Semiconductor designers weighing graphics and AI acceleration options often land on two distinct families from the same supplier. Imagination Technologies builds both PowerVR graphics IP and the newer PowerVR E-Series GPU IP, and the two lines address different balances of rendering horsepower, compute density, and power draw. This comparison looks at how each architecture fits into mobile, automotive, data centre, and edge AI designs, and where the lines diverge technically.
Understanding Imagination’s GPU IP Architecture
Imagination Technologies licenses its GPU designs as soft IP, meaning partners integrate the architecture into their own silicon rather than buying a finished chip. That approach has put the company’s GPU technology inside more than 12 billion devices, spanning cars, phones, wearables, and PCs. Every design, whether aimed at graphics or AI, is built on the same Tile Based Deferred Rendering foundation, which reduces data movement and strips out redundant operations early in the rendering pipeline.
This shared base matters because it lets licensees move between product lines without relearning an entirely new programming model. The architecture supports industry-standard APIs such as Vulkan and OpenCL, so software teams can bring existing code across different PowerVR and E-Series implementations with fewer rewrites.
PowerVR Graphics Solutions: From Mobile to Cloud Gaming
PowerVR has gone through 14 generations of development, starting from early mobile graphics and extending into cloud gaming and desktop workloads. Imagination Technologies offers several named architectures within this family, including PowerVR Rogue, PowerVR Furian, the ray-tracing-focused PowerVR Photon, and the newer IMG A-Series, each building on the TBDR foundation with its own set of improvements.
The PowerVR line also covers IMG DXD GPU, a design aimed squarely at Windows and cloud environments. It supports DirectX 11, DirectX 12, Vulkan 1.4, OpenGL 4.6, and OpenCL 3.0, which positions it for PC gaming, cloud rendering, Windows applications, and mobile ports running in cloud settings. On the ray tracing side, Imagination Technologies points to its mobile ray tracing solution as the first hardware-based implementation built for that market, and it currently sits at Level 4 on the Ray Tracing Levels Scale.
E-Series GPU IP: Next-Generation AI and Compute Acceleration
Where PowerVR spans graphics-first use cases, the PowerVR E-Series GPU IP is built with AI and compute acceleration as a core design goal rather than an add-on. It still handles rendering tasks, but its internal structure is tuned to also run inference, vision, and general compute workloads inside the same compact power envelope. That dual capability makes it a candidate for SoCs that need to do more than one job without adding a separate accelerator block.
It describes the PowerVR E-Series GPU IP as suited to multifunctional system-on-chip designs, since its parallel structure can run AI inference and graphics tasks at the same time. An open, developer-friendly compute software stack is part of the offering, which is meant to keep the programming experience familiar even as workloads shift toward AI-heavy pipelines.
Burst Processors and Power Efficiency Across Product Lines
One of the defining features introduced with E-Series is Burst Processors technology, which works alongside the graphics and AI pipelines to lower average GPU power consumption. Imagination Technologies states this cuts average power use by 35% compared with the previous D-Series generation. The mechanism behind that figure is a reduction in data movement within the GPU, which frees up memory bandwidth and keeps compute units fed more consistently.
That efficiency gain has knock-on effects for GPU utilisation as well, since less time is spent waiting on memory transfers. For power-constrained designs such as wearables or battery-powered consumer electronics, this kind of saving can matter as much as raw throughput. Imagination Technologies positions Burst Processors as a direct successor to the approach used in D-Series silicon.
Neural Cores and On-Device Intelligence in E-Series
Alongside Burst Processors, E-Series introduces Neural Cores, a denser and more deeply integrated form of acceleration for AI operations. Imagination Technologies reports up to 4x more dense acceleration compared with its D-Series GPU IP. This effectively turns what used to be a conventional GPU Core into a Neural Core capable of handling heavier AI workloads within the same silicon footprint.
These Neural Cores scale up to 200 TOPS using INT8 or FP8 precision, a level of throughput aimed at on-device features like super resolution and ambient occlusion. Because the acceleration sits inside the GPU itself rather than a separate chip, designs built around the PowerVR E-Series GPU IP can apply this compute capacity to both rendering enhancements and standalone AI inference tasks.
- PowerVR graphics IP spans 14 generations of architecture development, from PowerVR Rogue through to the ray-tracing-capable PowerVR Photon and IMG A-Series.
- IMG DXD GPU supports DirectX 11, DirectX 12, Vulkan 1.4, OpenGL 4.6, and OpenCL 3.0 for Windows and cloud gaming scenarios.
- Imagination’s mobile ray tracing technology sits at Level 4 on the Ray Tracing Levels Scale.
- E-Series Burst Processors reduce average GPU power consumption by 35% compared with D-Series.
- E-Series Neural Cores deliver up to 4x more dense AI acceleration than D-Series GPU IP.
- E-Series Neural Cores scale up to 200 TOPS using INT8 or FP8 precision for on-device AI features.
Choosing Between PowerVR and E-Series for Different Markets and Workloads
Market fit tends to drive the decision between the two families more than any single benchmark. Automotive designs that need ISO 26262-capable, safety-oriented graphics for infotainment, cockpit displays, and ADAS functions often draw on PowerVR’s long track record in that space. Mobile devices chasing richer gaming experiences or cloud-ready rendering lean on PowerVR’s ray tracing and DXD GPU capabilities as well.
For designs where AI inference, data pre-processing, or computer vision carry equal weight alongside graphics, the PowerVR E-Series GPU IP offers a combined rendering and compute path without a separate neural accelerator block. Consumer electronics with voice and gesture features, or SoCs handling language model workloads, fit naturally into that category. Imagination Technologies frames both lines as drawing on a single underlying GPU architecture, which lets SoC architects weigh rendering needs against compute density rather than choosing between unrelated technology stacks.
Looking Ahead for GPU IP Selection
Graphics and AI requirements continue shifting as silicon lifecycles stretch across years of deployment. A general-purpose, programmable GPU architecture gives chipmakers room to adapt as workloads change after launch, rather than locking designs to a fixed function block.
