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.

Politics : Ask Michael Burke -- Ignore unavailable to you. Want to Upgrade?

To: Steve Lee who wrote (76873)	3/1/2000 8:38:00 PM
From: Ed Beers	Respond to of 132070

Steve, Did you just include the royalty on the memory and forget the chipset side? Surely Intel, Sony, and Nintendo all pay royalties on their use of the interface also. Ed

To: Steve Lee who wrote (76873)	3/1/2000 9:29:00 PM
From: Ed Beers	Respond to of 132070

RDRAM performance. There are two key performance criteria for a memory system: latency and throughput. For DRAM of any interface variant, latency is largely determined by the underlying DRAM cells and sense circuitry. RDRAM does not appear to have any latency advantages over other DRAM variants. Serialization and synchronization may even make latency worse than the more conventional interfaces. The only option to reduce the latency of DRAM main memory is to add cache memory that reduces latency when cache hits occur. Transfer rate is limited by transfer rate of each DRAM part multiplied by the number of parts. Therefore, it is easy to obtain a high transfer rate from a large memory array but difficult for a small memory array. The general techniques are to use wider data busses and interleaving. RDRAM allows very high data rates from small memory arrays. This is perfect for game machines where a single 8 MByte or 16 MByte part can be the entire memory. RDRAM has had significant design wins in game machines for this reason. Large memory arrays, with many memory chips, don't have big problems with the transfer rate to the DRAM parts themselves. That is why servers were among the last devices to move from EDO to SDRAM. I don't expect RDRAM to break out of the "very small total memory" niche unless they achieve cost parity with the other options. Remember that have to compete with all of the other ways to build a faster system: larger cache, faster cpu, larger memory, faster graphics card, faster disk drive, better chip set... There also an increasingly large segment of the market where performance is already adequate and lower cost is the primary goal. Good luck with your investment but I will bet against you if I can find puts at reasonable premiums. Ed

To: Steve Lee who wrote (76873)	3/1/2000 10:45:00 PM
From: Earlie	Read Replies (2) \| Respond to of 132070

Steve: Ok, let's have a look. Big picture In 1999, slightly over 100.0 million PCs were sold world- wide. This year, the cheerleaders expect the total to rise to 123.5 million units, but already, sales results indicate that this is just not likely (particularly with corporate sales virtually non-existent), so let's use 110.0 million units. What percentage of the whole PC market might RMBS address? Not much of it, I'm afraid. Here's why. By far, the vast majority of PC sales are in the bottom end, where Intel has already stated that RMBS won't cut the mustard, so let's knock off a minimum of 70% from our possible PC sales universe. As 1999 progressed, AMD's sales of micros moved steadily higher. In the last quarter, they shipped 6.0 million micros. This quarter, the company expects sales to be "flat to slightly higher", so let's use 6.0 million units as a reasonable assumption. In the next two quarters, since AMD has a recognized performance advantage, it makes sense to expect the company's sales to accelerate. Also, the last two quarters are the heavy sales quarters, and Dresden should be really cooking by then, so let's assume a bit of unit growth and call for 7.0 and 8.0 million units. Not hard to see that AMD's unit sales will be well north of 25.0 million units and could push 30.0 million units. Scratch 25% of the PC market with respect to RMBS sales potential. As I see it, and putting wishful thinking aside, During the remainder of 2000, RMBS might have a shot at sales into high end PC boxes that have "Intel inside". In a best case scenario, that works out to under 25.0 million boxes (.75 x .3 x 110.0 million). However this figure has to be modified by the fact that the current quarter will produce "nominal" sales at best and it will likely require an additional quarter to get things moving. Consequentially, we have to once again reduce the PC box universe that might be available to RMBS by perhaps a third. We are now down to 16.0 million boxes. What subtraction should we now use to account for Apple and the other peripheral players? Let's be generous and say 2.0 million. Hello 14.0 million available boxes. Since RMBS chips are considerably more expensive than SDram chips and since Intel has as much as admitted that for the next while, SDram will be "as good as RMBS", (do you want me to supply evidence of this, as I can), then once again, logic suggests a significant if not massive reduction of possible rides for RMBS chips. Should we cut by 50%? More than 50%? Or perhaps less? At this point, does it matter? And finally, let's make a final reduction as a result of the arrival of DDR chips which after all, have provided superior results in every test administered on a head-to-head basis. Should we cut the available universe again by 50%? More than 50%? Apologies for the lengthy commentary, but a total available universe of possible PC "plug-ins" for RMBS chips that is in the "low single digit" millions, is not going to provide anywhere near the revenues projected for year 2000. Will things get better in 2001? Perhaps, but both RMBS and Intel expect RMBS sales to climb only if serious price reductions occur as a result of high production volumes. The above-noted numbers will not do the trick. Also, unless Intel comes up with a miracle, AMD is likely to further erode Intel's market share, particularly if AMD's performance leadership continues. But I left the key argument to the end. Just how many people really need a super high performance PC? So far, the trend of the last few years has been a powerful shriveling of high end sales as the mid and low end PCs get better and better. Does Joe Six-Pack really need or even covet a 1 Ghz box? Not a chance! Sure, RMBS might gain penetration in telco applications (although the price competition there is getting to be just as brutal), or some esoteric wireless devices, but I doubt it. Those devices just don't require scads of memory capability compared with a PC. Dare we even bring up the dreaded word "profits"? I could address the technical considerations as well, but this post is far too long already. Suffice to say that at anything over $20., RMBS stock price appears "over-valued" Best, Earlie