Heterogeneous Parallel Programming
HSA opens the door to a new class of solutions that are difficult to achieve in historical architectures where inter-task/process communication is important across heterogeneous compute devices. In response to that challenge, HSA addresses a broad class of approaches to application parallelism.
- Data Parallelism – Embarrassingly Parallel Applications
- Task-Parallelism
- Nested-Parallelism
- Braided-Parallelism
This capability opens up more efficient implementations of MapReduce, Hash Tables, Sparse Matrix Vectors and Conjugate Gradients solvers, FETI-DP Method, List Ranking, and spatial search.
More importantly this is just a small sampling of problems that can be addressed by HSA enabled platforms.
Design, Create and Innovate a New Class of Applications, Tools, Libraries and Runtimes for HSA
The HSA Foundation developer program offers a complete set of resources that will help you explore how heterogeneous system architectures can free you to drive a new level of innovation into your application.