User:Soon-Heum Ko (NSC)

Soon-Heum Ko (NSC) Application expert in Computational fluid dynamics, 100% full time equivalent, financed by SNIC (30%) and PRACE (70%) General activities PRACE European Project, SMHI HYPRE Parallelisation, SNIC Parallelisation Support Office National Supercomputing Center Room 181, House G Linköping University SE-581 83 Linköping, Sweden Phone (Office) +46 (0)13 82 25 11 (Mobile) +46 (0)734 61 31 75

Quick facts

• If you feel troubled in pronouncing Soon-Heum, then just call me Jeff
• Work at NSC since 2011
• Ph.D. in 2008; Two years of postdoctoral experience in computational science (Code porting to scientific frameworks, hybrid CFD-MD simulation with scheduling of coupled applications)
• Details on experiences/researches and biography can be found here

Expertise

• MPI / OpenMP
• Fortran / C / C++
• DDT

Projects

A. From PRACE

• Parallel I/O Implementation on the Multiple Sequence Alignment Software ClustalW-MPI
• Performance Benchmark of NEMO Oceanography Code
• Enabling Xnavis for Massively Parallel Simulations of Wind Farms
• Computer-Aided Drug Design

B. With SMHI

• HYPE Code Parallelisation

C. From SNIC

• Dalton CPP-LR parallelization

Old Projects

Ended in 2012

• LES_Code_Parallelization NSC-promoted code parallelization support. A collaboration with Prof. L. Davidson from Charmers University on parallelizing his LES code.
• Performance_Analysis_of_ad_OSS_Program Performance Analysis for ad_OSS Program. The work performs profiling and analysis of an in-house water molecule modeling code, called 'ad_OSS'. Work has been achieved for the consultation of Prof. Lars Ojamae @ LiU.
• Synthetic_Benchmark_on_Curie Synthetic benchmark analysis on Curie Tier-0 system. A 1IP task in the PRACE project, which objects to develop the European HPC Ecosystem.
• NSC_GPU_and_Accelerator_Pilot GPU/Accelerator Pilot Project at NSC GPU/IntelMIC system construction and the programming model test.

Ended in 2013

• HYPE Code Parallelisation Performance analysis of SMHI's HYPE code which solves the ground flow for pollution analysis
• Parallel I/O Implementation on the Multiple Sequence Alignment Software ClustalW-MPI Designing and implementing parallel I/O interface for a faster I/O operation on massive-sized sequence string dataset running on thousands of cores
• Performance Benchmark of NEMO Oceanography Code Compilation-level tuning and performance benchmarking of NEMO oceanography code on CURIE PRACE resource
• Enabling Xnavis for Massively Parallel Simulations of Wind Farms Implementation of MPI-I/O for reading/writing stuructured multiblock CFD dataset

Ended in 2014

• Computer-Aided Drug Design Enabling LSDALTON's DFT method (enabling a real 64-bit environment) for simulating large molecule of biological interest
• Dalton CPP-LR parallelization Parallelisation of one branch of DALTON code