EDEA: Efficient Dual-Engine Accelerator for Depthwise Separable Convolution with Direct Data Transfer
IEEE International System-on-Chip Conference (SOCC), pp. 1–6
Abstract
Depthwise separable convolution (DSC) has emerged as a crucial technique, especially for resource-constrained devices. In this paper, we propose a dual-engine for the DSC hardware accelerator, which enables the full utilization of depthwise convolution (DWC) and pointwise convolution (PWC) processing elements (PEs) in all DSC layers. To determine the optimal dataflow, data reuse, and configuration of the target architecture, we conduct a design space exploration using MobileNetV1 with the CIFAR10 dataset. In the architecture, we introduce an additional non-convolutional unit, which merges the dequantization, batch normalization (BN), ReLU, and quantization between DWC and PWC into a simple fixed-point multiplication and addition operation. This also reduces the intermediate data access between the DWC and PWC, enabling streaming operation and reducing latency. The proposed DSC dual-engine accelerator is implemented using the 22nm FDSOI technology from GlobalFoundries, occupying an area of $0.58 \mathbf{~ m m}^{2}$. After signoff, it can operate at 1 GHz at TT corner, achieving a peak energy efficiency of 13.43 TOPS/W with a throughput of 973.55 GOPS with 8-bit precision. The average energy efficiency of all DSC layers on MobileNetV1 is 11.13 TOPS/W, demonstrating substantial hardware efficiency improvements for DSC-based applications.
Authors 5
-
Affiliation as printed
RWTH Aachen University,Chair of Integrated Digital Systems and Circuit Design,Germany
-
Affiliation as printed
RWTH Aachen University,Chair of Integrated Digital Systems and Circuit Design,Germany
-
Affiliation as printed
RWTH Aachen University,Chair of Integrated Digital Systems and Circuit Design,Germany
-
Affiliation as printed
RWTH Aachen University,Chair of Integrated Digital Systems and Circuit Design,Germany
-
Affiliation as printed
RWTH Aachen University,Chair of Integrated Digital Systems and Circuit Design,Germany
Cited by 7 stored of 7
7 results
No patents citing this paper on Lens.org (checked 2026-10-06).
References 19
-
W6737664043details pending0citations
-
W6760069825details pending0citations
-
W2973980443details pending0citations
-
W2989331028details pending0citations
-
W3048234025details pending0citations
-
W3093796669details pending0citations
-
W3162993841details pending0citations
-
W3206196385details pending0citations
-
W3215511279details pending0citations
-
W4285146087details pending0citations
-
W4291910419details pending0citations
-
W4307927141details pending0citations
-
W4388976448details pending0citations
-
W4390776943details pending0citations
-
W4393034313details pending0citations
19 results