CSE Reports and Case Studies
This section contains reports from completed distributed CSE projects along with other miscellaneous technical reports
produced by the CSE team following shorter periods of activity with users' codes. There are also examples of dCSE case studies including user
comments on how the performance and efficiency of their codes have been improved by the help of
dCSE support.
Distributed CSE Success Stories
Distributed CSE Reports
- Bands parallelism in CASTEP Keywords: Parallel matrix diagonalisation, ScaLAPACK, FFTW
- NEMO on HECToR Keywords: I/O improvement, netCDF 4.0, HDF5
- CP2K Keywords: Load balancing, FFT performance optimisation, Compiler comparison
- CASINO Keywords: Hybrid OpenMP / MPI, System V shared memory, I/O improvement
- Performance enhancements for the GLOMAP aerosol model Part 1 Keywords: Compiler optimisation, Memory striding, Loop ordering, Profiling
- Performance enhancements for the GLOMAP aerosol model Part 2 Keywords: Mixed mode parallelism, MPI task placement, OpenMP scheduling
- Porting OpenFOAM to HECToR Keywords: Dynamic libraries, C++, Optimum cores
- Citcom Keywords: Multigrid, C, Mesh Refinement
- Scaling the EBL Turbulence Application to Thousands of Cores Keywords: FFT, Spectral, 2-D domain decomposition
- Future Proof Parallelism for Electron-Atom Scattering Codes with PRMAT Keywords: Sub-task management, Load balancing, Optimization, Multi-core, Fortran 2003
- WRF code Optimisation for Meso-scale Process Studies (WOMPS) Keywords: Cache Optimisation, I/O improvement, Hybrid OpenMP / MPI
- Efficient Massively-Parallel Tools for the Study of Catalytic Chemistry with ChemShell Keywords: Task farming, Tcl, Parallel numerical optimisation
- Fluidity-ICOM: High Performance Computing Driven Software Development for Next-Generation Modelling of the Worlds Oceans Keywords: PETSc, Unstructured Mesh, Adaptivity, Profiling, Mesh Optimisation, Parallel I/O
- Optimisation of the Input and Output (I/O) in DL_POLY_3 Keywords: Molecular Dynamics, Domain decomposition, Performance, Parallel I/O, netCDF
- Hybrid Time-Dependent Density Functional Theory in CASTEP Part 1 Keywords: Eigensolver, ARPACK, Multi-core Scalability
- Hybrid Time-Dependent Density Functional Theory in CASTEP Part 2 Keywords: Geometry optimisation, Excited state forces, Modular code
- Massive Remote Batch Visualizer (Porting AVS/Express to HECToR) Keywords: Visualization, Image compositing, AVS/Express
- Upgrading the FFTs in GS2 Keywords: FFTW2/3, SSE instructions, Indirect addressing
- Performance Optimizations for CARP Keywords: PETSc, Unstructured decomposition, Adaptivity, Preconditioning, Mesh optimisation
- Improving the scalability of CP2K on multi-core systems Keywords: Compiler comparison, Hybrid OpenMP / MPI, FFTW
- SPRINTing with HECToR Keywords: Computational statistics, Bioinformatics, R, Correlation, Permutation
- Improving 3D FFT performance in GWW Keywords: Quantum Espresso, GWW, 3D FFT, Pencil decomposition
- Parallelisation of CABARET Keywords: Unstructured Mesh, Compressible CFD, Hybrid OpenMP / MPI
- Direct Numerical Simulations (DNS) of Turbulent Fluid Flows Keywords: Incompressible CFD, Domain decomposition, FFT, System V shared memory
- Metal Conquest Keywords: ScaLAPACK, Matrix diagonalisation, K-point parallelism
- Implementation of established algorithms to extend HELIUM Keywords: Scalable code, Hybrid OpenMP / MPI, Many-core
- Micromagnetic modelling of naturally occurring magnetic mineral systems Keywords: PETSc, Sparse matrix construction, Unstructured
- Implementing dynamic load balancing and rigid body dynamics within DL_POLY_3 Keywords: Molecular Dynamics, Dynamic load balancing, Static domain decomposition
- Optimising the performance of the VASP code on HECToR Keywords: Ab initio electronic structure calculations, Collective communications, k-point parallelism
- HECToR enabled Step Change in Turbulent Multiphase Combustion Simulations Keywords: Direct Numerical Simulations, Mixed mode parallelism, Parallel I/O
