Accelerating Depthwise Separable Convolutions on Ultra-Low-Power Devices
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866911924081393664 |
|---|---|
| author | Daghero, Francesco Burrello, Alessio Poncino, Massimo Macii, Enrico Pagliari, Daniele Jahier |
| author_facet | Daghero, Francesco Burrello, Alessio Poncino, Massimo Macii, Enrico Pagliari, Daniele Jahier |
| contents | Depthwise separable convolutions are a fundamental component in efficient Deep Neural Networks, as they reduce the number of parameters and operations compared to traditional convolutions while maintaining comparable accuracy. However, their low data reuse opportunities make deploying them notoriously difficult. In this work, we perform an extensive exploration of alternatives to fuse the depthwise and pointwise kernels that constitute the separable convolutional block. Our approach aims to minimize time-consuming memory transfers by combining different data layouts. When targeting a commercial ultra-low-power device with a three-level memory hierarchy, the GreenWaves GAP8 SoC, we reduce the latency of end-to-end network execution by up to 11.40%. Furthermore, our kernels reduce activation data movements between L2 and L1 memories by up to 52.97%. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2406_12478 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Accelerating Depthwise Separable Convolutions on Ultra-Low-Power Devices Daghero, Francesco Burrello, Alessio Poncino, Massimo Macii, Enrico Pagliari, Daniele Jahier Machine Learning Distributed, Parallel, and Cluster Computing Depthwise separable convolutions are a fundamental component in efficient Deep Neural Networks, as they reduce the number of parameters and operations compared to traditional convolutions while maintaining comparable accuracy. However, their low data reuse opportunities make deploying them notoriously difficult. In this work, we perform an extensive exploration of alternatives to fuse the depthwise and pointwise kernels that constitute the separable convolutional block. Our approach aims to minimize time-consuming memory transfers by combining different data layouts. When targeting a commercial ultra-low-power device with a three-level memory hierarchy, the GreenWaves GAP8 SoC, we reduce the latency of end-to-end network execution by up to 11.40%. Furthermore, our kernels reduce activation data movements between L2 and L1 memories by up to 52.97%. |
| title | Accelerating Depthwise Separable Convolutions on Ultra-Low-Power Devices |
| topic | Machine Learning Distributed, Parallel, and Cluster Computing |
| url | https://arxiv.org/abs/2406.12478 |