StarPU is a run-time system that offers support for heterogeneous multicore machines. While many efforts are devoted to design efficient computation kernels for those architectures (e.g. to implement BLAS kernels on GPUs), StarPU not only takes care of offloading such kernels (and implementing data coherency across the machine), but it also makes sure the kernels are executed as efficiently as possible.

TopoMatch leverages on the Scotch library to handle any type of topologies and not only trees. Its main features are: - Handling any type of topologies (tgt Scotch format or hwloc format). - Handle large communication patterns (up to hundreds of thousands of processes and processing units) . - Manage binding constraints: you can specify a subset of the node onto which you want to do the mapping. - Manage oversubscribing: you can specify that more than one processes can be mapped onto a each processing unit. - Deal with logical numbering. Physical core numbering can be used with XML/HWLOC topologies. - Provide exhaustive search for small cases. - Adaptive algorithmic that provide a good trade-off between quality and speed. - Crucial sections of the code are multithreaded. - Optimize I/O to read large input files. - Portable on Unix-like systems (Linux, OS-X, etc.). - Many useful options (level of verbosity, topology optimization, partitioning, etc.).

SCOTCH is a set of programs and libraries which implement the static mapping and sparse matrix reordering algorithms developed within the SCOTCH project. Its purpose is to apply graph theory, with a divide and conquer approach, to scientific computing problems such as graph and mesh partitioning, static mapping, and sparse matrix ordering, in application domains ranging from structural mechanics to operating systems or bio-chemistry.

SCOTCH is a set of programs and libraries which implement the static mapping and sparse matrix reordering algorithms developed within the SCOTCH project. Its purpose is to apply graph theory, with a divide and conquer approach, to scientific computing problems such as graph and mesh partitioning, static mapping, and sparse matrix ordering, in application domains ranging from structural mechanics to operating systems or bio-chemistry.

IOR is a parallel IO benchmark that can be used to test the performance of parallel storage systems using various interfaces and access patterns. The IOR repository also includes the mdtest benchmark which specifically tests the peak metadata rates of storage systems under different directory structures. Both benchmarks use a common parallel I/O abstraction backend and rely on MPI for synchronization.

The open source library MpiFx provides modern Fortran (Fortran 2003) wrappers around routines of the MPI library to make their use as simple as possible. Currently several data distribution routines are covered.

Fypp is a Python powered preprocessor. It can be used for any programming languages but its primary aim is to offer a Fortran preprocessor, which helps to extend Fortran with condititional compiling and template metaprogramming capabilities. Instead of introducing its own expression syntax, it uses Python expressions in its preprocessor directives, offering the consistency and versatility of Python when formulating metaprogramming tasks. It puts strong emphasis on robustness and on neat integration into developing toolchains.

No description on the website

CHAI is a C++ libary providing an array object that can be used transparently in multiple memory spaces. Data is automatically migrated based on copy-construction, allowing for correct data access regardless of location. CHAI can be used standalone, but is best when paired with the RAJA library, which has built-in CHAI integration that takes care of everything

CAMP collects a variety of macros and metaprogramming facilities for C++ projects. It's in the direction of projects like metal (a major influence) but with a focus on wide compiler compatibility across HPC-oriented systems.

BLT is a streamlined CMake-based foundation for Building, Linking and Testing large-scale high performance computing (HPC) application.

Pugixml is a C++ XML processing library, which consists of a DOM-like interface with rich traversal/modification capabilities, a fast XML parser which constructs the DOM tree from an XML file/buffer, and an XPath 1.0 implementation for complex data-driven tree queries. Full Unicode support is also available, with Unicode interface variants and conversions between different Unicode encodings which happen automatically during parsing/saving.

LvArray is a collection of container classes designed for performance portability in that they are usable on the host and device and provide performance similar to direct pointer manipulation

HDF5 tools for use by GEOS

HDF5 is a suite that makes possible the management of extremely large and complex data collections.

RAJA offers portable, parallel loop execution by providing building blocks that extend the generally-accepted parallel for idiom. RAJA relies on standard C++14 features.

Adiak is a library for recording meta-data about HPC simulations. An HPC application code may, for example, record what user invoked it, the version of the code being run, a computed time history showing density changes, or how long the application spent performing file IO. Adiak represents this metadata as Name/Value pairs. Names are arbitrary strings, with some standardization, and the values are represented by a flexible dynamic type system

Conduit is a project from Lawrence Livermore National Laboratory that provides an intuitive model for describing hierarchical scientific data in C++, C, Fortran, and Python. It is used for data coupling between packages in-core, serialization, and I/O tasks.

Umpire is a resource management library that allows the discovery, provision, and management of memory on machines with multiple memory devices like NUMA and GPUs.

librsb is a library for sparse matrix computations featuring the Recursive Sparse Blocks (RSB) matrix format. This format allows cache efficient and multi-threaded (that is, shared memory parallel) operations on large sparse matrices. The most common operations necessary to iterative solvers are available, e.g.: matrix-vector multiplication, triangular solution, rows/columns scaling, diagonal extraction / setting, blocks extraction, norm computation, formats conversion. The RSB format is especially well suited for symmetric and transposed multiplication variants. Most numerical kernels code is auto generated, and the supported numerical types can be chosen by the user at build time. librsb can also be built serially (without OpenMP parallelism), if required. librsb also implements the Sparse BLAS standard, as specified in the [BLAS Technical Forum] documents.

Mongoose is a graph partitioning library. Currently, Mongoose only supports edge partitioning, but in the future a vertex separator extension will be added.

Not Another Completely Heuristic Operating System, or Nachos, is instructional software for teaching undergraduate, and potentially graduate level operating systems courses. As this stage, the package only processes the kernel part (but not the ./test/ part).

PajeNG (Paje Next Generation) is a re-implementation (in C++) and direct heir of the well-known Paje visualization tool for the analysis of execution traces (in the Paje File Format) through trace visualization (space/time view). The tool is released under the GNU General Public License 3. PajeNG comprises the libpaje library, and a set of auxiliary tools to manage Paje trace files (such as pj_dump and pj_validate). The space-time visualization tool called pajeng is deprecated (removed from the sources) since modern tools do a better job (see pj_gantt, for instance, or take a more general approach using R+ggplot2 to visualize the output of pj_dump). This effort was started as part of the french INFRA-SONGS ANR project. Development has continued through a collaboration between INF/UFRGS and INRIA.