Skip to content

v0.11.0
Compare
Choose a tag to compare
@rhornung67 rhornung67 released this 27 Jan 23:06
v0.11.0
0502b9b
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.

This release contains new features, several notable changes, and some bug fixes.

Please download the RAJA-v0.11.0.tar.gz file below. The others will not work due to the way RAJA uses git submodules.

Notable changes include:

  • New features:

    • HIP compiler back-end added to support AMD GPUs. Usage is essentially the same as for CUDA. Note that this feature is considered a work-in-progress and not yet production ready. It is undocumented, but noted here, for friendly users who would like to try it out.
    • Updated version of camp third-party library, which includes variety of portability fixes. Most users should not need to concern themselves with the details of camp.
    • Added new tutorial material and exercises.
    • Documentation improvements.

  • API Changes:

    • None.

  • Build changes/improvements:

    • RAJA version number is now accessible as #define macro variable constants so that users who need to parameterize their code to support multiple RAJA versions can do this more easily. See the file RAJA/include/RAJA/config.hpp for details. RAJA version numbers are also experted as CMake variables.
    • Added support to link to external camp library. By default, the camp git submodule will be used. If you prefer to use a different version of camp, set the RAJA CMake variable 'EXTERNAL_CAMP_SOURCE_DIR' to the location of the desired camp directory.
    • BLT submodule (CMake-based build system) has been updated to latest BLT release (v0.3.0). The release contains a new version of GoogleTest, which required us to modify our use of gtest macros and our own testing macros. For the most part, this change should be invisible to users. However, the new GoogleTest does not work with CUDA versions 9.1.x or earlier. Therefore, if you compile RAJA with CUDA enabled and also wish to enable RAJA tests, you must use CUDA 9.2.x or newer.

  • Bug fixes:

    • Fixed various issues to make internal implementations more robust, resolved issues with non fully-qualified types in some places, and work arounds for some compiler issues.
Assets 3
Loading