ACIMS: Analog CIM Simulator for DNN Resilience
Analog Computing In Memory (ACIM) combines the advantages of both Compute In Memory (CIM) and analog computing, making it suitable for the design of energy-efficient hardware accelerators for computationally intensive DNN applications. However, their use will introduce hardware faults that decrease...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-03-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/10/6/686 |
| _version_ | 1797541226384195584 |
|---|---|
| author | Dong Ding Lei Wang Zhijie Yang Kai Hu Hongjun He |
| author_facet | Dong Ding Lei Wang Zhijie Yang Kai Hu Hongjun He |
| author_sort | Dong Ding |
| collection | DOAJ |
| description | Analog Computing In Memory (ACIM) combines the advantages of both Compute In Memory (CIM) and analog computing, making it suitable for the design of energy-efficient hardware accelerators for computationally intensive DNN applications. However, their use will introduce hardware faults that decrease the accuracy of DNN. In this work, we take Sandwich-Ram as the real hardware example of ACIM and are the first to propose a fault injection and fault-aware training framework for it, named Analog Computing In Memory Simulator (ACIMS). Using this framework, we can simulate and repair the hardware faults of ACIM. The experimental results show that ACIMS can recover 91.0%, 93.7% and 89.8% of the DNN’s accuracy drop through retraining on the MNIST, SVHN and Cifar-10 datasets, respectively; moreover, their adjusted accuracy can reach 97.0%, 95.3% and 92.4%. |
| first_indexed | 2024-03-10T13:13:11Z |
| format | Article |
| id | doaj.art-1427d66b58404ebd903f240fe8d0f0dd |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-10T13:13:11Z |
| publishDate | 2021-03-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-1427d66b58404ebd903f240fe8d0f0dd2023-11-21T10:35:20ZengMDPI AGElectronics2079-92922021-03-0110668610.3390/electronics10060686ACIMS: Analog CIM Simulator for DNN ResilienceDong Ding0Lei Wang1Zhijie Yang2Kai Hu3Hongjun He4College of Computer Science and Technology, National University of Defense Technology, Changsha 410000, ChinaCollege of Computer Science and Technology, National University of Defense Technology, Changsha 410000, ChinaCollege of Computer Science and Technology, National University of Defense Technology, Changsha 410000, ChinaCollege of Computer Science and Technology, National University of Defense Technology, Changsha 410000, ChinaCollege of Computer Science and Technology, National University of Defense Technology, Changsha 410000, ChinaAnalog Computing In Memory (ACIM) combines the advantages of both Compute In Memory (CIM) and analog computing, making it suitable for the design of energy-efficient hardware accelerators for computationally intensive DNN applications. However, their use will introduce hardware faults that decrease the accuracy of DNN. In this work, we take Sandwich-Ram as the real hardware example of ACIM and are the first to propose a fault injection and fault-aware training framework for it, named Analog Computing In Memory Simulator (ACIMS). Using this framework, we can simulate and repair the hardware faults of ACIM. The experimental results show that ACIMS can recover 91.0%, 93.7% and 89.8% of the DNN’s accuracy drop through retraining on the MNIST, SVHN and Cifar-10 datasets, respectively; moreover, their adjusted accuracy can reach 97.0%, 95.3% and 92.4%.https://www.mdpi.com/2079-9292/10/6/686DNN acceleratorcompute in memoryanalog computingfault resilience |
| spellingShingle | Dong Ding Lei Wang Zhijie Yang Kai Hu Hongjun He ACIMS: Analog CIM Simulator for DNN Resilience Electronics DNN accelerator compute in memory analog computing fault resilience |
| title | ACIMS: Analog CIM Simulator for DNN Resilience |
| title_full | ACIMS: Analog CIM Simulator for DNN Resilience |
| title_fullStr | ACIMS: Analog CIM Simulator for DNN Resilience |
| title_full_unstemmed | ACIMS: Analog CIM Simulator for DNN Resilience |
| title_short | ACIMS: Analog CIM Simulator for DNN Resilience |
| title_sort | acims analog cim simulator for dnn resilience |
| topic | DNN accelerator compute in memory analog computing fault resilience |
| url | https://www.mdpi.com/2079-9292/10/6/686 |
| work_keys_str_mv | AT dongding acimsanalogcimsimulatorfordnnresilience AT leiwang acimsanalogcimsimulatorfordnnresilience AT zhijieyang acimsanalogcimsimulatorfordnnresilience AT kaihu acimsanalogcimsimulatorfordnnresilience AT hongjunhe acimsanalogcimsimulatorfordnnresilience |