All the winners of Ateji Parallelism Award receive a bottle of Champagne. You too can be the winner of next Award. Experiment with Ateji PX: explain why your samples are important, and show how Ateji PX makes them different.

Maxence Schmitt and Romain Guidoux, ISIMA

Maxence Schmitt and Romain Guidoux compared algorithms written in Java and Ateji PX, and showed that it is very easy to convert sequential Java code into parallel Ateji PX code. They achieved pretty good performance, obtaining a speed-up of 7 on an 8-core Intel Nehalem with the Game of Life example.

Maxence and Romain are students at ISIMA (master-level engineering school) and were supervised by Professor David Hill, expert in High Performance Computing.

Andrea Vandin, University of Pisa

As part of a project at the University of Pisa with Professor Fabio Gadducci, Andrea Vandin implemented and benchmarked different parallel versions of the convolution algorithm, a fundamental tool for most image processing transformations. He demonstrated that the message passing version is more efficient than the shared memory version (both written with Ateji PX).

Philippe Coucaud

In reaction to a question we mentioned in an newsletter, Philippe Coucaud provided insightful comments regarding syntax design for the select statement, using arguments comparing Ateji PX's select statement and Java's switch.

Jonathan Derque, R&D engineer at Exalead

Jonathan Derque demonstrated how Ateji PX makes it quick and simple to develop a parallel application based on the Actor Model. His benchmarks show that Ateji PX is faster than Scala and on-par with the Jetlang library in terms of run-time performance.