THE HPC LEARNING TOOL TO HELP YOU LEARN ABOUT HPC AND GET YOUR JOB DONE!
Parallelware Trainer is an interactive, real-time code editor with features that facilitate the learning, usage, and implementation of parallel programming by understanding how and why sections of code can be parallelized.
Users are actively involved in learning parallel programming through observation, comparison, and hands-on experimentation.
Bridging the gap between training and real code development:
- Use real-world code examples (including the opportunity for trainees to use their own code), rather than the typical ‘toy examples’ used in traditional training.
- Improved productivity of trainees after the course: they start working on their own code either in the course or immediately afterwards.
- Improved targeting of training: they learn what they need to learn for their code.
- Improved retention of information given in training as it is more relevant to the problems the trainees are facing in their real work.
- Trainees can choose the material relevant to them via Parallelware Trainer, rather than listening to lectures where some of the material is not relevant.
- Reduced staff costs and/or increased class sizes as Parallelware Trainer can act as a mentor by identifying the users unique challenges and solutions for parallelizing their code.
- Identifies opportunities for parallelism in real applications
- Shows you what your options are for parallelizing a specific code region, providing a ranked list of options to try
- Aids you in adding syntactically and semantically correct OpenMP (for shared memory parallelism) and OpenACC (for accelerators) directives
- Speeds up benchmarking to identify the most efficient parallelization techniques, including combinations of techniques and variations within specific directives
- Helps you understand why certain directives work in specific situations and not in others, including due to data scoping and how performance is impacted by different choices of directives
Interested in trying Parallelware Trainer?
Want to try out Parallelware Trainer and see how it can help you? Let us know and we’ll send you a complimentary 2 week trial license.
We currently only support Linux installations of Parallelware Trainer, but we will soon be providing support for Mac and Windows.
How does this work?
Once you’ve received your trial license, simply follow the instructions in the email. In three simple steps you will have Parallelware Trainer up and running and will be able to start using it to learn how to parallelize simply examples.
Using Parallelware Trainer to get to HPC
Once you are comfortable with the simple examples the next step is to import your own code and see how you can use Parallelware Trainer to learn how to parallelize your own software and how to make best use of OpenMP and OpenACC for shared memory and GPU parallelization.