Software

     
     
     
     
     
     
     
     
     
     


     

TOPS draws upon years of research investments by DOE, NSF, NASA, and other agencies. Software for PDEs and linear algebra that we brought into this project has already
  • empowered science that has appeared on the covers of Science and of Nature
  • been the core of a Gordon Bell Prize-winning computation
  • enabled many DOE scientists and engineers to meet their aggressive milestones under the ASCI program on the ASCI platforms, the world's most capable scientific computers, each consisting of thousands of processors
    • In fact, you have probably already used some of our software, either directly or indirectly, e.g., through commercial packages such as MATLAB, whose high-performance kernels are based on our performance-optimization technology.
    • Hypre - A family of powerful, scalable preconditioners
    • PARPACK - Scalable solvers for sparse generalized eigensystems
    • PETSc - Scalable nonlinear and linear solvers for PDEs and related problems
    • SLEPc , A scalable eigenvalue solvers package
    • SUNDIALS - Parallel ODE integrators and nonlinear solvers
    • SuperLU - Parallel direct sparse LU factorization
    • TAO - Scalable optimization software
    • Veltisto - Scalable PDE-constrained optimizer

    One of our main technical objectives under the SciDAC is to make all of this software interoperate, so that you may switch and tune solution algorithms at your convenience, with minimal recoding and even minimal recompilation. We also want to allow your codes to reach maximum performance potential, by minimizing the number of times your data has to be recopied into and out of our data structures, or between data structures internal to our solution routines.

    Our goal is to combine functionality that includes the "latest and greatest" in optimal algorithms with high performance and scalability.

    Please visit this page frequently for updates.