The latest version of Parallware Trainer is now available. This release (0.4) has new and interesting features such as:
- Support for OpenACC “data” directive for offloading to GPU.
- User can now select two implementations of parallel scalar reductions:
- Built-in reductions (e.g. OpenMP/OpenACC reduction clause)
- Atomic access (e.g. OpenMP/OpenACC atomic construct)
- Generation of pragma templates where users should specify array ranges
- Console messages now guide users to fill in array access ranges
- Improved user messages
- Bugfix in compound operators (e.g. +=, *=)
- Bugfixes in GUI
In order to try Parallware Trainer, you must register for the Early Access Program. Users will have full and free access to the tool for a month and will be able to learn parallel programming while improving their code. We look forward to receiving feedback from the participants on the usability and parallelization capabilities of Parallware Trainer.
New versions, currently under development are regularly made available for all participants.