OpenMP ARB Releases Technical Report with Support for a Tool Interface
Salt Lake City, Utah – November 11, 2016 The OpenMP® Architecture Review Board (ARB) announces the release of Technical Report 4 as a preview for the future OpenMP 5.0 API describing language features for support for a tool interface and for task and taskloop reductions as well as many improvements for the OpenMP device constructs.
The key features that are included in Technical Report 4 (pdf) include:
- Support for a tool interface. To enable development of high-quality, portable tools that support monitoring and performance analysis of OpenMP programs developed using any implementation of the OpenMP API, the OpenMP API now includes a tool interface.
- Support for task and taskloop reductions. Language features have been added to OpenMP API to support reductions amongst tasks.
- Improvements to OpenMP device constructs. Major improvements have been made to the usability of OpenMP support for devices such as GPUs, particularly for C++.
This Technical Report is a work in progress. The goal of this release is to get feedback on the proposed API. Feedback can be posted on the TR4 Discussion Forum (registration is required).
In addition, the OpenMP ARB also announces the release of the OpenMP Examples document. This documents contains a collection of programming examples meant to supplement the OpenMP 4.5 API specifications. The source codes of the examples are available for general access.
“OpenMP 5.0 will be the next version of the OpenMP specification, which we expect will be officially released in 2018,” says Bronis R. de Supinski, the Chair of the OpenMP Language Committee and CTO of Livermore Computing. “TR4 can be viewed as an alpha release of OpenMP 5.0: Its new features will be included in OpenMP 5.0, but the final release will include several other advances to the specification, as well as refinements to the specification of the tool interface.”
“The new OpenMP tools interface, also known as OMPT, will enable the development of tools that work reliably across all implementations of the OpenMP API,” says Martin Schulz, the lead of the working group for tools in the OpenMP language committee. “This will be the catalyst for a new generation of third party tools, which will significantly easy the development and optimization process of OpenMP applications.”
“TR4 contains several features that will help OpenMP maintain its strong position as the gold standard in the domain of shared memory programming,” says Michael Klemm, CEO of the OpenMP ARB.