Co-Design: Deploying Leading-Edge Computing Capabilities

Current Proposed Set of LLNL Proxies

The following table outlines the proposed suite of proxy applications either developed, in development, or planned at LLNL as part of our co-design effort. Several are available today for download, while others will require you to contact us for further information while we build a comprehensive site for distribution and feedback.


Name Type Language(s) Size(loc) Description
AMG2013 mini C, MPI, OpenMP 75,000 Algebraic Multi Grid (subset of hypre library)
Kripke mini C++, MPI, OpenMP 4,200 Sweep based, structured mesh deterministic transport
Laghos mini C++, MPI 2,000+
dependency on MFEM
Compressible shock hydrodynamics using unstructured high-order finite elements.
LULESH mini C++, MPI, OpenMP 5,250 Explicit Lagrangian shock hydrodynamics on unstructured mesh representation
MACSio mini

C, C++, Fortran, MPI + various I/O libs

10,000 Multi-purpose, Application-Centric, Scalable I/O
Pynamic skeleton C, MPI, python   Dynamically-linked library loading test
Quicksilver mini C++, MPI, OpenMP, Cuda ~8,000 Monte Carlo Particle Transport, multigroup cross sections

No plans for further development. Provided as-is. Access restricted per EAR99. Available to select partners on a case-by-case basis.

Name Type Language(s) Size(loc) Description
Lassen mini C++, Charm++ 3,500 Front-tracking through a 2D mesh. Highly load-imbalanced
Mulard mini C++ 46,000 Unstructured mesh, finite element, implicit multigroup radiation diffusion

Unsupported and replaced by a newer supported proxy app, see individual app page for details.

Name Type Language(s) Size(loc) Description
LCALS kernel C++ 4,000 Suite of kernels in a unified framework for testing compilers SIMD and threading
MCB mini C++, MPI, OpenMP 13,200 Monte Carlo Particle Transport