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.

Strategies & Market Trends : Gorilla and King Portfolio Candidates -- Ignore unavailable to you. Want to Upgrade?

To: Apollo who wrote (5649)	8/31/1999 9:31:00 PM
From: Mike Buckley	Respond to of 54805

Stan and other Rambus officionados, I am not a techie, but DRDRAM would seem to be discontinuous in that it scales way up, as compared to SDRAM, A much more highly scalable product is not necessarily discontinuous. Remember that the discontinuous innovation requires that the customers "adopt new technologies and put in place new infrastructures that" were previously incompatible. In that context, what is discontinous about Rambus? If someone has already answered this question, please understand that I didn't take the time to read the 70 messages. :) --Mike Buckley

To: Apollo who wrote (5649)	9/1/1999 6:45:00 AM
From: unclewest	Read Replies (1) \| Respond to of 54805

stan, i saw your invitation and request for more info... this is it in layman's terms. this was explained to me by an ee friend who has worked with rambus architecture and has worked on sdram and ddram design projects. i am not a techie...i am a rmbs stock student. :o) rdrams are mounted on rimms. sdrams are on simms. the architectures are quite different, not interchangeable at any level. that is why rambus is considered revolutionary and not evolutionary. granularity increases are a huge issue for sdram due to its parallelism. a 512 Mbit part has 64 Mbytes in it (these will be available soon). a 256 has 32 Mbyts in it. with rambus if you want to add memory you can add one part. with an 8-bit wide sdram part you need to put 8 devices in parallel. with a 16-bit wide sdram part, you need to put 4 devices in parallel. this means, at these granularities,(which is where we are headed) the smallest addition to sdram memory will be 512 Mbytes in one case or 256 in the other. that is a lot of memory to pay for if you do not need it. with rambus you can just add what you need one part at a time. rambus is cheaper is guaranteed for smaller amounts of memory. OEMs don't want to ship 10x the memory that W2000 needs. rambus offers other important advantages as well. 300% bandwidth improvement is the biggie. this is needed for video, voice and graphics apps. this is very important to me in evaluating rmbs...the dram industry is projected at $25 billion this year, $40 b in 2000, and $60 b in 2001...rmbs ceo expects to collect royalties on 1/2 of that. estimated rate per cfo 1.7% rmbs also gets royalties on peripherals, rimms etc at up to 5%. total shares <25 million. rambus has no debt, $90 million in cash, and is currently profitable...these new revs flow straight down to the bottom line after taxes. i have seen estimates that demonstrate the total cost to convert the dram industry to rambus exceeds $20 billion. sony and toshiba spent $1 billion on one new rambus rdram fab this year and toshiba has others. the point is this committment has already been made...the industry has spent the money. this enormous financial committment not to mention the effort required to get some 100 major tech companies coordinated coupled with the fact that intel says current rambus speeds are good for at least 3-5 years represents a huge bar to entry for any competition. rambus says they can increase speed 100-200mhz per year going forward. intel makes approx 85% of the world's chipsets...intel says rambus goes in 100% of them by mid 2001. the chipset schedule is due for update today at the idf. unclewest