L2R-CIPU: Efficient CNN Computation with Left-to-Right Composite Inner Product Units
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2024
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866914862365409280 |
|---|---|
| author | Nisar, Malik Zohaib Ibrahim, Mohammad Sohail Usman, Muhammad Lee, Jeong-A |
| author_facet | Nisar, Malik Zohaib Ibrahim, Mohammad Sohail Usman, Muhammad Lee, Jeong-A |
| contents | This paper proposes a composite inner-product computation unit based on left-to-right (LR) arithmetic for the acceleration of convolution neural networks (CNN) on hardware. The efficacy of the proposed L2R-CIPU method has been shown on the VGG-16 network, and assessment is done on various performance metrics. The L2R-CIPU design achieves 1.06x to 6.22x greater performance, 4.8x to 15x more TOPS/W, and 4.51x to 53.45x higher TOPS/mm2 than prior architectures. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_00360 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | L2R-CIPU: Efficient CNN Computation with Left-to-Right Composite Inner Product Units Nisar, Malik Zohaib Ibrahim, Mohammad Sohail Usman, Muhammad Lee, Jeong-A Hardware Architecture This paper proposes a composite inner-product computation unit based on left-to-right (LR) arithmetic for the acceleration of convolution neural networks (CNN) on hardware. The efficacy of the proposed L2R-CIPU method has been shown on the VGG-16 network, and assessment is done on various performance metrics. The L2R-CIPU design achieves 1.06x to 6.22x greater performance, 4.8x to 15x more TOPS/W, and 4.51x to 53.45x higher TOPS/mm2 than prior architectures. |
| title | L2R-CIPU: Efficient CNN Computation with Left-to-Right Composite Inner Product Units |
| topic | Hardware Architecture |
| url | https://arxiv.org/abs/2406.00360 |