Skip to main content
The Khronos Group, arbiters of the OpenGL 3D specification and a range of related specifications, has announced version 3.1 of OpenGL ES, its 3D API oriented toward mobile systems.

The headline feature of the new spec is support for compute shaders. OpenGL and OpenGL ES are built around a graphics pipeline that feeds in geometry (lines and points) and textures at one end while producing pixels at the other. Compute shaders break out of that graphics orientation by providing a model for GPU-based computation that's not tied to any specific part of that pipeline.

Compute shaders can still be used graphically—manipulating or producing geometry, for example—but they can also be used for non-graphical things, such as physics computations.

ES 3.1 also adds support for indirect drawing, which enables the GPU to draw objects taken from a buffer in GPU memory rather than CPU memory. This is important especially in conjunction with compute shaders, as those shaders can be used to create objects within GPU memory. Indirect drawing is more efficient, as it can reduce the amount of CPU involvement.

The new specification makes shaders more flexible by allowing different kinds of shader to be programmed separately. It also adds new texturing options and makes the shader language more powerful, with new bitfield and arithmetic operations.

OpenGL ES is a subset of the full OpenGL specification used on the desktop. The previous version, OpenGL ES 3.0, largely took features from OpenGL 3.x. This new update uses OpenGL 4.4 to provide its menu of potential features, with selection driven by developer demand and suitability to mobile usage.

Khronos plans to make further announcements later this week at the GDC gaming conference. It's leading off with the GL ES 3.1 launch because many mobile vendors at GDC want to talk about their plans for OpenGL ES 3.1. The new features may or may not need new hardware. While some mobile GPUs, such as NVIDIA's Tegra K1, will support 3.1 with nothing more than driver updates, other GPU families might need new chips.

Source
Be the first to reply!

Reply