To: Jdaasoc who wrote (77094 ) 8/10/2001 12:45:30 AM From: Bilow Read Replies (1) | Respond to of 93625 Hi Jdaasoc; Re: "I would have to believe that some varients of 512 Mbit SDRAM chips organized to make 128 or 256 MB DIMMS could be larger die size than 512 Mbit 4i RDRAM. " Yeah, you're a believer all right. Grasp at any straw. Re: "I have yet to see non-"niche" market for sub $1,000 desktop computers which will be calling for 8 512 Mbit SDRAM chips organized as a 512 MB DIMMs as standard fare anytime soon if the actual cost for the DIMMs were at acutal cost plus decent profit margin being charged for them. " You damn well know that it only requires 4 SDRAM chips to make a DIMM. Here's one in mass production at Samsung:samsungelectronics.com Your granularity calculation is already off by a factor of two, and that means we're down to a minimum DIMM memory size of 256MBytes. That's $18 on Price Watch right now, hardly a niche restrictive price. Honestly. Sometimes I think that Rambus morons believe that engineers are too stupid to get out of bed. What do you think we'll do if we run into a granularity problem? It doesn't take a genius from Stanford to answer this question. We've already faced this issue a dozen times in the past 20 years. We'll just add more pins to the memory chip, just like we've done so many times before. They already have the x32 JEDEC specs in both SDRAM and DDR. They're in massive production and are commonly used on graphics boards. Yes, they're somewhat more expensive than x16 chips (but a heck of a lot cheaper than RDRAM), but if mainstream memory has to go to x32, those graphics chips will become mainstream products and will become the cheap commodity version. The memory makers are already talking about x64 chips for graphics. That gets your granularity issue down to one. Think about it. This is what the industry has done over and over and over again, with not just DRAM, but also with SRAM &c. Granularity is not an issue, if pins are cheap enough, and with new bump packaging, pins are like water. The x16, x8 and x4 SDRAM chips are in a TSOP 54, while the x16, x8 and x4 DDR SDRAM chips go into a TSOP 66. The x32 SDRAMs are in a TSOP 86 and the x32 DDRs are in a TSOP 100. The package type is a tiny fraction of the DRAM's cost; this isn't going to make RDRAM cost competitive. You're basically spewing the same old tired Rambus BS from 2 years ago. Why haven't you at least updated it? At least look through the product guide from a couple memory makers before you go off on this BS. Where did you copy that argument from, Yahoo? Haven't I answered it enough without you bringing it back? Re: "My link is real and the 3% smaller die size is plain as day. " Yeah sure, and every "real" link to a source translated from a foreign language where a reporter interviewed an industry flack but didn't quote him fully and exactly is a great source of engineering fact. Especially if it reports something extremely surprising, BWAHAHAHAHAHA!!! Here's a link to the original copy of your link:Samsung begins mass production of third-generation 256Mb Rambus DRAM koreaherald.co.kr Here's their most recent article quoting Samsung about a DDR product, LOL!!!:Samsung unveils 300MHz graphic DDR DRAM chip May 25, 2001... The global DDR memory market is expected to grow to 1.4 billion units in 2002 from 160 million units this year. ... koreaherald.co.kr Do you get the impression, perhaps, that the Korea Herald is published in Lake Woebegon, where all products are better than average? As long as I'm on that kick, here's another Korea Herald article:Steep fall in DRAM price raises fears for Korea's exports June 7, 2001... Meantime, corporations grew wary of Rambus memory due to Intel's initial problems with it . RDRAM based on Rambus designs is the only memory available with Pentium 4 processors. Intel plans to offer "Brookdale," a Pentium 4 chipset that will work with more common synchronous DRAM (SDRAM), in the third quarter of this year. ... koreaherald.co.kr -- Carl