Benchmarks are used to quantify the ability of computers to solve problems effectively. Widely used benchmarks enable comparison between different computers, yet there are also cases where performance on many benchmarks is needed to assess performance of a particular application without running that particular application. Community agreement on a good choice of candidate benchmarks to fully evaluate parallel and distributed systems is lacking. 

Many algorithms have been proposed and appear in the state of the art of many application domains such as image processing, cryptography, pattern recognition, signal processing, document analysis, etc… In order to quantify the performance and the benefits of a new algorithms, it is necessary to benchmark them and compare them to older algorithms.  The computer architecture on which the benchmark is run is also important since it may favor some kinds of algorithms over other algorithms. 

In most cases, the benchmark problem size is really important. One wants to understand what computational resources will effectively solve the problem quickly and correctly for different problem sizes. 

This special session focuses on research works aimed at benchmarking parallel and distributed systems for solving real world problems. Furthermore, benchmarking related contributions concerning the definition of new open platforms, studies on the distribution of computation tasks and applications in different fields in computer science or from other domains are sought.