Richard,
Thanks for posting the pointer to the documents.
It is reassuring to see that the court saw through Rambus' obfuscation on transaction size. Their claims describe transaction size as a part of each transaction. This is completely different from the operation of SDRAM and DDR, where transaction size is set during initializatin and cannot be adjusted on each transaction.
"Moreover, the construction urged by Rambus utterly ignores the clear language of the claim that block size is associated with a transaction request. (See e.g., '918 Patent, Claim 1-7 and all other claims (8 through 38) dependent upon Claim 1-7)." page 46. Memorandum Opinion Civil Action No. 3:00cv24.
It is so easy to see Rambus action in retrospect, they developed a new way to move data from memory to cpu in isolation, dismissing all previous designs. Got the patent and started licensing the pin-saving RDRAM devices. Then existing DRAM makers, mostly Japanese at the time, dismissed the new concepts. From my exerience the DRAM departments were so fat and happy at the time they didn't want to change a thing, in addition they like simple simple simple logic in their DRAMs. But bus speeds were increasing, and at some point Hitachi made SDRAM to improve the performance of SH-2 processors in Sega Saturns. Hitachi controlled both controller and memory, made cooperation easier. This was not a huge leap, Hitachi had a DRAM controller in their SH-1, so they just moved the controller from the CPU to the memory chip, passed a clock over and viola, an SDRAM. Rambus recrafted their continuance patents to try to cover SDRAM, and they shook down the DRAMurai for money. Hitachi was sick of their losses in DRAM after Samsung destroyed the market so they rolled over.
The future, 0 external DRAM, everything embedded. 0 royalties for RAMBUS. Think of that hungry chipset on the mother board, it has eaten practically ever other chip on the board, only DRAM is left. When the DRAM is embedded the power used goes down and the performance goes way up, due to very wide access.
Paul |
