GPGCN: A General-Purpose Graph Convolution Neural Network Accelerator Based on RISC-V ISA Extension
In the past two years, various graph convolution neural networks (GCNs) accelerators have emerged, each with their own characteristics, but their common disadvantage is that the hardware architecture is not programmable and it is optimized for a specific network and dataset. They may not support acc...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-11-01
|
| Series: | Electronics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2079-9292/11/22/3833 |
| _version_ | 1797465441420967936 |
|---|---|
| author | Wenkai Tang Peiyong Zhang |
| author_facet | Wenkai Tang Peiyong Zhang |
| author_sort | Wenkai Tang |
| collection | DOAJ |
| description | In the past two years, various graph convolution neural networks (GCNs) accelerators have emerged, each with their own characteristics, but their common disadvantage is that the hardware architecture is not programmable and it is optimized for a specific network and dataset. They may not support acceleration for different GCNs and may not achieve optimal hardware resource utilization for datasets of different sizes. Therefore, given the above shortcomings, and according to the development trend of traditional neural network accelerators, this paper proposes and implements GPGCN: a general-purpose GCNs accelerator architecture based on RISC-V instruction set extension, providing the software programming freedom to support acceleration for various GCNs, and achieving the best acceleration efficiency for different GCNs with different datasets. Compared with traditional CPU, and traditional CPU with vector expansion, GPGCN achieves above 1001×, 267× speedup for GCN with the Cora dataset. Compared with dedicated accelerators, GPGCN has software programmability and supports the acceleration of more GCNs. |
| first_indexed | 2024-03-09T18:22:28Z |
| format | Article |
| id | doaj.art-6aeccbf88bfc4fd1bdc15be88225d05f |
| institution | Directory Open Access Journal |
| issn | 2079-9292 |
| language | English |
| last_indexed | 2024-03-09T18:22:28Z |
| publishDate | 2022-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Electronics |
| spelling | doaj.art-6aeccbf88bfc4fd1bdc15be88225d05f2023-11-24T08:11:15ZengMDPI AGElectronics2079-92922022-11-011122383310.3390/electronics11223833GPGCN: A General-Purpose Graph Convolution Neural Network Accelerator Based on RISC-V ISA ExtensionWenkai Tang0Peiyong Zhang1School of Micro-Nano Electronics, Zhejiang University, Hangzhou 310058, ChinaSchool of Micro-Nano Electronics, Zhejiang University, Hangzhou 310058, ChinaIn the past two years, various graph convolution neural networks (GCNs) accelerators have emerged, each with their own characteristics, but their common disadvantage is that the hardware architecture is not programmable and it is optimized for a specific network and dataset. They may not support acceleration for different GCNs and may not achieve optimal hardware resource utilization for datasets of different sizes. Therefore, given the above shortcomings, and according to the development trend of traditional neural network accelerators, this paper proposes and implements GPGCN: a general-purpose GCNs accelerator architecture based on RISC-V instruction set extension, providing the software programming freedom to support acceleration for various GCNs, and achieving the best acceleration efficiency for different GCNs with different datasets. Compared with traditional CPU, and traditional CPU with vector expansion, GPGCN achieves above 1001×, 267× speedup for GCN with the Cora dataset. Compared with dedicated accelerators, GPGCN has software programmability and supports the acceleration of more GCNs.https://www.mdpi.com/2079-9292/11/22/3833GCNsgeneral GCNs acceleratorRISC-Vsoftware programmable |
| spellingShingle | Wenkai Tang Peiyong Zhang GPGCN: A General-Purpose Graph Convolution Neural Network Accelerator Based on RISC-V ISA Extension Electronics GCNs general GCNs accelerator RISC-V software programmable |
| title | GPGCN: A General-Purpose Graph Convolution Neural Network Accelerator Based on RISC-V ISA Extension |
| title_full | GPGCN: A General-Purpose Graph Convolution Neural Network Accelerator Based on RISC-V ISA Extension |
| title_fullStr | GPGCN: A General-Purpose Graph Convolution Neural Network Accelerator Based on RISC-V ISA Extension |
| title_full_unstemmed | GPGCN: A General-Purpose Graph Convolution Neural Network Accelerator Based on RISC-V ISA Extension |
| title_short | GPGCN: A General-Purpose Graph Convolution Neural Network Accelerator Based on RISC-V ISA Extension |
| title_sort | gpgcn a general purpose graph convolution neural network accelerator based on risc v isa extension |
| topic | GCNs general GCNs accelerator RISC-V software programmable |
| url | https://www.mdpi.com/2079-9292/11/22/3833 |
| work_keys_str_mv | AT wenkaitang gpgcnageneralpurposegraphconvolutionneuralnetworkacceleratorbasedonriscvisaextension AT peiyongzhang gpgcnageneralpurposegraphconvolutionneuralnetworkacceleratorbasedonriscvisaextension |