In the past 50 years, the increase of computational power was often achieved by increasing the speed of a single processor. Today it is quite clear that this strategy is not practical anymore and multicore/multiprocessor platforms are becoming increasingly dominant.
Unfortunately, most of the results derived for single processors cannot be easily extended to multiprocessors for a number of reasons. First, schedulability analysis is far more complex than the uniprocessor case and depends on the way computational activities are allocated to the various processors. Second, interprocessor communication has a strong impact on task response times and resource access protocols, and hence must be carefully estimated. Third, there are many possibilities to split the work over multiple machines, and it is not clear which approach is more suitable. Finally, considering that algorithms intrinsically represent a sequential execution of code on a single machine, programming multiprocessor platforms exploiting parallelism is not trivial and prone to many subtle faults.
The workshop aims at stimulating the discussion on the mentioned topics by proposing selected presentation and dedicating a significant amount of time to discussion on open problems and fruitful research directions.