| Abstract: |
|
OpenMP is twenty this year. It continues to be one of the most important systems used to exploit parallelism inside shared memory, with many users and active support for the development of the standard and its implementations from multiple hardware vendors, open-source projects and academic researchers.
This BoF will describe the evolution of OpenMP, including its support for accelerators, and present the latest developments in OpenMP which are important to users of modern hardware.
The BoF will consist of a few, short, presentations (less than five minutes each) on key new OpenMP features, followed by a question and answer session in which an expert set of panelists, including many who are involved in OpenMP standardization, will engage in discussion with the audience and answer their questions about anything OpenMP related.
If you use OpenMP, or are wondering whether you should, this is your chance to find out more about its current state, future direction, and discover that there it has more features than "!$omp parallel for"!
Targeted Audience Users of OpenMP who are interested in modern OpenMP and the future development of the OpenMP standard. MPI programmers with single threaded codes who realise that hardware trends make this a non-scalable approach and that they therefore need to exploit in-node parallelism explicitly. Anyone who is interested in OpenMP. |
|
;){kind=link}