Chips that can support two of the most demanding and high-profile processor use cases – 5G and advanced AI – at the same time have been among the most-hyped technologies of the past year. The company that has achieved the highest recognition among a host of start-ups targeting this dream is EdgeQ, which is now sampling its first products for 5G infrastructure, and claims to have its first design win with “a large North American OEM”, which has committed to productizing the EdgeQ chip.
There were no more details about that deal, but the company has released more information about its 5G base station-on-a-chip, which consists of a physical layer (PHY) product with software stack. It is based on an architecture that is designed to be able to support different types of 5G product – distributed unit, radio unit or integrated gNodeB – with adjustments made in firmware. In future, its programmability should be extended to support AI too, on the same physical layer.
True to the heritage of board member Paul Jacobs, former CEO of Qualcomm, one of the chief areas of focus has been to reduce power consumption. EdgeQ claims up to 50% total cost of ownership (TCO) reduction for a 5G base station, mainly because of power savings. According to Ravuri, half the TCO spend for a base station is in the power system (including cooling, uninterruptible power supply and the cost of electricity).
CEO Vinay Ravuri told EETimes: “Within two weeks we actually got full Linux running and it looks solid, and we are able to send 5G traffic through the chip.
The chip is based on a signal processor running on 50 identical cores, using the open RISC-V architecture. These are dynamically programmable to process the PHY layer functions, which are implemented in software and can be adapted according to the type of base station. The core is licensed from Andes and customized with EdgeQ’s 5G-specific instruction extensions (and others for AI workloads).
Some of the specialised operations supported by EdgeQ’s own instructions are common to both 5G and AI workloads, so the same chip could accelerate both, in parallel or, more efficiently, alternately. This commonality is the basis of many start-ups’ claims to be able to target two of the hottest applications for advanced processors with the same product, though EdgeQ has gone further than most in detailing and demonstrating its capabilities, at least to a closed circle of possible customers and partners.
It claims processor cycles between 5G transmissions can be used to perform AI tasks, effectively multiplexing the workloads together, “much like how virtual machines run,” as Ravuri put it. The customer can decide on the balance between the two types of workload.
The first focus of the dual-function products will be on systems such as AI-enabled cyberattack detection in 5G networks, and then device-side use cases will follow such as industrial robots or vehicles that require both AI processing and 5G connectivity.
In the 5G base station, as well as the processor, EdgeQ’s chip features a network-on-chip (NoC), RF interface, forward error correction (FEC) accelerator, protocol accelerator for Layers 2 and 3, and secure boot. A host processor, based on eight ARM Neoverse cores, performs control and configuration functions such as diagnostics and on-chip software upgrades.
Like many new players in the mobile chip market, EdgeQ sees Open RAN as a potential entry point. It claims that, by offering production-grade PHY software, rather than just reference software, it saves customers from developing their own implementations, which end up being proprietary. EdgeQ says its software supports key 5G functionality such as beamforming, Massive MIMO and interference cancellation out of the box, in a standard way, which will help enable multivendor networks for operators, while saving cost and time for smaller vendors to develop advanced base stations.
EdgeQ offers a silicon-as-a-service model. Customers pay a base fee for the chip and a basic 5G implementation and can then pay to enable more advanced functions such as ultra-low latency, location services, RAN sharing, slicing or machine learning, which are delivered via firmware updates.
“It became clear that 5G is so broad that not everybody wants to pay upfront for everything,” Ravuri said. “They don’t want to pay for sophisticated features up front because they may not know what they’ll use it for.”
EdgeQ now has 120 employees, mainly in San Diego, USA and Bangalore, India. It was founded in 2018 and has raised a total of $51m. In 2020, it added former Qualcomm CEO Paul Jacobs and former Qualcomm CTO Matt Grob, both co-founders of XCOM, to its advisory board.