DDR2 - An Overview Print E-mail
Written by Michael Schuette   
Dec 30, 2002 at 12:00 AM

DDR (I) is approaching the end of its dominating role in the desktop space to be replaced by the second generation of double data rate memory starting at 400 MHz and using a conventional 64/72 bit parallel bus interface. Lower operating voltage, new features like on-die termination, off-chip driver calibration, Posted CAS and variable write latency sound intriguing but what is really lurking behind the new standard. We have seen claims of up to 72% power savings over DDR I and other miracle cures for any bandwidth issues faced by the PC industry. We have done the math ourselves found a few convenient errors in some of the manufacturer's descriptions of DDR2. Otherwise, DDR2 looks very promising but who will benefit in the end? If the DRAM makers have their way, DDR or DDR I as we know it will be succeeded at some point in the coming year by the next generation of mainstream SDRAM: DDR2 The second generation of DDR has been anticipated by consumers and DRAM manufacturers alike, however, for rather different reasons.

Hypothetical DDR2 Registered ECC DIMM using Elpida components (altered after Elpida). The pin count on the DIMM has increased from 184 pins in DDR (I) to 240 pins in the new DDR2 format.

Most consumers primarily associate higher performance with DDR2. Starting speed grades of 400 MHz data rate for system memory and already going up to 1 GHz in graphics applications, even before the new technology has hit the streets sound like some real progress has been made in DRAM technology. This is especially true in light of the difficulties in current DDR I technology to yield DDR400 MHz components in sufficient quantity.

So what are the magic ingredients that makes DDR2 wiggle so much faster than DDR I? More importantly, we have heard all the rumors of DDR2 being backward compatible with DDR I but we also heard the opposite. For the graphics card manufacturers, the compatibility may not be such a crucial factor but what will happen to the consumer who wants to upgrade his memory to DDR2 on an existing mainboard? Last not least, since there is no such thing as a free lunch, will DDR2 be more expensive than current DDR (I) or what other trade-offs will there be e.g. in terms of latency vs. bandwidth?

We have seen the wildest speculations about DDR2 floating on the web, including the use of very inventive math by some DRAM manufacturers to hi-light the benefits of the new standard. We have heard terms like "DDR2 in emulation mode" and "poor man's DDR", we have seen articles using all the correct buzz words albeit somewhat out of context. But what is really behind the new technology?



Last Updated ( Dec 10, 2009 at 05:43 AM )