DDR to DDR2

As with the move from SDR (single data rate) SDRAM to DDR SDRAM was a logical move, so is the choice from DDR to DDR2. Unlike the move from SDR to DDR however, the architectural changes between DDR and DDR2 are more evolutionary. For those who remember some math from the senior years of high school will recall what a sine wave looks like. Computer clocks signals look something like a sine wave but squared off. With SDR, data was only transferred on one edge of the clock cycle, the rising edge. DDR introduced the concept of transferring data on both the rising and falling clock edges resulting in double the theoretical performance.

DDR has continually risen in speed. Starting from the days of DDR200 we now have memory that is hitting the DDR500 mark but lack of widely available high speed memory beyond the DDR400 spec implies that continually raising the core speed alone will not cut it.

DDR2 takes a different approach, the clock rate is dropped but amount of data transferred per clock cycle is again doubled, not by a clocking trick but a second memory chip is used in parallel. In essence, the actual clock rate of each individual memory chip can be halved while maintaining the same data rate as DDR. The main advantages are twofold; chips need not be clocked as fast so the scaling problem that is affecting the high end of the DDR market is not yet relevant here; secondly the lower clockrate means a reduction in heat and subsequently, a lower voltage.

The current problem of course is the additional cost of DDR2, partially because of the second chip which accounts for at least some the price differential. I am sort of surprised that DDR2 did not make its way over to the laptop market initially where a higher cost is offset by the less than cutthroat pricing on the PC side. Laptops would also benefit from the lower voltage requirements of DDR2. Nonetheless, most 915G and 925x boards will utilize DDR2 and not regular DDR.