To: Knighty Tin who wrote (62004 ) 6/10/1999 8:13:00 PM From: Robert Read Replies (2) | Respond to of 132070
MB and the thread -- That article on RMBS gave me some nice chuckles, and when I showed it to a buddy of mine who is a senior microprocessor designer he didn't stop laughing for a few minutes. I think it unintentionally brings to light the whole RMBS debate. In order: 1. "Die size remains a concern, suppliers conceded, but all of them maintained that neither the 10-to-25-percent die-size penalty nor a higher price tag should hamper the acceptance of RDRAM for high-performance PCs." Are you kidding me! The 2 big variables in manufacturing cost are die size and yield. 2. "The panel, organized at the Warburg Dillon Read technology conference, included Rambus Inc. chief executive Geoff Tate along with representatives from DRAM suppliers and test and assembly executives." Translated, the quotes in this piece are all from industry shills and RMBS mouth pieces. 3. "Die size is being sacrificed as manufacturers concentrate on keeping yields high, said Avo Kanadjian, senior vice president of memory marketing for Samsung." So, they avoid the disaster of larger die size AND worse yields, but larger die size still means fewer devices per wafer. Thus the higher cost. 4. "Rambus' price premium over SDRAM was not seen as a barrier by panelists, who believe that PC OEMs will pay extra for the added performance." Must be news to the OEMs! <vbg> 5. "Rambus remains more expensive than SDRAMs, but panelists refused to classify the two as competitors, saying they aim for different uses . . . Panelists' belief was that rather than compete with Rambus, the PC133 architecture will be relegated to the bottom of the sub-$1,000 PC zone" This is both incredibly insulting to PC133 and also totally wrong. Relegated not just to the sub-$1000 zone, but the BOTTOM of the sub-$1000 PC zone! Also, "aim for different uses"? Huh? I think for virtually ALL applications, PC133 is more than good enough. Only 3D games require the additional performance. 6. "Memories account for 5 to 7 percent of a PC's cost these days, Kanadjian said, and panelists believe PC OEMs are willing to see that inch up to near 10 percent for Rambus." Again, I bet this is news to the OEMs. At a time with even the most die hard PC sector touts on Wall Street coming around and agreeing with the "Nuclear Winter in PC land" thesis now -- margins falling, revenue growth nonexistent, etc. -- I just don't see how the fraction of the total cost of a box devoted to memory GOES UP to 10%. 7. "Tate of Rambus said his goal, admittedly "aggressive," was to have RDRAM match SDRAM for testability, packaging and PC-board cost by the end of 2000 . . . Tate said that money from high-end PCs — which he said account for 20 percent of the PC market — should suffice to fund engineering improvements to drive down costs, which in turn would help Rambus find eventual acceptance in mainstream PCs." Can you say, "wishful thinking"? In short, RDRAM is a neat technology, but it is not nearly as revolutionary as was originally advertised. Is it worth the extra cost? Clearly the answer is no in most cases. So the real question becomes how well a marketing job RMBS and its allies do in convincing people to buy something they don't really need. And with everything migrating towards the sub-$1000 market, I certainly would not want to be trapped in the high end segment of the market. My only worry is that the mo-mo idiots might bid RMBS up on the hype and hope. But otherwise, dem poots lookin' mighty fine.