MemBox: Shared Memory Device for Memory-Centric Computing Applicable to Deep Learning Problems
Large-scale computational problems that need to be addressed in modern computers, such as deep learning or big data analysis, cannot be solved in a single computer, but can be solved with distributed computer systems. Since most distributed computing systems, consisting of a large number of networke...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-11-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/10/21/2720 |
| _version_ | 1797512667596849152 |
|---|---|
| author | Yongseok Choi Eunji Lim Jaekwon Shin Cheol-Hoon Lee |
| author_facet | Yongseok Choi Eunji Lim Jaekwon Shin Cheol-Hoon Lee |
| author_sort | Yongseok Choi |
| collection | DOAJ |
| description | Large-scale computational problems that need to be addressed in modern computers, such as deep learning or big data analysis, cannot be solved in a single computer, but can be solved with distributed computer systems. Since most distributed computing systems, consisting of a large number of networked computers, should propagate their computational results to each other, they can suffer the problem of an increasing overhead, resulting in lower computational efficiencies. To solve these problems, we proposed an architecture of a distributed system that used a shared memory that is simultaneously accessible by multiple computers. Our architecture aimed to be implemented in FPGA or ASIC. Using an FPGA board that implemented our architecture, we configured the actual distributed system and showed the feasibility of our system. We compared the results of the deep learning application test using our architecture with that using Google Tensorflow’s parameter server mechanism. We showed improvements in our architecture beyond Google Tensorflow’s parameter server mechanism and we determined the future direction of research by deriving the expected problems. |
| first_indexed | 2024-03-10T06:04:54Z |
| format | Article |
| id | doaj.art-b92e0307e18440e6a8a3705ccf35bbbf |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-10T06:04:54Z |
| publishDate | 2021-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-b92e0307e18440e6a8a3705ccf35bbbf2023-11-22T20:39:51ZengMDPI AGElectronics2079-92922021-11-011021272010.3390/electronics10212720MemBox: Shared Memory Device for Memory-Centric Computing Applicable to Deep Learning ProblemsYongseok Choi0Eunji Lim1Jaekwon Shin2Cheol-Hoon Lee3Artificial Intelligence Research Laboratory, ETRI, Daejeon 34129, KoreaArtificial Intelligence Research Laboratory, ETRI, Daejeon 34129, KoreaAviation Drone Laboratory, LIG Nex1, Yongin 16961, KoreaDepartment of Computer Engineering, Chungnam National University, Daejeon 34134, KoreaLarge-scale computational problems that need to be addressed in modern computers, such as deep learning or big data analysis, cannot be solved in a single computer, but can be solved with distributed computer systems. Since most distributed computing systems, consisting of a large number of networked computers, should propagate their computational results to each other, they can suffer the problem of an increasing overhead, resulting in lower computational efficiencies. To solve these problems, we proposed an architecture of a distributed system that used a shared memory that is simultaneously accessible by multiple computers. Our architecture aimed to be implemented in FPGA or ASIC. Using an FPGA board that implemented our architecture, we configured the actual distributed system and showed the feasibility of our system. We compared the results of the deep learning application test using our architecture with that using Google Tensorflow’s parameter server mechanism. We showed improvements in our architecture beyond Google Tensorflow’s parameter server mechanism and we determined the future direction of research by deriving the expected problems.https://www.mdpi.com/2079-9292/10/21/2720distributed systemshared memorydeep learningbig dataFPGAASIC |
| spellingShingle | Yongseok Choi Eunji Lim Jaekwon Shin Cheol-Hoon Lee MemBox: Shared Memory Device for Memory-Centric Computing Applicable to Deep Learning Problems Electronics distributed system shared memory deep learning big data FPGA ASIC |
| title | MemBox: Shared Memory Device for Memory-Centric Computing Applicable to Deep Learning Problems |
| title_full | MemBox: Shared Memory Device for Memory-Centric Computing Applicable to Deep Learning Problems |
| title_fullStr | MemBox: Shared Memory Device for Memory-Centric Computing Applicable to Deep Learning Problems |
| title_full_unstemmed | MemBox: Shared Memory Device for Memory-Centric Computing Applicable to Deep Learning Problems |
| title_short | MemBox: Shared Memory Device for Memory-Centric Computing Applicable to Deep Learning Problems |
| title_sort | membox shared memory device for memory centric computing applicable to deep learning problems |
| topic | distributed system shared memory deep learning big data FPGA ASIC |
| url | https://www.mdpi.com/2079-9292/10/21/2720 |
| work_keys_str_mv | AT yongseokchoi memboxsharedmemorydeviceformemorycentriccomputingapplicabletodeeplearningproblems AT eunjilim memboxsharedmemorydeviceformemorycentriccomputingapplicabletodeeplearningproblems AT jaekwonshin memboxsharedmemorydeviceformemorycentriccomputingapplicabletodeeplearningproblems AT cheolhoonlee memboxsharedmemorydeviceformemorycentriccomputingapplicabletodeeplearningproblems |