EdgeQ reveals more details behind its next-gen 5G/AI chip – ProWellTech
5G is the current revolution in wireless technology, and every chipmaker old and new is trying to get into this highly competitive but hugely lucrative market. One of the most interesting new players in the room is EdgeQ, a startup with a strong tech pedigree via Qualcomm that we reported on last year after raising a Series A valued at nearly $ 40 million.
The company was pretty stealthy about its technology when it was working on its design (their website literally says, “Welcome to WordPress. This is your first post. Edit or delete it, then start writing!”) But today the company announced more details for the first time (and presumably updated its website as well).
The most interesting facet of System-on-a-Chip (SoC) design is that it is based on RISC-V. In contrast to processor architectures such as x86 and Arm, RISC-V is open source and one of the first open architectures to achieve lasting popularity and an enduring ecosystem. That has led to a number of new companies building on it, including now EdgeQ and also SiFive, which we covered late last year.
Vinay Ravuri, Founder and CEO of EdgeQ, stated that by using RISC-V chips, EdgeQ can offer the flexibility of reprogrammable processors called FPGAs while offering a more cohesive and integrated product with better energy savings. In his view, it was one of the major challenges in the wireless communications market with the introduction of 5G.
“When you’re in a closed system, it compresses well and everything fits together,” he said, referring to market leaders like Huawei and Ericsson, whose vertically integrated base stations are widely used around the world. The problem is, customers feel choked when all of their equipment comes from one irreplaceable vendor. With a purely open system based on standards such as OpenRAN, you get a “clunky solution” that is cobbled together from off-the-shelf parts. This leads to an increased power consumption, as the components in these boxes were never designed to be used together.
According to Ravuri, EdgeQ sits midway between open and closed and offers an expandable system that is also built in and in some cases can save up to 50% on the power needed for a wireless base station. The key is to combine machine learning over a better SoC with wireless communication and let all the pieces work together seamlessly. “The uniqueness of the communication chips lies in the algorithms,” he said. “You don’t sell sand, you don’t connect gates and say this is a processor. They connect gates and here is an algorithm for the physical communication layer. “
Adil Kidwai, Vice President and Product Manager at EdgeQ, said, “Under the hood, hardware instructions are controlled by software. It’s a ‘soft’ modem with very low power consumption. ”Since EdgeQ is based on RISC-V, the toolchain available in this ecosystem also applies to the company’s product, so engineers can use compilers and debuggers that are specific to RISC-V were developed. Ravuri found that EdgeQ added around 50 to 100 of its own vector extensions to the base RISC-V implementation to optimize performance.
With the product’s design more firmly established, he said the company would like to test its system with customers in the first half of this year. “Once we get the sample, there is a production cycle for customers to run,” he said, and intends to start growing sales by 2022. The EdgeQ base station is compatible with OpenRAN options 7.x and option 6, according to the company.
The company announced today for the first time that Paul Jacobs, former Qualcomm CEO, and Matt Grob, former CTO of the company, have both officially joined EdgeQ’s advisory board. The two of them met Ravuri when he was at Qualcomm and have been in touch throughout EdgeQ’s development.