them. :)

To: Bilow who wrote (36156)	1/3/2000 1:55:00 AM
From: Bilow	Read Replies (6) \| Respond to of 93625

Hi all; While Samsung is saying whatever it can to try and keep the failing body alive, it is quite obvious to us design engineers that RDRAM is dead, dead, dead. I should explain Samsung's position. Since Samsung is pretty much the only memory maker that managed to produce decent quantities of direct RDRAM, they would love for the industry to move to that technology. It would mean that they would get a big lead on the rest of the industry, they would end up minting money. In order to do this, they have to convince design engineers that RDRAM is here to stay. To that end, they are going to widely publicize their efforts in that area. Note that they are also developing DDR DRAM, as that is what the industry is quite obviously going to. Go take a look at the Samsung web site. They've got 256Mb DDR in development, and 128Mb DDR sampling. At the very least they are hedging their bets, as are all the memory makers. They would love for RDRAM to survive, it would be a windfall for them, but the rest of the industry isn't making it happen. Engineers prefer to use components that have multiple sources. The reason is so that you don't end up in a position where you are on "allocation". Allocation means that your company has to wait for components, and that really puts the manufacturing people in a bad position. The failure of the memory making community to come to the support of RDRAM (and, therefore, Samsung,) is not good news for Rambus. Because of its difficult interface, RDRAM was always hard to sell to design engineers. But the problems that surfaced last year put the nails in the coffin. The technology is no longer being specified by design engineers, at least not the ones that want product that ships easily. Anyway, Samsung has its obvious reasons for trying to get the zombie back out of the grave. The other memory companies don't have that reason, and they are not supporting RDRAM beyond what Intel paid them to. Now Intel is abandoning the technology as well. Micron has released its "confidential" design guide for DRAM over the next five years, and it is perfectly in line with what memory design engineers are thinking, and with what I have been saying. I quote from their pdf files below, but first, I should explain some terms... "5+ Year Compliance" means being able to buy more of the same part, or another part that complies with the guarantees of the original part, for five years into the future. This is a necessary requirement for engineers designing new designs for long life and mass production. "TSOP" and "TQFP" refer to the package. These package types are far cheaper than direct RDRAM's chip scale packaging. Designers who are going to be specifying a lot of DRAM prefer to use the cheaper packages. Engineering is mostly about saving money through the reduction of resource usage, in this case, package costs. (Engineering school, on the other hand, is largely about trying to get prototypes to work. But that is where world class engineering begins, the race actually gets won by the team that puts together the cheapest design to satisfy requirements. Engineering is largely about money.) The Micron document dates from last month, and is in the form of a collection of overhead displays. These are being shown to design engineers all over this country when they invite Micron in to talk about their memory products. This is sales literature for sales to design engineers. There would be an applications engineer standing next to the projector and talking about these slides. Some excerpts from the pdf file: DRAM Design Guide for 5+ Year Compliance 54-pin TSOP SDRAM (x4, x8, x16) Medium to large memory arrays PC main memory 50-pin TSOP SDRAM (x16) 86-pin TSOP SDRAM (x32) Minimum memory arrays PC graphics 66-pin TSOP DDRAM (x4, x8, x16) Medium to large memory arrays Mainframe, servers, workstations and routers 100-pin TQFP DDRAM (x32) Minimum memory arrays High end graphics and routers micron.com Those of you who are investing based on the future of memory should take a good look at what memory design engineers are seeing today. This is information about what kinds of memory will be designed into products during 2000, it is not information about what kinds of memory were designed into products last year. Stuff that was designed last year is a lot more likely to have RDRAM, and you will continue to see product announcements from previous year's design wins. But the place to look is to the future, not the past. Design wins that RDRAM took last year are subject to redesign, as are pretty much any design win. That is what Intel is doing as fast as it can, with DDR and PC133 support coming out this year. Since last year, there has been a change in Rambus' ability to sell to its customers, design engineers. Rambus screwed up incredibly badly, and they are now off the road maps of most design engineers. We just can't afford to design in a technology that even Intel can't make work. In the above sales literature, note that RDRAM isn't listed as an option, even though it appears as a possiblity on the front page. Micron makes RDRAM. But they aren't trying to sell it. Rambus is quite dead. -- Carl