INPHI WAS SHOWING off working 32GB LRDIMMs at IDF this year, upping maxing the memory of a Nehalem-EP system with only 2/3rds of the slots filled. If you want more than a mere 384GB in your box, you can up that to up that to 8TB of DRAM with an 8-way Beckton box.
Inphi 32GB LRDIMM
The first thing to make clear is that Inphi doesn’t make DIMMs per se, they make iMBs or Isolation Memory Buffers that go on to DIMMs. Those IMBs conform to the JEDEC LRDIMM, or Load Reduced DIMM spec, basically a standards based version of MetaRAM.
Inphi LRDIMM test board and 16GB module
Put one of those on a DIMM, and you can pack in a lot of memory. The 32GB module, above, uses 4 ranks of high density 4Gb DRAM dies stacked into 8Gb packages. If you put 36 of those on a board, you end up with 32GB of ECC DDR3/1333 on a DIMM. If that isn’t enough for you, the iMB is capable of 8 ranks, and DRAM is easy enough to stack.
Nehalem-EP maxed out
A theoretical 64GB 8 rank version of this LRDIMM could theoretically max out the memory of a Nehalem-EP using only one DIMM slot per channel. While each one these sticks may cost more than the rest of the server, that price differential won’t last for long, it never does.
With the current technology, you can take the old 2TB max from a ‘Beckton’ Nehalem-EX system and quadruple it to 8TB. Yes, you will soon be able to buy an 8 socket Westmere-EX system that sports 8TB of memory using off the shelf components. It may cost more than the data center it sits in, but who cares about that? An 8 rank 128GB LRDIMM built with 4 high stacks of 4Gb DRAMs would up the total to 32TB per system, but we at SemiAccurate central see that as a bit excessive. Until next year.S|A
Latest posts by Charlie Demerjian (see all)
- Intel talks about Knights Landing architecture - Mar 31, 2015
- Gigabyte’s ARM servers represent a sea change in the market - Mar 30, 2015
- VESA shows off working USB-C Alt Mode with Displayport - Mar 24, 2015
- UzBrain’s Rail Gun turns a toy weapon into an FPS controller - Mar 18, 2015
- Fotonation uses computational imaging to focus faces - Mar 17, 2015