Portable Application Development for Next Generation Supercomputer Architectures

PI Name: 
Tjerk Straatsma
PI Email: 
str@ornl.gov
Institution: 
Oak Ridge National Laboratory
Allocation Program: 
ALCC
Allocation Hours at ALCF: 
60 Million
Year: 
2015
Research Domain: 
Computer Science

Supercomputers are powerful scientific and mathematical instruments capable of calculating quadrillions of operations in a single second. Science harnesses the compute speed of supercomputers to tackle some of science’s biggest challenges including exploration of complex and important theories, simulation of otherwise impossible experiments on sub-atomic to cosmological scales, and rapid prototyping to speed time-to-solution for industrial technology. Although supercomputers can significantly advance science and technology, a great barrier to benefitting from supercomputers exists in the challenge of programming application software that is capable of using these massively parallel systems.

Three science driven DOE supercomputing facilities; Argonne Leadership Computing Facility (ALCF), National Energy Research Scientific Computing Center (NERSC) and the Oak Ridge Leadership Computing Center (OLCF), are supporting efforts to prepare scientists and engineers for the next-generation supercomputers that are planned for these centers through early application readiness programs that will prepare a range of application codes. The ALCF has the Early Science Program (ESP), NERSC the NERSC Exascale Science Application Program (NESAP), and the OLCF, the Center for Accelerated Application Readiness (CAAR) program.

This allocation supports these three facilities as they launch these application readiness programs addressing 40 key application codes. In order for the application readiness programs to be successful it is crucial that application developers have allocations not only at their sponsored supercomputing facility, but also at a facility with an alternate architecture so portability issues can be explored. This project will support the computational resources on the current systems to start this collaborative effort and help scientists and engineers leverage future HPC systems.