HPX 1.1.0 Released!

The STE||AR Group is proud to announce the release of HPX 1.1.0, 10 years after the first commit! This release contains 2300 commits since the previous release and has closed over 150 issues. HPX 1.1.0 brings users full control over how HPX uses processing units, improvements to parallel algorithms and many other usability improvements. This release would not have been possible without the help of all the people who have contributed bug reports, questions, testing, code and improvements to the documentation. Thank you!

Download the release using the links below, from our download page, or from our GitHub page.

The highlights of this release are:

  • We have changed the way HPX manages the processing units on a node. We no longer implicitly bind all available cores to a single thread pool. The user has now full control over what processing units are bound to what thread pool, each with a separate scheduler. It is now also possible to create your own scheduler implementation and control what processing units this scheduler should use. We added the hpx::resource::partitioner that manages all available processing units and assigns resources to the used thread pools. The runtime, thread pools and individual threads can now be suspended and resumed independently. This functionality helps in running HPX concurrently to code that is directly relying on OpenMP or MPI.
  • We have continued to implement various parallel algorithms. HPX now almost completely implements all of the parallel algorithms as specified by the C++17 standard. We have also continued to implement these algorithms for the distributed use case (for segmented data structures, such as hpx::partitioned_vector).
  • The parallel algorithms adopted for C++17 restrict the iterator categories usable with those to at least forward iterators. Our implementation of the parallel algorithms supported input iterators (and output iterators) as well by simply falling back to sequential execution. We have now made our implementations conforming by requiring at least forward iterators.
  • We have added a compatibility layer for std::thread, std::mutex, and std::condition_variable allowing for the code to use those facilities where available and to fall back to the corresponding Boost facilities otherwise.
  • We have added a new launch policy hpx::launch::lazy that allows to defer the decision on what launch policy to use to the point of execution.
  • We have added several improvements to how components can be constructed.

For a complete list of new features and breaking changes please see our release notes. If you have any questions, comments, or exploits to report you can comment below, reach us on IRC (#stellar on Freenode), or email us at hpx-users@stellar.cct.lsu.edu. We value on your input!

Download links:

File MD5 Hash
zip (6.3M) 623d089b226cb2ba041330f9c41b015c
gz (4.2M) b160d31322e2205e1d17a510942ad135
bz2 (3.5M) dae616283f91157b99902e82037065b4
7z (3.1M) b5bf3df03194275d9e42ca6c74b841cd

 

GD Star Rating
loading...
    This entry was posted in HPX, Release and tagged by Hartmut Kaiser. Bookmark the permalink.

    About Hartmut Kaiser

    Hartmut is an Adjunct Professor of Computer Science at Louisiana State University. At the same time, he holds the position of a senior scientist at the Center for Computation and Technology (LSU). He received his doctorate from the Technical University of Chemnitz (Germany) in 1988. He is probably best known through his involvement in open source software projects, mainly as the author of several C++ libraries he has contributed to Boost, which are in use by thousands of developers worldwide. His current research is focused on leading the STE||AR group at CCT working on the practical design and implementation of the ParalleX execution model and related programming methods. In addition, he architected and developed the core library modules of SAGA for C++, a Simple API for Grid Applications.

    Leave a Reply

    Your email address will not be published. Required fields are marked *