- SPRINTing further with HECToR Keywords: Computational statistics, Bioinformatics, R, Correlation, Permutation
- Improvements for multi-core performance and domain choice within DL_POLY_4 Keywords: Molecular Dynamics, Code optimisation, FFT, Domain decomposition
- Optimization of the MPI parallel RMT code for HECToR and likely successors Keywords: Red-black load balancing, Finite-differencing, Numerical integration
- A New CASTEP and ONETEP Geometry Optimiser Keywords: Non-linear Optimization, Memory scaling, DFT Geometry Optimization, L-BFGS
- Porting and Optimisation of Code_Saturne on HECToR Keywords: Mesh partitioning, Metis, ParMetis, PT-Scotch, Zoltan
- Implementation of a Divide and Conquer Strategy for the Materials Modelling Code CRYSTAL Keywords: Linear scaling DFT, Multipole expansion, Task farming
- Adding the molecular dynamics functionality to the quantum Monte Carlo code CASINO Keywords: Petascale, Diffusion Monte Carlo, Data re-distribution / load balancing
- CP2K - Sparse Linear Algebra on 1000s of cores Keywords: Block Compressed Sparse Row, Sparse matrix-matrix multiplication, Mixed-mode
- Improving Load Balancing and Parallel Partitioning in Fluidity Keywords: Adaptive partitioning, Dynamic load balancing, Unstructured decomposition
- Improved Data Distribution Routines for Gyrokinetic Plasma Simulations Keywords: Indirect addressing, Unbalanced decomposition, Real space to k-space transfer
- RMT for High Harmonic Generation Keywords: Scalable code, Finite-differencing, Numerical integration
- Boosting the scaling performance of CASTEP: enabling next generation HPC for next generation science Keywords: I/O improvement, Band parallelism, MPI buffer optimisation
- Direct Numerical Simulations (DNS) of Turbulent Fluid Flows II Keywords: CFD, Turbulence simulation, Parallelisation, Code modernisation
- Bootstrapping and support vector machines with R and SPRINT on HECToR Keywords: Computational statistics, Bioinformatics, R
- Scalable coupling of Molecular Dynamics and Direct Numerical Simulation of multi-scale flows Part 1 Keywords: CFD, Parallelisation, Multi-scale modelling
- Scalable coupling of Molecular Dynamics and Direct Numerical Simulation of multi-scale flows Part 2 Keywords: Continuum-Molecular coupling, MPMD Parallelisation, MD farming
- Adding Spin Polarisation and van der Waals Energy Correction to Conquest Keywords: Order N DFT, Radial FFT, Matrix diagonalisation
- Improved Scaling for Direct Numerical Simulations of Turbulence Keywords: Pseudospectral, 2-D domain decomposition, FFTW3
- Enhancement of a high-order CFD solver for many-core architecture Keywords: Compiler performance comparison, Red-black shared memory parallelism, Mixed-mode
- Combined-Multicore Parallelism for the UK electron-atom scattering Inner Region R-matrix codes on HECToR Keywords: Shared memory segments, Asynchronous parallel I/O and passive one-sided communication, Object-oriented Fortran 2003
- Parallelisation and porting of UKRMol-in the electron-molecule scattering inner region R-matrix codes Keywords: PETSc matrix construction, SLEPc, Hamiltonian construction
- Developing NEMO for Large Multi-core Scalar Systems Keywords: Array index re-ordering, Mixed cell domain decomposition, Code for vector architectures
- CABARET on Jet Flap Noise and Quasigeostrophic Ocean Circulation Models Keywords: Global communications, FFT, Unstructured Mesh
- Optimisation of VASP Density Functional Theory (DFT)/Hartree-Fock (HF) hybrid functional code using mixed-mode parallelism Keywords: OpenMP-MPI, VASP5, FFTW3
- Adding Parallel I/O to PARA-BMU Keywords: ASCII I/O, Binary I/O, netCDF, HDF5, File compression
- Improving Scaling in Conquest for HECToR Phase 3 - OpenMP-MPI Hybrid Implementation Keywords: Order N DFT, DGEMM, Weak-scalability
- Developing Hybrid OpenMP/MPI Parallelism for Fluidity-ICOM - Next generation Geophysical Fluid Modelling Technology Keywords: FEM, NUMA, Graph colouring
