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 : Intel Corporation (INTC) -- Ignore unavailable to you. Want to Upgrade?

To: kash johal who wrote (82406)	6/1/1999 10:28:00 PM
From: Elmer	Read Replies (1) \| Respond to of 186894

Re: "SO again for this years systems what does RDRAM buy over PC 133 in projected system performance." Kash, it would seem that 133MHz is 133MHz, no matter who is feeding it. The advantage of RamBus is that because of the low pincount, the memory hub can afford to have multiple banks of RDRAM while only adding a small number of pins. From a purely CPU standpoint that doesn't mean a whole lot but from a system standpoint that could be a big factor. To access another bank do you want to add 16 pins for another 800MHz RamBus port or 128 pins for 100MHz SDRAM? EP

To: kash johal who wrote (82406)	6/2/1999 1:17:00 AM
From: Tenchusatsu	Read Replies (2) \| Respond to of 186894

<for this years systems what does RDRAM buy over PC 133 in projected system performance.> I can tell you that RDRAM will perform better in systems which have multiple sources all accessing memory concurrently, like the Coppermine processor, AGP-4x graphics card, and PCI devices (A3D sound card, network card, disk controller). PCI by itself won't present much of a load to RDRAM, nor will Coppermine's 133 MHz bus, nor will any AGP-4x card. But all of them bursting traffic onto RDRAM at random times is exactly the kind of load that RDRAM can better handle compared to any brand of SDRAM. <In addition I am hearing folks now saying that DDR SDRAM will happen mid to late next year. At which point it is similar in preformance to RDRAM.> DDR SDRAM has problems of their own. Anyone who says that it's an easy transition from PC133 to DDR isn't giving the whole picture. No way will we see any DDR SDRAM products until 2000 at the earliest. Plus, because of the higher speeds, DDR SDRAM has a higher pincount than regular SDRAM, not to mention RDRAM. One of the trends we'll see in future microprocessors is the integration of memory controllers onto the die. RDRAM's low pincount makes it ideal for this kind of integration. In comparison, DDR SDRAM will have four to five times the number of pins of RDRAM. Try convincing a computer architect to add that many number of pins to a processor package. Finally, although DDR SDRAM might have a higher theoretical bandwidth than RDRAM (2.1 GB/sec vs. 1.6 GB/sec), it won't be able to use that bandwidth as efficiently as RDRAM. <Experts at IBM and other places are saying that it buys nothing!!!> I don't think they're saying that, although the market is taking it this way. Rather, I think IBM thinks the return on investment isn't high enough to warrant pushing RDRAM as hard as Intel is. And like I said before, I can understand this point of view, even if it's only delaying the inevitable. Tenchusatsu