There are a number of next generation memory technologies on the horizon that hold great promise to meet the evolving needs of consumer devices and enterprise storage systems and applications. Some have been in development for a number of years, and are close to a critical turning point that will see them widely adopted. Here’s a few that merit watching in the next year, including several DRAM alternatives.
DDR4: Years in the making
The next iteration of DRAM has been in the making for a number of years with initial manufacturing by some vendors in early 2011 and release of the initial JEDEC standard in September 2012. Recent speculation
suggests Intel will be incorporating DDR4 into its high-end computers in the third quarter, and that it will take 12 to 18 months before DDR4 finds its way into laptops and PCs. The standard is already expected to receive an update by JEDEC
. DDR4 DRAM standard doubled the speed as well as improving performance scalability, capacity, and power efficiency in comparison to its predecessor.
Hybrid Memory Cube: The next DRAM?
With DDR4 only now starting to hit the market, there’s time to figure out what its replacement might be, and it may not be DDR5. Both MRAM and RRAM hold the promise of being able to instantly resume a computing session after the machine has been powered off. Another alternative to DRAM is the Hybrid Memory Cube (HMC)
, a concept that dates back to 2006. HMC has been described as a complete paradigm shift from current memory architectures, combining high-speed logic process technology with a stack of through-silicon-via bonded memory die. It is aimed at applications ranging from high performance computing to consumer technologies such as tablets and graphics cards.
UFS: The inevitable transition from eMMC
The ability of Universal Flash Storage (UFS) to support full duplexing, where reading and writing occur between host processor and UFS device at the same, is one of the primary reasons it’s expected to eclipse eMMC in the long run
, as it provides a significant boost in performance that accommodates features such as higher quality video and faster network speeds on mobile devices. The latest specification from JEDEC addresses mobile applications and computing systems that need high performance and low power consumption. UFS 2.0 exceeds current SSD interface of SATA 3.0 (6 Gbit/s) and supports up to Gear 3 at 600 Mbit/s per lane with multi-lane support. Adoption of UFS will likely pick up in 2015, according to Gartner, with eMMC sticking around for years to come.
3D NAND: Scaling beyond NAND Flash
This will be a year of growing pains for 3D NAND. A key hurdle for its adoption will be developing a cost-effective production process if it is to be a viable successor to 2D NAND. Jim Handy, principal analyst at Objective Analysis, recently authored comprehensive series of blogs
looking at 3D NAND development and its challenges. For now, conventional NAND components are likely to maintain their cost advantage through 2015.
NVDIMM: Where storage and memory merge
With the continuing adoption of SSDs in the datacenter, memory and storage are becoming more tightly intertwined. The Storage Networking Industry Association has seen a need for a bridge technology to prepare industry for that reality
. Enter NVDIMMs, persistent memory modules residing on the DDR DRAM channel, combining volatile DRAM memory and non-volatile flash memory. An NVDIMM operates exactly like a regular DRAM module, but during a power failure or system crash, it is powered by a supercapacitor pack. The data within the DRAM is transferred to the flash and can be restored to the DRAM when normal conditions resume. Applications for NVDIMMs include applications that run at high speeds and those that can’t afford downtime.
In its recently released "Global Next Generation Memory Market (2012 -2017)" report, Research and Markets cites faster switching time, high endurance, and power efficiency as the major drivers for the next generation memory market, with design cost as its major hurdle.