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.

Technology Stocks : Rambus (RMBS) - Eagle or Penguin -- Ignore unavailable to you. Want to Upgrade?

To: pompsander who wrote (44077)	6/11/2000 7:45:00 PM
From: Bilow	Respond to of 93625

Hi pompsander; Re DDR not being out there already... The basic problem is that current processors do fine with the current level of memory bandwidth performance. At 100MHz, an 8-byte wide GTL+ P6 interface can only move 0.1GHz * 8Bytes = 0.8GB/sec. And that is before the bandwidth that is provided by cache is included. In short, there is no reason to put a 1.6, 2.1, or 2.6GB/sec memory interface onto a 0.8GB/sec processor. So why should they? In order to really need higher bandwidth, you have to go to faster FSBs or dual processors. Athlon does have a fast enough FSB to make use of DDR, but it has only been out for a year and was a surprise winner. Consequently, the independent chipset makers (i.e. VIA, ALi, SiS) haven't had time to get the chipsets out. They will get here. Listen to three minutes of the following (rather long) Q&A presentation from VIA. 33:00 to 36:00. The speaker, Dean Hays of VIA, talks about that earlier DDR solution, and why it wasn't mass produced. He also talks about what is different this time, and makes a reference to RDRAM's high cost: silicontech.com Due to this lack of need for high bandwidth memory in PCs, I expect to see the dual use VIA chipsets (which are compatible with both DDR and SDRAM) ship significant amounts of SDRAM, particularly for their Intel products. One would have thought that maybe Timna could have made use of RDRAM's bandwidth, since it doesn't have a FSB, but I guess it wasn't a significant enough performance improvement to matter to Intel. My guess is that even if the chip could use the high bandwidth of RDRAM, there was still no reason to use it because of cache memory doing its job well. We forget that when SDRAM first came out, it was marketed as a memory that would provide the same system performance as DRAM, but without the cache memory. In other words, it was marketed the same as RDRAM, a solution that provides unneeded headroom. Naturally, SDRAM wasn't used by PCs until the processors actually needed it, or more likely, when the cost per bit crossed under the DRAM price. This problem with DDR being too fast to be necessary for PCs wasn't present in graphics. Consequently, as soon as a graphics house (Nvidia) had the need for more memory bandwidth, they folded DDR in immediately. Other graphics houses with DDR chipsets in design or early testing are ATI, Matrox, 3Dfx, and S3. They will have their stuff out by the end of the year. -- Carl