Imagination launches IMG Labs to create new technologies
Imagination Technologies has announced IMG Labs, a specialist division tasked with developing innovations fundamental to new, advanced semiconductor products. IMG Labs’ mission is to understand and accelerate future trends in the semiconductor industry, translating these into new licensable technologies that will enable world-leading products for Imagination’s partners.
Headed up by Tim Mamtora, IMG Labs’ initial focus is on AI, GPU, heterogeneous compute, and ray tracing.
Tim Mamtora, Chief of Innovation, Imagination Technologies, said: “IMG Labs is Imagination doubling down on what it does best, core R&D for new technology that our customers want and need. The creation of Labs further demonstrates our long-standing commitment to developing commercial IP products that offer unique differentiation for customers. It enables our partners to be first to market with cutting-edge solutions and the exceptional quality that they have come to expect.”
Mamtora joins Imagination’s executive management team having spent over 14 years at Broadcom creating and managing world-leading technology and has a decade of experience architecting and designing graphics processors. Mamtora is responsible for IMG Labs and directing innovation efforts across the company. He holds an MBA from Cambridge Judge Business School and an MEng degree from the University of Cambridge.
Mamtora continued: “Heterogeneous compute is a key element of Imagination’s strategy going forward and a strong focus for IMG Labs. Increasingly our customers want to accelerate a variety of workloads with flexible arrays of processors, all the while operating within challenging power and bandwidth budgets. With a rich portfolio of processor IP, and highly-skilled research teams, Imagination is in a unique position to bring novel solutions to market and deliver clear system differentiation.”
Senior members of the IMG Labs team include serial inventor Simon Fenney, Research Fellow, author or co-author of over 50 patents in the field of rendering hardware, texture and image compression, and mathematical operations, alongside industry heavyweights Clifford Gibson, VP GPU Architecture; Steve Mackenzie, VP Systems and SoC Architecture; and Alex Pim, VP of AI Research and Architecture who leads on applied AI and high- performance computing.
Mamtora concluded: “IMG Labs is comprised of some of the brightest minds at Imagination and continues to grow its pool of world-class engineers across a spectrum of disciplines from graphics and multimedia to AI and neural network acceleration.”
Jon Peddie, President, Jon Peddie Research, added: “Imagination has a pedigree of delivering world-first, best-in-class technologies that have inspired the semiconductor and silicon engineering sectors. These advancements have been at the cornerstone of millions of critically important systems across myriad disciplines with thousands of partners. Those partners have produced revolutionary 3D GPU technology, and multi-award-winning neural network acceleration. The company has also been a leader in advanced automotive GPU IP. IMG Labs will enable Imagination to carry on with its pipeline of innovation while continuing to help their customers to differentiate.”
Deeper into IMG Labs
A key part of IMG Labs is Strategic Technology Research, a cross-functional horizon-scanning initiative to anticipate trends and identify disruptive innovations for more detailed investigation. Second-level research groups contextualise these trends, establishing how to harness the value of new technologies to meet future customer needs.
- AI: As the next wave of AI takes hold, research and architectural staff within the AI teams will analyse industry verticals, collate customer requirements, and anticipate high-performance compute trends – all in preparation for the conception and implementation of new and innovative solutions that give Imagination customers that competitive edge.
- GPU: The demand for high frame rate, high-fidelity graphics continues to grow across a variety of segments. IMG Labs stands ready to stretch what is possible through novel compression techniques, industry-leading functional safety and exceptional power and bandwidth efficiency.
- Heterogeneous compute: Modern systems are comprised of numerous dedicated processors and are required to support a plethora of workloads. By helping customers take advantage of the compute power at their disposal while retaining the flexibility to meet ever-greater demands, IMG Labs ensures every ounce of performance can be extracted from its network of IP cores, with optimal efficiency.
- Neural network acceleration: With 18 patents filed in relation to the IMG Series4 NNA, IMG Labs is set to be a novel solutions powerhouse, pushing the IP envelope with its patents, white papers, architectural proposals and regular blog post updates for the machine learning community. Whether it’s ADAS for the automotive industry, natural language processing and speech synthesis for data centres, or low-power, low-area, high-performance inferencing at the edge, IMG Labs will drive patented solutions directly into the engineering group, allowing them to deliver unique, high-quality, and award-winning products.
- Ray tracing: With more than 15 years of experience and over 250 patents filed, Imagination is the ideal partner to commercialise this ground-breaking technology. IMG Labs combines this pedigree with world-class AI expertise to bring unparalleled solutions to customers in both mobile and desktop segments. Imagination’s Ray Tracing Levels System white paper continues its thought-leadership position, educating the industry on the varying capabilities of the ray tracing offerings available in the market today.
You can read how the Imagination Tensor Tiling algorithm is empowering automotive system architects to achieve low-power, scalable and distributed compute for ADAS and autonomous vehicles with its Series4 NNA. Discover how Reduced Operation Set Computing (ROSC) can help AI and compute solution providers stay ahead of the divergence in machine learning technologies and a continually evolving software landscape.