JEDEC Fiddles With DDR4 While LRDIMM Burns 67
An anonymous reader writes "JEDEC hasn't finalized the upcoming DDR4 standard yet, but it seems they left out licensing some crucial IP for (the already finalized and shipping) LRDIMMs (for use on data center servers). As a result they are only produced by one source which is facing some hurdles justifying their copying of IP. This article discusses how DDR4 is based on LRDIMMs and the future of memory. Quoting: 'JEDEC finalized the LRDIMM standard without securing licensing on load reduction and rank multiplication. Inphi, currently the only maker of LRDIMM buffer chipsets – others have backed off – lost a challenge of Netlist IP at the USPTO. As a result the Netlist patents have become stronger and are going to come back and bite Inphi in Netlist vs. Inphi, which was stayed pending these patent reexaminations – patents which survive re-examination can never again be challenged in court. NLST patents ’537 and ’274 survived with all claims intact, which is a powerful statement on the strength of their IP – Inphi has appealed to the BPAI, but the USPTO decision is telling.'"
Wrong (Score:5, Informative)
"patents which survive re-examination can never again be challenged in court"
Wrong. A party that files an inter partes reexamination petition and that petition for inter partes reexamination is granted by the PTO (initiating an inter partes reexamination), may not "assert[] at a later time, in any civil action. . . the invalidity of any claim finally determined to be valid and patentable on any ground which the third-party requester raised or could have raised during the inter partes reexamination proceedings." 35 USC 315(c)
So, anyone other than Inphi (or its privies) may challenge the patents' validity in Court on any grounds, including the prior art that was before the PTO in the Inphi v. Netlist reexam. Moreover, Inphi could challenge the patents on "any ground which" Inphi did not "raise" and "could" not "have raised" in the inter partes reexamination. This would include, at a minimum, any ground for invalidity other than prior art in the form of other patents printed publications. 35 USC 301, 311(a). I.e., invalidity under 35 USC 101 ( patentable subject matter, utility), 112 (indefiniteness, written description, enablement) and 102(b) (public use, on sale).
Also, Netlist amended a bunch of claims, which typically raises major worries about validity for written description.
Re:WTF Is this shit (Score:4, Informative)
Layman speculation, don't take this as gospel. I think it's mostly accurate, though.
JEDEC is a big consortium of technology companies that designs RAM standards. SDRAM, DDR, DDR2, DDR3, DDR4? Those were all designed and standardized by JEDEC.
As for LRDIMMs: all sorts of problems start to pop up when you start adding more than 8 or so sticks of RAM per processor. I'm not an electrical engineer, but the basic gist is "things do not work properly". There have been various schemes to combat this in the past: FB-DIMMs (fully buffered), R-DIMMs (registered), and probably some others I can't think of; they generally involve some sort of tradeoff to beat that whole more-than-8-sticks-of-RAM problem. LRDIMMs are the "next generation" of this. Useful for data centres and servers where you need a huge amount of memory in a small space.
Anyway, it appears that JEDEC finalized the LRDIMM standard, but failed to license a few very important patents. So it seems that this one company (Inphi) went ahead and started making chips for LRDIMMs, but didn't realize (or didn't care) that they were infringing on some other company (Netlist)'s patents. Netlist sued for infringement and won. This is a troublesome situation because it seems JEDEC has created a standard which requires third-party licensing. The reason that is bad is that Netlist could completely choke off the supply of chips for LRDIMMs if they decide to set their license fees too high. Or Netlist could try to gain a monopoly as the only producer of these chips. It's possible for Netlist to really screw with adoption of LRDIMMs.
And there's some mention that DDR4 (not yet finalized) may have similarities to the LRDIMM standard, which may have similar consequences, just for DDR4.
The article isn't that great. It seems to be more of a financial hot-tips kind of thing, as RichMan somewhat suggests elsewhere in these comments.