MediaTek and Imagination Technologies work together to create optimal heterogeneous multi-processing SoCs

SUNNYVALE, CA., Aug. 28, 2013 (GLOBE NEWSWIRE) — Imagination Technologies (LSE:IMG), a leading multimedia, processor, communications and cloud technologies company, recognizes its partner MediaTek Inc.  for its unique deployment of heterogeneous processing. Confirmed by leading industry sources, Imagination sees the company's new MT8135 SoC as the industry's first publicly announced mobile SoC enabling heterogeneous multi-processing (HMP) utilizing asymmetric processor configurations. This innovative SoC features Imagination's latest generation PowerVR Series6 'Rogue' graphics processor (GPU).

True heterogeneous multi-processors enable dynamic configuration of the operating modes of CPUs of varying capabilities, GPUs, and other processing engines in the system, to ensure ideal task allocation, optimal performance and minimal power consumption under all real workload conditions. Imagination believes that by using true heterogeneous multi-processor technologies, the full potential of all the complex processing resources of an SoC can be realized. This can be seen in applications such as GPU compute, where Imagination has demonstrated power savings in excess of 30% with 2x performance gain when the higher performance and lower power consumption of PowerVR GPUs are used to augment the CPU compared to a CPU-only solution.

“Heterogeneous processing architectures represent the future of mobile and embedded SoCs. We have been working with MediaTek to advance their HMP technology on multiple fronts — as founder members of the Heterogeneous System Architecture (HSA) Foundation, on development of this new SoC, on future co-development and more,” says Tony King-Smith, EVP marketing at Imagination. “MediaTek's HMP implementation is a great example of how we see our partners using their unique expertise combined with our extensive GPU, CPU, and SoC experience to differentiate their products.”

MediaTek's HMP solution enables all processors in a multi-processor system to be individually configured for an optimal balance of power and performance under variable workloads. The solution is completely scalable and can be configured from any number of small and large CPU or GPU 'engines.' It is also flexible and architecturally neutral in terms of what types of processors are used and how they are connected.