Tenchusatsu,
It's great seeing the technical discussion going on here. "The truth shall set you free", and a wide-open understanding of RDRAM and comparison to the competition is what's required to make everyone feel comfortable that Rambus and Intel can do what they claim they're doing, and do it better than any competitive products.
Having said that, I found one of your statements slightly misleading, unless I'm missing something here. With respect to the speed of the data lines in RDRAM, it sounded as if you were saying that the data lines moved at 800Mhz, twice as fast as the clock. But the data lines are not going to change any faster than the clock, I believe. You're going to get a clock transition (up or down) latching the data, then the data is going to change (up or down), then the clock will transition (down or up) latching the data again, then the data is going to change, and so on. You get one transition on the data line for each transition on the clock, which means that they run at a maximum of the same frequency (400 Mhz). Did I misinterpret your response? (I once had a discussion on the board with someone about the fact that the maximum rate on any line was 400Mhz, but because you get data on both the rising and falling edges, you get 800Mbps, but you still wouldn't call it 800Mhz).
Dave |