초고성능컴퓨팅 계산과학공학 연구개발 및 출연(연) R&D 적용

Abstract

ㅇ 계산과학공학 선도 연구개발 및 초고성능컴퓨팅 기반 R&D 효율화 - 초고성능컴퓨팅 기반 R&D 효율화 수행 - 초고성능컴퓨팅 기반 고에너지 물리, 생물물리, 나노소재 및 천체물리 연구 - 연합 인증, 협업 클라우드 및 데이터 중심 네트워킹 기술 개발

ㅇ 거대 컴퓨팅 가시화 기술 개발 및 계산과학 적용 - 세계 수준의 거대 컴퓨팅 계산과학공학 가시화 기술 개발로 R&D 효율화 및 선도 기술 확보 - 기술의 실용화 및 기술이전으로 거대 컴퓨팅 가시화 기술 활용확대

ㅇ R&D 효율화를 위한 계산과학공학 플랫폼 개발 및 체계 구축 - 계산·실험·데이터 공유 플랫폼 개발 - R&D 효율화를 위한 출연연 협력체계 구축 및 활성화 - 유동해석 분야 출연연 공동연구 및 효율화 사례 발굴

(출처 : 보고서 초록 3p)

IV. Results of the study □ The massive data visualization technology development - The performance improvement for massive data visualization · The real-time visualization performance for 3TB data has reached to of 5.2 seconds by applying parallel data processing technology · The volume rendering performance has been improved to 7.26 times faster than existing CPU-based algorithm by developing GPU-accelerated technology · The grid/cell based visualization performance has been improved to 5.5 times before by developing and applying the hierarchical-tree data structure for grid exploration · The out-of-core visualization performance has been improved to 1.5 times by developing data fetching technique for large data visualization · The developed technology has been applied to KFX project by technology transfer

- The application of real-time thermal hydraulic power data visualization technology for virtual nuclear reactor · The memory efficiency for parallel calculation has been improved 77% by applying the massive form data partitioning technique · The prototype of implementing the integrated analysis environments for multi-physics has been developed to verify the normal status of full core by applying thermal hydraulic power-neutron dynamics related analysis methods(The nuclear reactors in service, Hanbit unit #3, #4) · The prototype of customized visualization tool for nuclear reactor thermal hydraulic power analysis data has been developed

□ Advanced research and development of computational science engineering and High Performance Computing based National R&D efficiency - National R&D efficiency based on High Performance Computing · Computational screening of surface properties for nano catalysts (KISTI-KRICT) · Nano-Particle / Graphene Nanocomposites Energy-based Materials Computational Science-based Experimental Research Integrated Property Analysis Research(KISTI-KIER) · Computational Study of Zintl Phase for Electrode Materials of Rechargeable Batteries(KISTI-KBSI) · Development of nano-patterned surface material based on super-computing (KISTI-KIMM) - Advanced research and development of computational science engineering · Research on dark matters · Research on condensation phenomenon within the bacterial chromosome · Numerical simulation on spinning black holes · Federated access management technologies for worldwide e-resources · Technology development of OpenSciWay platform · Technology development of data named networking for big data science

□ Computational Science Engineering Platform for R&D Efficiency - Development of open computational science engineering platform for research · Designing a data model suitable for solving the diversity problem of computational science simulation data · Improving data quality and providing the basis of analysis for computational science data through development of data pre-processing framework with flexible structure · Development of specialized computational science data sharing portal and building computational science database of materials field · Development of material property data analysis tool such as Battery Explorer, 3-D Chart, and etc. · Design and development of customized data representation technology based on data type · Development of machine learning based material property prediction service that dramatically shortens the time required for material experiment or simulation

- Turbomachinery R&D efficiency using high performance computing (joint with KIMM) · Achieves 3% accuracy compared to wind tunnel test for wing flow analysis problem · Establishs and demonstrates a leading web-based research environment for the analysis of gas turbine blades · Develops key element algorithms (grid generation automation tool, adaptive orthogonal decomposition reduction modeling tool) for data-based engineering analysis service

- Promoting computational science & engineering in government R&D agencies and Supporting human resource development · Community workshops to strengthen cooperation for R&D efficiency, and promote computational science engineering in the bio/nano/materials fields (5/24~25, KT HRD, 167 persons attending) · Operation of Korea e-Science Forum to activate computational science engineering community (a quarterly consultation meeting) · Seasonal school held to train human resources in the computational science engineers field (8/21, 8/28, 10/26) · Signed a business agreement to promote computational science engineering technology (MOU with SNU Hospital and Society for CSE) · Development of computational science engineering education contents (Computation chemistry) for training of computational science human resources

(출처 : SUMMARY 11p)