RDRAM volumes.
Ignoring DDR as a factor in the discussion, it seem's the cost's of RDRAM implementation, (chip's, PCB's, RIMM's, mobo's, test and QA), currently are limiting factors to volume production of RDRAM. The only effective way to decrease costs is to generate large volumes of platforms using RAMBUS.
But beyond the 820 and 840 for high end pc's and W/S's, the platforms coming out now are SDRAM based. (For example, the 810 and 815 chip set platforms. And the BX chip set apparently will have a longer life than anticipated).
Even Timna, which has an rdram bus, appears to be initially supported by Intel mobo's with an SDRAM hub. (What's the message to third party mobo suppliers?)
Of course Willamette will be RDRAM based, but significant volumes are at least a year away. Considering the large expenditures made to create RDRAM as the industry standard memory going forward, why not RDRAM in every platform? The answer probably lies in the ramp of RDRAM. (A chicken/egg situation if there ever was one). <g> Major shifts in technology that require significant changes to legacy design, manufacture and test aren't easy.
JMHO's |
