PC800 RDRAM never really proved to be substantially superior to DDR266 memory in most real-world uses, but the real comparative advantages of RDRAM, we are ever told by Rambus backers, are just around the corner. RDRAM fans point to lower prices as a sign that RDRAM production problems are now solved. It is true that RDRAM lowers pin counts and thus makes motherboard design and production potentially cheaper. The Rambus camp is also claiming there are growing stability problems with running fast DDR memory types in three- and four-DIMM configurations. The case is made most elegantly on this page of this article by the always-excellent Johan DeGelas. He states in no uncertain terms: "DDR SDRAM technology is clearly reaching it's [sic] signal integrity limits." And the newest processors need faster memory to really perform as they should.
So will newer forms of DDR memory arrive in time to meet the demand, or will PC1066 offer the best way to feed a multi-gigahertz P4 beast once Intel unloads a 533MHz bus on us? Discuss.