UltraSoC’s technology will allow SMI to gain a more detailed understanding of the behaviour of the hardware and software within the company’s products, which are targeted at a diverse range of applications such as security, visual cognition, language comprehension and web-scale personalisation.
SMI’s solutions employ a patented processor architecture which is designed to be fully customisable to achieve maximum utilisation of on-chip resources, making it suitable for a vast range of applications from artificial intelligence (AI) at the edge to web-scale and high-performance computing. UltraSoC’s embedded analytics will be designed into SMI’s system-on-chip (SoC) semiconductor products, allowing monitoring of internal bus transactions, processor execution and other system-wide behaviours within the device.
Gajinder Paneusar, CTO of UltraSoC, said, “SimpleMachines’ Composable Computing concept is a genuine game-changer for emerging data-driven applications. The problems they’re addressing are traditionally seen as needing a hard-wired solution – and often that means costly and time-consuming ASIC development. Clever new hardware architectures also often overlook how the software developer will use, debug, or understand that architecture. We’re delighted to have the opportunity to work with SMI and to support the team as they deliver solutions to arguably some of the world’s trickiest computing problems.”
SimpleMachines CEO, Karu Sankaralingam, added: “UltraSoC is the only company that can provide intimate visibility of the operation of our chips – which is vital as we implement our architecture.”
SMI addresses a pressing need in the market, in that hardware (chip)-only solutions are not keeping pace with rapid changes in technology; and are expensive and take too long to develop and deploy.
The company’s breakthrough design simplifies complex workloads into four basic behaviours that are then composed and compiled onto the silicon on the fly to create a processor that is future-proofed against the growing needs of artificial intelligence, virtual reality, machine learning, robotics, and big data.
The Composable Computing Platform is a radically new software-centric approach that enables the programmer to easily optimise the hardware and get the performance of custom silicon with a platform that supports hundreds of different use cases.