OpenGL ES 3.1 goes official at GDC 2014

PowerVR Rogue GPUs already running demos of the latest mobile API at GDC 2014

Almost two years after revealing their last major revision to the mobile-oriented OpenGL® ES graphics API, the Khronos Group is ready to unveil a brand new upgrade to the specification aimed at bringing a selection of desktop-level yet power efficient features to mobile devices.

The new version is OpenGL ES 3.1 (formerly known as OpenGL ES Next) and it provides essential functionality for mobile applications aimed at enhancing the feature level, reducing CPU overhead and simplifying interactions between graphics and compute tasks. New features in OpenGL ES 3.1 include compute shaders with atomics and image load/store capability, separate shader objects, indirect draw commands, enhanced texturing functionality (texture gather, multisample textures, stencil textures etc.) and enhanced shading language functionality.

OpenGL-ES_1500Until now, shading operations were mostly graphics-oriented (vertex and fragment); generic computation in mobile could only be performed by using OpenCL™ or Renderscript. However, the use of two separate APIs (one for graphics and one for compute) was not always the optimal solution for certain applications.

Since the developer ecosystem was asking for a viable alternative to access compute shaders, Khronos decided to include this feature in OpenGL ES 3.1. The basic principles of OpenCL compute shaders are the same in OpenGL ES 3.1: the compute pipeline is a single-stage machine that runs generic shaders; compute workloads are formed from groups of workitems called workgroups and processed by the executable code for a compute program. A workgroup is a collection of shader invocations that can execute the same code in parallel.

What does OpenGL ES 3.1 mean for chipset vendors and OEMs?

When OpenGL ES 3.1 was first previewed at CES 2014, we were one of the first members of the Khronos Group to announce support for the new API.

Perhaps the most interesting aspect of the announcement above is that all PowerVR Rogue GPUs are able to deliver the full feature set of OpenGL ES 3.1, therefore future-proofing a whole generation of existing PowerVR Series6-based chipsets designed for smartphones, tablets, smart TVs or smart cars. This includes application processors from Allwinner (UltraOcta A80 – PowerVR G6230), LG (H13 – PowerVR G6200), Renesas (R-Car H2 – PowerVR G6400), Intel (Atom Z3460/Z3480 – PowerVR G6400, Atom Z3560/Z3580 – PowerVR G6430), MediaTek (MT8135 – PowerVR G6200, MT6595 – PowerVR G6200) and many others.

PowerVR Series6 GPUs have already achieved conformance for OpenGL ES 3.0 and OpenCL 1.2.

PowerVR Rogue GPUs - OpenGL ES 3.1

The other important aspect is related to PowerVR Series6XE, our family of GPUs aimed at next-generation, highly affordable devices. Since PowerVR Series6XE GPUs have been designed to fully support OpenGL ES 3.1, silicon vendors can be certain that even their PowerVR Rogue-based, entry-level SoCs will be capable of delivering the best possible performance per mm2, with no compromises in API feature set.

What does OpenGL ES 3.1 mean for developers?

Developers who want to quickly move to OpenGL ES 3.1 can rely on Imagination and our silicon partners to deliver an optimal combination of high-performance, low power graphics IP; a reliable, robust driver stack; and a comprehensive, fully-featured SDK.

The great news out of GDC 2014 is that we have already updated some of our tools so developers can try out OpenGL ES 3.1. For example, PVRVFrame will allow OpenGL ES 3.1 applications to run on desktop development machines that don’t natively support the OpenGL ES APIs.

Expect a new version of the PowerVR Graphics SDK and its utilities to become available in the very near future.

What does OpenGL ES 3.1 mean for consumers?

Simply put, consumers that own PowerVR Rogue-based devices can rest assured that next-generation applications will be fully supported by our GPUs. OpenGL ES 3.1 introduces a number of features that bring the mobile graphics API one step closer to console-level performance while still focusing on optimal use of available hardware features. Mobile users will be able to enjoy more feature-rich applications that combine graphics and compute functionality.

A recent example highlighting the superior performance efficiency of our PowerVR GPUs has been the latest announcement from Futuremark. 3DMark Cloud Gate, their latest OpenGL ES 3.0 graphics benchmark was demonstrated at MWC 2014 exclusively on PowerVR Rogue-based devices.

You can find out more information about OpenGL ES 3.1 by accessing the Khronos website and visiting their booth at GDC 2014 and other future events.

Make sure you follow us on Twitter (@ImaginationPR, @PowerVRInsider, @GPUCompute) for the latest announcements from Imagination and our ecosystem.

Editor’s Note

OpenCL and the OpenCL logo are trademarks of Apple Inc. used by permission by Khronos.

OpenGL is a registered trademark and the OpenGL ES logo is a trademark of Silicon Graphics Inc. used by permission by Khronos.

 

, , , , , , , , , , , , ,

  • Pingback: Khronos Group anuncia su API gráfico estándar OpenGL ES 3.1 | SINETEC()

  • TimBob

    Hi Alex, thank you for the write up on OPGL ES 3.1, was looking forward to hearing more after the initial blog entry about OPGL Next.

    Just a quick query regarding the Rogue IP graphics road map illustration. Is it just me or does the illustration appear to show that the GX6240 is closer to series 7 and therefore more advanced than the rest of the GX line up? I do know based on what I have read here that in reality the GX6240 is ‘area’ optimized and is therefore not as powerful as the other GX IP’s, such as the GX6250 and GX6450 etc. I am sure I’m simply reading too much into the positioning of the IP’s in relation to series 7 on that illustration, if this is the case please ignore.

  • Pingback: Khronos Group anuncia su API gráfico estándar OpenGL ES 3.1 | | Blog Desktop()

  • http://withimagination.imgtec.com/index.php/author/alexvoica Alexandru Voica

    I was limited in how I could represent the cores on the roadmap (we now have two extra families). I meant to show that GX6240 will be coming a bit later compared to the other three.