CPU-GPU간 긴밀성을 위한 효율적인 공유메모리 접근 방법과 검증 시스템 구현
Implementation of Integrated CPU-GPU for Efficient Uniform Memory Access Method and Verification System
0
조회
0
다운로드
1
인용
0
참고문헌
키워드
초록
In this paper, we propose a system for efficient use of shared memory between CPU and GPU. The system, called Fusion Architecture, assures consistency of the shared memory and minimizes cache misses that frequently occurs on Heterogeneous System Architecture or Unified Virtual Memory based systems. It also maximizes the performance for memory intensive jobs by efficient allocation of GPU cores. To test between architectures on various scenarios, we introduce the Fusion Architecture Analyzer, which compares OpenMP, OpenCL, CUDA, and the proposed architecture in terms of memory overhead and process time. As a result, Proposed fusion architectures show that the Fusion Architecture runs benchmarks 55% faster and reduces memory overheads by 220% in average.
Abstract
In this paper, we propose a system for efficient use of shared memory between CPU and GPU. The system, called Fusion Architecture, assures consistency of the shared memory and minimizes cache misses that frequently occurs on Heterogeneous System Architecture or Unified Virtual Memory based systems. It also maximizes the performance for memory intensive jobs by efficient allocation of GPU cores. To test between architectures on various scenarios, we introduce the Fusion Architecture Analyzer, which compares OpenMP, OpenCL, CUDA, and the proposed architecture in terms of memory overhead and process time. As a result, Proposed fusion architectures show that the Fusion Architecture runs benchmarks 55% faster and reduces memory overheads by 220% in average.
참고문헌 (13)
- J. Power, "Heteroge-neous system coherence for integrated CPU-GPU systems", Proceedings of the 46th Annual IEEE, 2013, 457~467
- 박용훈, "GPGPU를 이용한 파데브-레브리어 알고리즘 구현 및 성능 분석", 대한임베디드공학회논문지, 8(3), pp.171~178 (2013), 대한임베디드공학회
- 정근용, "스테레오 영상 보행자 인식 시스템의 후보 영역 검출을 위한 GP-GPU 기반의 효율적 구현", 대한임베디드공학회논문지, 8(2), pp.121~128 (2013), 대한임베디드공학회
- S. Che, "Rodinia : A benchmark suite for heterogeneous com-putting", Proceedings of IEEE International Symposium on Workload Characterization, pp.44~54 (2009)
- J. Feehrer, "Coherency hub design for multisocket sun servers with coolthreads technology", IEEE Micro, 29(4), pp.36~47 (2009)
- I. Singh, "Cache coherence for GPU architectures", Proceedings of IEEE 19th International Symposium on High Performance Computer Architecture, pp.578~590 (2013)
- J. Lee, "Many-thread aware prefetching mech anisms for GPGPU applications", Proceedings of 43rd Annual IEEE, 2010, 213~224
- P. Rogers, "Heterogeneous system architecture overview", Proceedings of Hot Chips, 25 (2013)
- P. Hammarlund, "Haswell : The fourth-generation Intel core processor", IEEE Micro, 34(02), pp.6~20 (2014)
- K. Wang, "GDM : Device memory management for gpgpu computing", Proceedings of The 2014 ACM international conference on Measurement and modeling of computer systems, pp.533~545 (2014)
- O. Kayiran, "Managing GPU concurrency in heterogeneous architectures", Proceedings of the 47th Annual IEEE, 2014, 114~126
- Rodinia Benchmark Group, "The Rodinia Benchmark Suite version 3.1", ~ (2015)
- 박현문, "CPU-GPU환경에서 효율적인 메모리공유를 위한 퓨전아키텍처 개발", 한국전자통신학회 논문지, 11(02), pp.151~158 (2016), 한국전자통신학회