Accelerating Computation in HTML5

With the emergence of HTML5 the browser has become a powerful and productive platform for application developers. It is possible to develop wide range of application using html5. The JavaScript engines are further improved, by different vendors like Chrome, Mozilla etc., for faster execution of JavaScript. However, Compute-intensive functions, graph layout algorithms, and huge data processing, are still considered beyond the reach of JavaScript. This presentation looks at improving the performance of such applications implemented on the Web platform. It describes how high-performance compute capabilities of multi-core CPUs and programmable GPUs can be made accessible to web applications.

This presentation describes a real world use case related to layout algorithms used in graphics such as node-link diagram. Here the goal is to improve the computation time of graph layout algorithm.

To accelerate the computation and reduce the compute time, the layout algorithm is implemented using new technologies like OpenCL and WebCL .OpenCL is  a new standard for carrying out computation  in heterogeneous environments i.e It is possible to execute same OpenCL code on GPU ,CPU or any OpenCL compliant device.

Session Details:-
The session will cover following topics:-

  • Introduction to multicore architecture and its evolution
  • Brief description of compute resource used in the demo
    -Multicore CPU,many core GPU (Graphics processing unit),GPGPU computing (general purpose graphics processing unit)
  • Demo of HTML5 use case using parallel computations
  • Comparison between compute time  for single threaded JavaScript execution  and parallel execution    In HTML5.
  • Brief overview of new cutting edge technology like OpenCL and WebCL ,which make it possible to achieve accelerated computations in HTML5
  • Code walkthrough of the demo application

Audience  Take aways:-
After this session audience will know how to achieve compute performance in HTML5 using new technology
-Advantages GPGPU  computing
-Controlling native code execution from browser

Speaker:

Ashish is a Software Engineer in Development at SAS Research & Development, India.  He is involved in developing and maintaining various graphs that are deployed on mobile & desktops.

Enhanced by Zemanta

Comments are closed.