- Micromagnetic modelling of naturally occurring magnetic mineral systems: II Keywords: PETSc, Sparse matrix construction, Unstructured
- Cloud and Aerosol Research on Massively-parallel Architectures (CARMA) Keywords: FFT replacement, GMRES, Parallel BiCGStab
- Mesh reordering in Fluidity using Hilbert space-filling curves Keywords: Unstructured finite elements, Domain decomposition, NUMA
- Software Framework to Support Overlap of Communications and Computations in Implicit CFD Applications Keywords: Non-blocking MPI collectives, MPI-3.0, 3D FFTs
- Improving CONQUEST to allow ab initio molecular dynamics calculation on 100,000+ atoms Keywords: Hilbert space filling curve, Dynamic load balancing, QM/MM
- Optimising the Parallelisation of a Harmonic Balance Navier-Stokes Solver for the Ultra-rapid Analysis of Wind Turbine, Turbomachinery and Aircraft Wing Periodic Flows Keywords: Structured multi-block, MPI message aggregation, Re-structuring hybrid OpenMP and MPI
- Opening up HPC to the Discrete Element Method User Community Keywords: Discrete Element Modelling / Granular LAMMPS, Boundary conditions, Inter-grain bonding model
- Microiterative QM/MM Optimisation for Materials Chemistry Keywords: DL-FIND, Transition state / Reaction path, Shell model optimisation
- Performance Enhancement and Optimization of the TPLS and DIM Two-Phase Flow Solvers Keywords: CFD, PETSc, Preconditioned Krylov solvers
- Preparing DL_POLY_4 for the Exascale Keywords: Second level parallelism via OpenMP, Parallel link cell algorithm, 64 bit integers
- Improved Global High Resolution Chemistry-Aerosol Modelling for Climate and Air Quality Keywords: Parallel file access (IO), Code restructuring, Profiling
- Seismic wave calculation at the Hz level for a general Earth Keywords: Cray compiler, Profiling, Code optimisation
- Parallelisation of the adaptive kinetic Monte Carlo code, DL_AKMC Keywords: Parallel dimer search, Energy/Force evaluations, Diffusion modelling
- Enhancement of high-order CFD solvers for many-core architecture Keywords: Structured multi-block, Domain decomposition, Hybrid OpenMP / MPI
- Multigrid solver module for ONETEP, CASTEP and other codes Keywords: Multigrid, ONETEP, CASTEP
- Adapting QSGW to large multi-core systems Keywords: Electronic structure calculations, OpenMP / MPI parallelization, Loop parallelization
- Improving the scaling and performance of GROMACS on HECToR using single-sided communications Keywords: SHMEM, MPI point-to-point, Molecular simulation
- Further Improving NEMO In Shallow Seas (FINISS) Keywords: Array index re-ordering, Mixed cell domain decomposition, Code for vector architectures
- DL_POLY_4 Algorithmic Improvements Keywords: Verlet neighbour list, Symplectic multiple time stepping, Molecular interaction
- Massively Parallel Computing for Incompressible Smoothed Particle Hydrodynamics (ISPH) Keywords: Single-sided communications, Neighbour list searching, PETSc
- Implementing OpenIFS on HECToR Keywords: Integrated (Weather) Forecast System, Parallel I/O, Meteorological archival and retrieval system
Case Study CSE Reports
- How to make best use of the AMD Interlagos processor Keywords: HECToR Phase 3, Interlagos, Bulldozer, Benchmarks
- Optimising Combustion Code SoFTaR Keywords: Optimisation, global transposition, domain decomposition
- HECToR Fortran Compiler Performance Comparison Keywords: Cray, GNU, Intel, PGI, Pathscale
- CASTEP Quad Core Benchmarking and Optimisation Addendum to bandblocking Keywords: System V shared memory, FFT transpose, bandblocking
- Optimising HYPO4D on the X2 vector machine Keywords: Optimisation, Cray X2, vectorisation, Lattice Boltzmann
- Benchmarking CASTEP 4.3 on the X2 vector machine Keywords: Porting, benchmarking, profiling
- CSE optimisation and scaling advice for a CFD code Keywords: Compiler comparison, optimisation
- Guide to Partitioning Unstructured Meshes for Parallel Computing Keywords: Decomposition, efficient communication
For a list of current
dCSE projects that are under contract and/or underway please
see
here