Accelerating Depthwise Separable Convolutions on Ultra-Low-Power Devices

June 18, 2024 ยท Declared Dead ยท ๐Ÿ› International Conference / Workshop on Embedded Computer Systems: Architectures, Modeling and Simulation

๐Ÿ‘ป CAUSE OF DEATH: Ghosted
No code link whatsoever

"No code URL or promise found in abstract"

Evidence collected by the PWNC Scanner

Authors Francesco Daghero, Alessio Burrello, Massimo Poncino, Enrico Macii, Daniele Jahier Pagliari arXiv ID 2406.12478 Category cs.LG: Machine Learning Cross-listed cs.DC Citations 1 Venue International Conference / Workshop on Embedded Computer Systems: Architectures, Modeling and Simulation Last Checked 4 months ago
Abstract
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%.
Community shame:
Not yet rated
Community Contributions

Found the code? Know the venue? Think something is wrong? Let us know!

๐Ÿ“œ Similar Papers

In the same crypt โ€” Machine Learning

Died the same way โ€” ๐Ÿ‘ป Ghosted