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 (43560)	6/6/2000 8:37:00 AM
From: blake_paterson	Read Replies (4) \| Respond to of 93625

Pulling an all nighter, Carl? This could explain your rant... or are you in Taiwan? BP

To: Bilow who wrote (43560)	6/6/2000 9:07:00 AM
From: Bilow	Respond to of 93625

Hi all; A story of how Intel lost their lock on the chipset business... Back in '97, Intel had a lock on the chipset business for their processors. They decided one day to use that lock to push the memory makers to produce what turned out to be an incredibly bad idea for a memory interface. This post was prescient: Aug 28, 1997 To me that says that if the worlds DRAM manufacturers want to sell DRAMS into PC's using Intel chipsets, then they will be rambus DRAMS. Do you, or anybody else, have a different interpretation of these public statements? Perhaps you believe that Intel will no longer control the chipset business Can you present a case where anyone else can fill that position? Who else has the process technology AND the capacity AND the design expertise? I don't see anyone else fitting the bill and having the marketing clout to convince Compaq, Dell, HP,Gateway etc etc to switch. #reply-2076483 Elmer Intel chose Rambus but it is now clear that they failed to bring it to commodity status. There are now something like 14 DDR chipsets underdevelopment, most of them by companies other than Intel. Where once Intel dominated the business, VIA now hopes to hold 50% of the market by the end of the year. Just how did Intel lose control of the chipset business? A basic principle of engineering design is to avoid designing in specialty parts, when possible. Specialty parts are those available from only a single supplier, or available only in limited volume. Just about all designs will end up with some specialty parts, a great example would be an Athlon (or Pentium) processor chip. But just because you lost your virginity when you were forced to choose one specialty part doesn't mean that you should go around choosing more of them. Every specialty part that goes into a design is another opportunity for a disaster. So good designers minimize them. One common technique is to make a design that can get extra performance out of a specialty part, but is still compatible with the standard parts. A great example of that is the VIA chipset that supports Virtual Channel DRAM - an expensive specialty memory, but also allows cheap, standard SDRAM. The new VIA chipsets that allow both DDR and SDRAM are another example of this very basic principle. It's not about giving the customer choices, it's about giving your parts procurement department choices. The problem with Rambus is that it is so different from SDRAM that it is very difficult to make a memory controller that is compatible with both types. As long as the memory controller is kept off the CPU chip, this merely means that you need to design backup chipsets. That is, while you design a chipset to support RDRAM, you also design one to support SDRAM (or DDR). Where you get into real trouble is when you bring the memory controller onto the CPU die. Then if the memory type turns out to be a problem, your design will be in big trouble. Timna is a perfect example of this. In addition, Intel failed to design backup chipsets, and lost a substantial portion of their chipset market to a host of previously new risen competitors. The principle of avoiding specialty parts is ancient and quite revered in the engineering design community. When Intel tied themselves to Rambus it was widely known to be a risky proposition. I am sure that this decision was made at the upper levels of Intel management, not by the design engineers, who inevitably have these principles drilled into them the hard way. At the time that Intel made its fateful decision, I wrote: Aug 29, 1997 Unless you want to take the chance of making a chip that will only work with a memory type that is unavailable, you never, ever, ever design a processor that runs with only a single type of RAM. Memory is always going on allocation, and that means that your CPU sales will be limited, and your competitor capable of using the RAM types not on allocation grabs your sales. Designing only RAMBUS compatible processors would be an incredible gift to the competition, and isn't going to happen. Even generic DRAM sometimes goes on allocation. Specialty RAM does it all the time. Bilow Intel took the chance of making a chip (i.e. Timna, i820 and i840) that will only work with a single memory type. They lost. As I suggested above, their competitors grabbed their sales. Not only did Intel breathe life into VIA, ALi, and SiS, they also reinvigorated AMD. -- Carl