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: freeus who wrote (6820)	9/22/1999 12:42:00 PM
From: Charles R	Respond to of 54805

Folks, I caught wind of this thread recently. I was looking at RMBS as a potential investment vehicle but have not made any conclusions yet (other than that it is too pricey currently given the risk) and based on the dialogue on this thread over the past few days I thought this might be as good a forum as any to discuss RMBS. I am cutting and pasting a post I made to another thread a couple of days back. Before you read the attached clip, you may want to read my take on the technical side of RDRAM (Rambus's core technology): Message 11288868 **********begin clip***** It has been a few days since we discussed the technical aspects of RDRAM, PC133, DDR DRAM and looks like there is no major disagreement on the technical aspects so far. With that assumption let's do a application review and see where RDRAM can be successful. I am splitting the following in to two parts - short-term and long-term. Short being defined as something within the next 12- 18 months and long-term being anything later. SHORT TERM: - PCOAC: Most controversial are but potentially the most lucrative segment for RDRAM. The benefits are clear - low pincount and ability to support on-die graphics/3D controller. However, in the 12-18 month time frame it is unlikely that this solution would be cost effective for any system that uses more than say 32 MB per system. - High-end Game Players: This is clearly a winner for RDRAM. These systems need a good amount of bandwidth but do not need a lot of memory. RDRAM should serve this market well. - High-end Graphics: Many people have posted on this and it would seem that RDRAM would be a good choice in high-end graphics solutions because of the pincount issue. I can see the appeal but I am unsure how this would compete with embedded DRAM (on die or as MCM). My knowledge in this area is limited so am looking for comments from knowledgeable people in this regard. Also, it would be good to nail down the volumes in the high-end segemnt where RDRAM would be applicable. Is it hundreds of Ku or millions of units per year? - Integrated Graphics/Chipset: Right now no one other than Intel is pushing RDRAM in this market. Chipset vendors other than Intel sell low cost solutions and try to save every penny whereever they can. Given that RDRAM is not showing any real performance advantage for the mainstream desktop, it is not clear that this situation would change in the next 12-18 months. - Communication controllers: There is a new class of communication controllers that are being developed right now and it is immediately clear to me that RDRAM will be adopted into a lot of designs. The data streaming and pin count aspects make this a good candidate for this segments. The volume however is going to be light. My estimate: very likely to be tens of ku and no more than a few hundreds of ku per year. - Other: I have seen some talk about fancy applications such as HDTVs but I fail to see any applications with even tens of ku of volume in the near term. I would appreciate if anyone can point to applications with volume potential. LONG TERM: - PCOAC: This seems like the brightest market for RDRAM going forward but I wonder how much technology will change in year 2001. In the 0.13 generation will it be possible to cram a lot of memory on die sufficient to support needs of graphics and L2 cache? I have seen some suppliers touting very economical MCM BGAs - not sure what the cost penalties would be a few years out. Whether it is on-chip or on-die I am sure there will be some competition for RDRAM from embedded DRAM (could be custom, DDR, something else). The penetration here, IMHO, is going to be directly proportional to the cost/benefit of RDRAM compared to DDR DRAM and embedded DRAM. RDRAM definitely has a long way to go before it becomes cost competitive but there is definitely time to work on the cost issues. - Games: This should continue to be a winner for RDRAM. Embedded DRAM will start stealing market share over time but unless RDRAM guys become totally uncompetitive they should have decent market share as high-end game players migrate to low-end over time. The controller royalties would probably have to go down substantially to keep this market alive. - High-end Graphics: First of all I am not sure how big the high-end graphics chip business be after a couple of years. Most of the volumes will get sucked into some kind of integrated graphics/chipset or PCOAC and I keep wondering what the growth rate will be for the stand alone high-end graphics chip business (I am sure it it will turn negative this year and stay negative but I don't have any good guesstimates on the steepness of that drop). Even assuming this will continue to be a large segment we are already starting to see embedded DRAMs showing up from several vendors. I am hard pressed to see why this trend would reverse. Again, more knowledgeable people can comment. - Integrated Graphics/Chipset: One of the advantages of RDRAM over PC133 or DDR is die size savings because of lower pin count. Looking two years out, this will reduce for two reasons. Firstly, DDR will be the mainstream and DDR narrows the per pin advantage of RDRAM (compared to PC133). Secondly, advanced semiconductor packaging technology will become more accessible to chipset players. Currently C4-flip chip is not accessible to the chipset guys but two years out it (or some equivalent) should be accessible to the chipset guys. I also see a lot of potential for VC which can further reduce the reason for the chipset guys to go to RDRAM. The market share RDRAM can gain in this segment is going to be largely a function of how rapidly RDRAM guys can reduce the costs and how narrow the gap is with DDR. (I would love to hear what people think on how RDRAM costs will compare with DDR over time) - Communication controllers: I can see this as being the growth area. LSI putting RDRAM controller in their coreware library will definitely help adoption rate. But volume are unlikely to reach anything more than a few million units for a long time to come. And, even here, embedded DRAM should put a good show. I think it will be tough to predict market shares this early in the game. - Other: I can see some potential growth in areas like set top boxes but unless there is a change in display technology, I am hard pressed to see the need for a lot of bandwidth (i.e., we gotto move in volume to something other than the plain old TV). Some of the futurists on the thread can shed light into some potential high volume areas. Whew! That was longer than I thought it would be. Looking forward to your thoughts on the subject. ********** End clip ************* Looking forward to some itneresting posts. Chuck