High-end graphics on mobile: PowerVR Series5XT and Series6 GPUs take on OpenGL ES 2.0 and 3.0 (part 2)
If the first segment of my two-part article on the history of PowerVR GPUs and OpenGL ES focused on the two families that set the game rules for mobile graphics (PowerVR Series4 and Series5), this second installment gives some insight into PowerVR Series5XT and Series6, the two most recent GPU families that have re-shaped expectations for performance efficiency.
PowerVR Series5XT: high efficiency and low power with the 2nd gen USSE shader engine
The PowerVR Series5XT architecture builds on the underlying strengths of Series5, delivering significant performance advantages through increased floating point and instruction throughput as well as natively supported YUV and color space conversion.
PowerVR Series5XT GPUs introduces support for PVRTC2, a major upgrade to our PVRTC texture compression technology. The addition of sub-texturing, together with significant improvements in quality and NPOT support all further strengthen PVRTC2 as the texture compression format of choice for cross-platform, cross-OS textures in leading mobile and embedded platforms.
This architecture also supports advanced multi-core scalability by enabling multi-processor implementations, with up to 16 core instantiations possible that deliver better than 95% linear performance scaling of both geometry and pixel processing resources.
The PowerVR Series5XT family comes in three distinct implementations:
- PowerVR SGX543, a four-pipeline GPU delivering significant enhancements to the Series5 SGX architecture. PowerVR SGX543’s wide-ranging architectural enhancements include a second generation USSE2 (Universal Scalable Shader Engine) which provides extended USSE instruction set with comprehensive vector operations and co-issue capability. This leads to a typical increase of 40% in performance for ‘shader-heavy’ applications. The PowerVR SGX543 delivers a twofold increase in floating point and hidden surface removal performance compared to previous Series5 cores.
- PowerVR SGX544, a GPU core providing comprehensive market-proven support for full OpenGL ES 1.1/2.0, DirectX 9_3 and OpenCL 1.1 EP. The feature set delivered by PowerVR SGX544 GPUs goes well beyond OpenGL ES 2.0; many new features introduced by OpenGL ES 3.0 can be supported in hardware, including advanced effects such as Multiple Render Targets (MRTs), instancing and occlusion queries.
- PowerVR SGX554, featuring eight pipelines for a single core configuration. The PowerVR SGX554 GPU provides full support for a wide set of graphics and compute APIs (OpenGL ES 1.1/2.0, OpenGL 2.1, DirectX 9_3, OpenCL 1.1 EP, Renderscript/Filterscript, etc.) with maximum hardware acceleration, making it ideal for tablets, computing devices and smartphones.
PowerVR Series6 ‘Rogue’: Maximum performance efficiency with the latest generation USC shader cluster array
The latest implementation of our advanced TBDR architecture is the PowerVR Series6 ‘Rogue’ family of graphics IP cores. PowerVR Series6 GPUs include a scalable number of graphics and compute clusters designed to target the requirements of a growing range of dynamic markets. Using these arrays of shader clusters, PowerVR Series6 cores deliver class-leading graphics and GPU compute efficiency while minimising power and bandwidth demands.
The PowerVR Series6 ‘Rogue’ family brings a number of unique features designed to improve overall system performance, lower power consumption and reduce memory bandwidth:
- Scalar processing to achieve highest utilization of ALUs and ease graphics programming
- Triple compression: a combination of lossy (PVRTC) and lossless (geometry and image) compression techniques to reduce memory bandwidth
- PowerGearing™ to manage different power consumption levels and balance overall SoC power consumption
PowerVR Series6 is the first GPU IP family to have achieved full conformance with the latest version of OpenGL ES. We have worked closely with Khronos to make sure our PowerVR Series6 IP cores and drivers are optimized to efficiently deliver full OpenGL ES 3.0 support across all our ‘Rogue’ family.
This enables our partners to scale their innovative SoC solutions both vertically and horizontally across different markets, all the while ensuring they get the highest performance at the lowest power consumption and bandwidth usage.
Here is a list of the PowerVR Series6 cores announced so far:
- PowerVR G6200 and G6400 are designed to deliver the best performance at the smallest area possible for two and four cluster architectures respectively.
- PowerVR G6230 and G6430 ‘go all out’ in terms of performance by adding incremental extra area for a complete feature set and maximum performance whilst minimising power consumption
- PowerVR G6630 features six shader execution clusters. It is designed to deliver the best possible performance for the widest range of application use cases; the advanced PowerGearing power scaling technology enables USC cluster pairs to be shut down depending on the app load.
- PowerVR G6100 is the latest addition to the highly efficient Series6 family of graphics IP cores, making it the smallest member of the ‘Rogue’ family to date. A single core, single cluster GPU, it targets mass-market adoption of OpenGL ES 3.0 on a wide range of computing platforms.
OpenGL ES 3.0: the next big thing in graphics APIs is here
Khronos APIs are fundamental to creating future generations of converged application platforms. Open graphics and compute APIs enable apps to travel seamlessly from one platform to another, with little or no modification required.
The combination of all Khronos APIs as a key part of a total platform solution is vital for Imagination’s future. Over the past ten years, we have focused on delivering complete, heterogeneous GPU IP solutions that enable both graphics and compute on the same core, therefore offering our partner ecosystem the best of both worlds, with no sacrifices in performance or power.
With OpenGL ES 3.0, Khronos has put a lot of effort into improving the performance efficiency of the standard. The new enhancements brought to the rendering pipeline enable developers to create advanced visual effects, bringing applications ever closer to the image quality seen in PC and console graphics. Features such as occlusion queries, transform feedback, instanced rendering and support for four or more rendering targets make better use of the graphics and compute resources of our impressive PowerVR ‘Rogue’ architecture, keeping processing localized in the GPU, and thus saving memory bandwidth, lowering CPU usage and ultimately achieving lower power consumption.
PowerVR GPUs are at the heart of today’s iconic devices, from flagship smartphones (Samsung Galaxy S4, iPhone 5, Lenovo K900, Sony Xperia C S39h) and popular tablets (Amazon Kindle Fire HD, iPad 4, Samsung Galaxy Tab 3), to exciting new wearables (Google Glass) or high-resolution smart TVs (LG CINEMA3D smart TV). Furthermore, PowerVR-based affordable devices from Chinese OEMs are now shipping in millions of units, making our GPU IP the most widely adopted OpenGL ES solution on the market.
The infographic below shows a summary of the various OpenGL ES versions released within the last decade, our PowerVR GPU-related announcements and some of the most iconic products that integrated the latest generation GPU IP from Imagination.
Thanks to our close involvement throughout the development of the OpenGL ES 3.0 standard, all our PowerVR Series6 ‘Rogue’ GPU cores have been designed to fully support all features of OpenGL ES 3.0. PowerVR graphics IP cores continue to be selected by industry leaders such as AllWinner, LG, MediaTek, Renesas, and many others to power their next-generation, OpenGL ES 3.0-capable application processors, making way for the next revolution in mobile computing.
Do you find the new OpenGL ES 2.0 extensions for PowerVR Series5XT useful? Are you eager to get your hands on the first smartphone or tablet with a PowerVR Series6 GPU? Leave us your comments in the box below. Make sure you keep up with the latest developments on PowerVR by following us on Twitter (@GPUCompute, @PowerVRInsider and @ImaginationTech) and subscribing to our blog feed.