v5.0

v5.0 contains a few bug fixes in the CMake build system, but the key new features are:

• Additional programming models, including Fortran, SYCL 2020 with Unified Shared Memory, and Futhark
• Reporting of performance of the array initialisation and copy back of solutions to the host that happen before and after (respectively) of the main benchmarking.
• Preliminary support for building with Spack.

Find out more in our blog post.

v4.0

This update adds new implementations in parallel programming models, including ISO C++, SYCL 2020, TBB and Thrust. We also include new standalone implementations in the Julia, Rust and Scala languages. Our build system has been updated to use CMake exclusively, and provides recommended build parameters for major compilers for all programming models.

For more details, please see the accompanying blog post.

Full list of changes available in the CHANGELOG.

v3.4

Improvements added to support building for newer architectures.

Full list of changes available in the CHANGELOG.

Support added for measuring cache bandwidths

Results presented at SC'17:
Deakin T, Price J, McIntosh-Smith S. Portable methods for measuring cache hierarchy performance. 2017. Poster sessions presented at IEEE/ACM SuperComputing, Denver, United States.
You can view the poster and Extended Abstract

v3.2

OpenMP, Kokkos and RAJA have had minor updates to:

• align memory
• ensure streaming stores with the Intel compiler
• Kokkos and RAJA iterator types updated

v3.1: Build system updated.

The new build system uses a series of Makefiles, rather than CMake.
Due to the cross platform and cross programming model combinations,
different compilers on the same platform are often required.
Therefore it is simplest to use simple Makefiles to choose the right
compiler for the right model than rely to CMake for this.

v3.0

Version 3.0 includes the Dot kernel

v1.0

Bump version string

v0.9

Bump version to 0.9