Exascale Computing Project


Exascale Computing Project:  ADIOS

ORNL personnel led the following projects.


Distributed Tasking for Exascale, led by Jack Dongarra

Description: Increase scalability, interoperability, and productivity of the PARSEC Environment, which provides a runtime component for dynamic execution on heterogeneous distributed systems and a productivity toolbox that comprises a development framework supporting multiple domain specific languages and extensions, with debugging, trace collection, and analysis tools.


Open MPI for Exascale (OMPI-X), led by David Bernholdt

Prepare the MPI standard and its implementation in Open MPI for exascale through improvements in scalability, capability, and resilience.


EXA-PAPI: The Exascale Performance Application Programming Interface, led by Jack Dongarra

Description: Provide tool designers and application engineers with a consistent interface and methodology for the use of low-level performance-counter hardware found across the system (i.e. CPUs, GPUs, on/off-chip memory, interconnects, I/O system, energy/power, etc.). Enable users to see, in near real time, the relations between software performance and hardware events across the entire compute system.


PROTEAS: PROgramming Toolchain for Emerging Architectures and Systems, led by Jeff Vetter

Description: Provide productive and performance-portable programming solutions based on directive-based methodologies that support current language paradigms. Provide integrated performance assessment solutions that will enable automatic performance analysis and performance-driven optimization. Provide a integrated programming toolchain that is powerful enough to prototype the above solutions, while flexible enough to extend its functionality over time


Software for Linear Algebra Targeting at Exascale (SLATE), led by Jack Dongarra

Deliver routines for solving systems of linear equations,  least-squares problems, symmetric and non-symmetric eigenvalue problems, and singular value problems.  Similar functionality will be provided for real and complex matrices, in both single and double precision. Mixed-precision iterative refinement solvers and explicit matrix inversion routines will also be provided.


ALExa: Accelerated Libraries for Exascale, led by Wayne Joubert

Develop, enhance and support a strategic set of high-level mathematical libraries and frameworks for exascale applications:
AMP (Advanced Multiphysics Package): coupled physics services
DTK (Data Transfer Kit): multiphysics grid data transfer
TASMANIAN (Toolkit for Adaptive Stochastic Modeling and Non-Intrusive Approximation): sparse grid surrogate models, UQ


ADIOS Framework for Scientific Data on Exascale Systems, led by Scott Klasky

Support exascale applications by meeting the requirements of their I/O patterns and addressing their data management and in situ analysis needs.  Optimize I/O on exascale architectures and addressing new resources such as burst buffers, deep memory and storage hierarchies, and interconnects.  Make ADIOS easily maintainable, sustainable and extensible by the DOE community, while ensuring its performance and scalability.