5G not only represents a vast leap in communications compared to the flip phone days of 3G, it’s also going to be more memory hungry.
Today’s cellphone networks aren’t your dad’s cellphone networks. In fact, 5G not only represents a vast leap in communications compared to the flip phone days of 3G, it’s also going to be more memory hungry.
It makes sense when you think about how much computing power people are carrying around in their hands compared to even the early days of the Blackberry. Mobile networks are just as much about transmitting 4K video as they are talk and text. Connected devices not only include smartphones, but sensors, parking meters, smart cars, wearables, and utilities. Telecom infrastructure is now networking and compute infrastructure—flash and DRAM are supplanting SRAM and TCAM, and there might be room for emerging memories, too.
A high speed TCAM (ternary content-addressable memory) can search its entire contents in a single clock cycle and is faster than RAM. It’s a mainstay of networking gear, such as high-performance routers and switches, to increase the speed of route look-up, packet classification, packet forwarding and access control list-based commands. Despite its longevity, it is “exotic” in that it’s quite expensive and there are limited suppliers, according to Jim Handy, principal at Objective Analysis, but there’s solid payback from using them. “They streamline the routers. They make them far faster with less other processing hardware,” said Handy.
Conventional SRAM-based TCAM circuits are usually implemented with 16 CMOS transistors, which limits the storage capacity of TCAMs to tens of megabytes in standard memory structures. Handy believes that DRAM is a likely replacement in some applications as a gigabyte of commodity DRAM costs about 1/100th as much as a high-speed SRAM.
The past four decades of cellular network evolution have not only seen an increase in speed, but a growing demand for onboard memory in devices, including smartphones and the equipment that keeps the data moving.
One alternative might be ReRAM; late last year, research institute Leti demonstrated RRAM-based TCAM circuits can match the performance of CMOS-based SRAM circuits for multicore neuromorphic processor applications despite the performance and reliability tradeoffs. One of the challenges for ReRAM, as well as MRAM, another “emerging” memory, is temperature, according to Handy, and neither can handle high heat, although embedded MRAM has made some inroads. “The telephone companies want both high and low temperature stuff," he said.
Handy said one barrier for telecom networking that has disappeared is the need for longevity Forty years ago, telcos wanted a 30-year guaranteed lifetime for everything, but they're not so worried about that anymore. Five or 10 years is fine for them now. This is partly a reflection of how quickly networks are evolving. Handy noted that 4G has been around less than a decade and now it’s getting phased out by 5G.
The latest generation of mobile networks is just as much a technology platform as it is broadband communications infrastructure, according to Craig McGowan, Micron Technology’s senior customer program manager. Telecom has evolved from the initial need for mobility to making the requirement move from analog to digital. “Until the recent past, the focus really has been on consumer broadband — at least from a 3G/4G perspective,” said McGowan. He believes this was a cellphone perspective, whereas 5G is seen more holistically as a platform people can build on. There is still the consumer broadband aspect, but also the Internet of Things (IoT) aspect.
The low latency, mission critical aspect is the other biggest change, said McGowan, as 5G will enable the ultra-low latency, ultra-fast response time needed to support online gaming, artificial reality (AR) and virtual reality (VR), autonomous vehicles, and more exotic potential use cases such as remote surgery. This means we are moving to a more distributed model of computing, including the memory.
Because the compute needs are spread across the spectrum of core to edge, it starts to drive heterogeneity in terms of the requirements of the memory attached to the various compute elements along the way, according to Ryan Baxter, senior director for data center at Micron. Product characteristics and silicon matter, but so does the environment in which they will be used — standard operating temperature ranges become more important because of where compute is going to be done in the future.
While DDR (double date rate) and high-bandwidth memory will be used to keep packets moving, there’s a growing need for onboard storage throughout the network infrastructure, said Virtium vice president Hiep Pham, who oversees the company’s networking endeavors. For the immediate future, the transition from 4G means continuing to support legacy equipment. “The goal for 5G is basically to have a faster speed, a high capacity, and low latency,” he said. “But 5G will go together with some applications.”
When 5G blends with 4G, communications among the various transport points, towers, and other network links increase the performance, durability, and temperature-tolerance demands on solid-state storage, memory, and other crucial elements.
This is where the lines get blurry — the evolution from a telecom network to a technology platform means it’s not just about a better voice and data experience for handsets, but support for a wide range of edge computing and IoT scenarios, including industrial automation, as well as autonomous vehicles and smart cities. Pham believes the amount of artificial intelligence (AI) in the network will increase, starting in data centers but also distributed through the network, right to the edge. “The edge will have to handle some of the AI, even though it goes through 4G or 5G.” He said there will remain bandwidth limitations, so the edge must be capable of storing and process information.
The storage-class memory out at the edge will have to be low power, high endurance, and inherently secure. This poses a problem for 3D NAND, according to Pham, because it doesn’t have the longevity to sit out at the edge for a long period of time, so that means a lot of data will still need to be shipped back to a central location. And although 5G is fundamentally a wireless technology, it’s one that must support a wide range of communications relative to its predecessors — mobile, video, machine-to-machine and even machine-to-data center communication.
Baxter said the distributed nature of a 5G platform means there’s been a proliferation of different memory technologies being considered, and not just standard DDR3 and DDR4. LPDDR and GDDR-based technology are being looked at and the requirements of such broad spectrum are driving a fair amount of fragmentation because 5G presents itself as a blend of telecom and computing networking. “We're seeing significantly larger memory footprints required to essentially support the compute elements of the 5G deployments,” said Baxter.