Intensive computing with the abiquo platform

A few months ago, Santiago (one of our java architects at abiquo) developed in one morning a simple application to render scenes on a grid, using the yafray command line application as backend.

Without our platform, this image is rendered in 30 minutes (approximately). The same rendering task, using seven desktop computers and our platform, is reduced to 3 minutes, i.e. ten times faster.

In the application you can decide how to split your task; In this case, the yafray initialization state slows the process down a little. For this reason, if you decide to divide the image into more squares the execution will take longer. On the other hand, by using little squares, you can send more work to the best computer. However, this application is only a demostration :).

In order to use the intelligent agents layer, the platform learns about the service and the nodes, and the developer can apply its knowledge to split the work in dynamic squares, to send bigger squares to the best computers.

You can check here a video demostration.

Maybe it’s time to remind the data mining example using an application we gridified on the research group at the university few years ago. Now it’s working with the abiquo platform. The video here :)

Tags: abiquo platform, intensive computing, render, yafray