안녕하세요.
한국차세대컴퓨팅학회 주최로 Multicore Parallel Programming 및 Performance Optimization 교육이 실기되기에 회원님들께 알려드리고자 합니다.
자세한 사항은 아래의 내용을 참조해 주시기 바랍니다.
◇ 기 간 : 2013.1.30(수)-2013.2.1(금)
◇ 장 소 : 한국과학기술정보연구원(KISTI)
◇ 주 최 : 한국차세대컴퓨팅학회
◇ 강 사 : 명지대학교 이명호교수
◇ 참가비 : 900,000원(3일 교육, 강의자료) - 무통장입금처 : 우리은행 1005-201-651807 (사)한국차세대컴퓨팅학회
◇ 참가 문의 : 이명호교수(명지대학교, myunghol@mju.ac.kr, 031-330-6783)
◇ 강의 내용 ◇
- Multicore Processor Architecture 및 HPC Architecture Overview
- Multicore Processor Architecture
- Multicore processor architecture
- Manycore GPU Architecture
l Parallel Architecture
– Parallel Architecture Overview
- Shared-Memory Multiprocessor
- Message-Passing Multicomputer
– Cache Coherence
- Snooping-based coherence
- Directory-based coherence
– Interconnection Network
- Topology
- Routing and Control Flow
- Parallel Programming Overview Parallel Computing Models
- Theoretical Parallel Models
- Parallel Algorithm Design
l Data Dependence Analysis
– Basics of Data Dependence Analysis
l Parallel Programming Models
– Threaded, Shared-Memory Parallel Programming Model
– Distributed Memory Message-Passing Parallel Programming Model
– Partitioned Global Address Space Parallel Programming Model
– Heterogeneous Parallel Programming Model
- OpenMP Tutorial with Performance Optimizations
l OpenMP Directives and Clauses
– OpenMP를 활용한 병렬화를 위한 directive들 소개
– 각 directive들에 대한 예제 실습
l OpenMP Run Time Library Routines
– OpenMP run time library routine들 소개
– 예제 실습
l Environment Variables
– OpenMP environment variable들 소개
– 예제 실습
l Putting Things Together: Example Parallel Programs Using OpenMP
– 학습한 내용들을 활용하여 간단한 Parallel Program 작성 실습
– Performance optimization technique들을 함께 적용하여 실습
- CUDA Tutorial
l Introduction to CUDA
– Overview
– Parallel processing using CUDA
l Hardware Execution Model
– SIMD execution
– Multithreading
- Memory Hierarchy optimization
– Locality and data placement
– Bandwidth optimization
- Issues with Parallelization
– Data races
– Synchronization
– Branch Handling
– Predicated execution
- Floating-point issues
- Performance optimization
– Instruction throughput optimization
– Memory optimizations
– Latency hiding techniques
- OpenACC Overview
l OpenACC introduction: High-level GPGPU Programming API
l OpenACC APIs
l Performance issues
|
