Scalable PowerVR Series7 GPUs target applications from wearables to servers

November 11, 2014 // By Graham Prophet
Silicon IP provider Imagination Technologies has announced the PowerVR Series7, the latest generation of the PowerVR Rogue GPU architecture and an entire line-up of GPUs which scale from 16 to 512 arithmetic logic unit (ALU) cores, providing scalability, efficiency and performance, ultimately reaching teraflop performance.

Imagination says that the PowerVR Series7 is architected for the broadest range of next-generation products from wearables and IoT through automotive, mobile and consumer electronics devices, up to high-performance GPU compute. Highly configurable, scalable GPUs address market requirements across these segments, with new features such as full hardware virtualisation and Android Extension Pack (AEP) support with hardware tessellation, as well as features such as geometry shaders, PVR3C triple compression technology for bandwidth/power savings, and full ASTC (adaptive scalable texture compression) support. Series7 also includes Imagination's advanced PowerGearing technology for dynamic core and cluster level power management.

Tony King-Smith, EVP marketing, Imagination, says: “With graphics-enabled devices entering all aspects of everyday life, it is no longer possible to offer just point solutions. Each new generation requires a wavefront of GPUs from low to high end, to address the broad spectrum of applications. With PowerVR Series7, we have not only honed our architecture but also our IP support with tools such as our Design Optimization Kits for rapid physical and performance optimization. These allow our partners to choose right-sized solutions based on the optimum balance of power, performance and area for every segment.”

The PowerVR Rogue architecture is scalable via threads, cores and clusters. Rogue uses clusters of up to 32 multi-threaded ALU cores, working in close cooperation with dedicated texture units on the same region of the screen. This approach ensures that all accesses to memory are correlated and efficient, unlike older multi-core approaches where many cores all work in an uncoordinated way on different regions of the screen which results in many distributed accesses to memory and inefficient DRAM usage. With their highly scalable and efficient architecture, PowerVR Series7 GPUs provide unrivalled performance and efficiency.

Series7XT and Series7XE GPUs achieve up to a 60% architectural performance increase on the latest industry standard benchmarks compared to equivalent configurations of previous generation PowerVR Series6XT/6XE GPUs, claiming to be most efficient, highest performance,