Newsletter

Newsletter
 

Mobile GPU Computing for Accelerated Apps

 

This software is not yet available as a product. Mobile GPU consulting services are currently available. To get started with mobile GPU computing, email us at sales@accelereyes.com and include a description of your application.



Overview

AccelerEyes is building a software acceleration library for mobile devices. It contains commonly used compute functions. It delivers a 5 to 30X speed boost for convolutions and other common compute functions. It enables:

  • Realtime video processing
  • Better augmented reality
  • Better computational photography
  • Faster data processing

The AccelerEyes mobile GPU computing software delivers performance through its unique GPU computing capability. All mobile devices have a GPU which can deliver greater compute throughput than the CPU. Why not use the GPU to make your app faster?


Details

The AccelerEyes mobile library is a C/C++ library with a simple matrix API. The same code can be deployed on all platforms without any further code rewrites. For software and documentation, register below.


Supported Platforms

Android Logo
Android
Apple Logo
iOS

Supported GPUs

The mobile library supports the OpenGL ES 2.0 and OpenCL specs, compatible with all popular GPUs, including the following:

tegra logo
adreno logo
mali logo
powervr logo

Example - Tennis Volley Game

This tennis game demonstrates controller free gaming with mobile devices, made possible using GPU acceleration. The motion detection is done via Horn-Schunck optical flow.

Download the "Tennis Volley" game on the iTunes App Store or the Android Market.

iPad2 using PowerVR GPU

Motorola Xoom using Tegra2 GPU


Example - Traffic Video Processing

Simple demonstration of performance boost in a video processing application, computing the gradient of the image frames. Filmed in midtown Atlanta.

Google Nexus S: CPU at 1.3 FPS, GPU at 35 FPS

Motorola Atrix: CPU at 2.9 FPS, GPU at 29 FPS


Example - Tracking with Chan/Vese Active Contours

Simple demonstration of a real-time tracking application, using the popular Chan/Vese active contours. Filmed in midtown Atlanta.

Google Nexus S

Motorola Atrix