Scientific Computing - HPC Source SC14 Special Edition

sands, and then millions, of cores. We’re
talking about compute power that is truly
staggering, and we need HPC code that can
fully exploit these systems.”
Burroughs added that he is “very
encouraged by the positive response to
the IPCC program Intel has received from
organizations around the world interested
in updating public scientific code.”
For example, the IPCC project at
Georgia Institute of Technology is working
to modernize quantum chemistry codes
used in materials science, biochemistry,
pharmacology, chemical physics, chemical
engineering, geology and energy research.

According to Dr. Edmond Chow, Associate Professor, School of Computational Science and Engineering, Georgia Institute of Technology, the IPCC project began with the design of a new parallel code, called GTFock, for the most fundamental approximation in quantum chemistry, called Hartree-Fock. The objectives of the code are to achieve scalability to very large numbers of compute nodes, larger than what is currently possible; and heterogeneous CPU-Intel Xeon Phi computation.

Software engineers working in the IPCC program at Georgia Tech are integrating the new parallelized GTFock code into the PSI4 quantum chemistry software package, and they are planning future integration with the NWChem application. The new code will impact hundreds of research projects, because both applications are widely used. According to Chow, an earlier version of the PSI4 package has been downloaded more than 9,000 times, and the NWChem package is the U.S. Department of Energy’s flagship quantum chemistry code and arguably the best parallel code currently available.

Dr. Chow reports that the new code is performing well on Tianhe- 2, the world’s most powerful supercomputer, using 8100 nodes, which comprises more than half of the entire Tianhe- 2 machine. With each node comprising two Intel Xeon CPUs and three Xeon Phi coprocessors, the computations used more than 1.6 million cores, all working in parallel. These runs were used in a study of truncated models of pro-tein-ligand systems, very likely the largest study of its kind, according to Dr. Chow.

The IPCC at the Irish Centre for High

Focus Areas Code/Application

Weather/Climate ADCIRC Life Science - Molecular Dynamics AMBER Aerospace AOSOA TAU Aerospace AOSOA TRACE Computational Fluid Dynamics AVBP (Large Eddy)

Astronomy/Particle Physics BAGEL Genomics BLAST Life Science Bowtie2

Quantum Chemistry BQCD Molecular Dynamics BUDE Weather/Climate Cam- 5

Density Function Theory Castep Weather/Climate CESM Manufacturing CFD Auto Weather/Climate CFS Weather/Climate CliPhi (COSMO)

Computational Fluid Dynamics COSA Life Science Cryo-EM/DTF Life Science DL_MESO Quantum Chemistry DL-Poly Weather/Climate ECHAM6

FEM Elmer Curriculum, Computational Fluid Dynamics FrontFlow/blue code Astronomy/Particle Physics GADGET Chemistry GAMESS-US Physics/CERN Geant4

Molecular Dynamics GLAT Density Function Theory GPAW Molecular Dynamics GROMACS QCD, Key Archer Aps GS2

Chemistry GT-FOCK Weather/Climate Harmonie Physics HEOM Life Science Insight Toolkit (ITK)

Chemistry, Biochem Integral Computation Library Physics/CERN KNL Evaluation Molecular Dynamics LS1 MARDYN Optimization Tool MACPO User Group & Tools MAGMA Astronomy/Particle Physics MILC Astrophysics MODAL Weather/Climate MPAS Molecular Dynamics NAMD Nano NEMO5

Life Science - Molecular Dynamics NEURON Molecular Dynamics NW Chem Visualization OSPRay Visualization ParaView Astrophysics PICOLA Physics QCD Nano QE Life Science - Molecular Dynamics QM / MM Program Language R Astronomy/Particle Physics RAMSES