Integrated Circuits and Systems >
BASER: Bit-Wise Approximate Compressor Configurable In-SRAM-Computing for Energy-Efficient Neural Network Acceleration With Data-Aware Weight Remapping Method
# Shunqin Cai and Liukai Xu contributed equally to this work
WEIKANG QIAN, (Senior Member, IEEE) |
LIANG CHANG, (Member, IEEE) |
YANAN SUN, (Senior Member, IEEE) |
Received date: 2024-02-28
Accepted date: 2024-05-15
Online published: 2024-11-27
Supported by
National Key R&D Program of China under Grant(2023YFB4502200)
National Natural Science Foundation of China under Grant(62174110)
National Natural Science Foundation of China under Grant(62104025)
Natural Science Foundation of Shanghai under Grant(23ZR1433200)
SRAM-based computing-in-memory (SRAM-CIM) is expected to solve the “Memory Wall” problem. For the digital domain SRAM-CIM, full-precision digital logic has been utilized to achieve high computational accuracy. However, the energy and area efficiency advantages of CIM cannot be fully utilized under error-resilient neural networks (NNs) with given quantization bit-width. Therefore, an all-digital Bit-wise Approximate compressor configurable In-SRAM-computing macro for Energy-efficient NN acceleration, with a data-aware weight Remapping method (BASER), is proposed in this paper. Leveraging the NN error resilience property, six energy-efficient bit-wise compressor configurations are presented under 4b/4b and 3b/3b NN quantization, respectively. Concurrently, a data-aware weight remapping approach is proposed to enhance the NN accuracy without supplementary retraining further. Evaluations of VGG-9 and ResNet-18 on CIFAR-10 and CIFAR-100 datasets show that the proposed BASER achieves 1.35x and 1.29x improvement in energy efficiency, as well as limited accuracy loss and improved NN accuracy, as compared to the previous full-precision and approximate SRAM-CIM design, respectively.
SHUNQIN CAI , LIUKAI XU , DENGFENG WANG , ZHI LI , WEIKANG QIAN , LIANG CHANG , YANAN SUN . BASER: Bit-Wise Approximate Compressor Configurable In-SRAM-Computing for Energy-Efficient Neural Network Acceleration With Data-Aware Weight Remapping Method[J]. Integrated Circuits and Systems, 2024 , 1(2) : 80 -91 . DOI: 10.23919/ICS.2024.3419630
[1] |
|
[2] |
|
[3] |
|
[4] |
|
[5] |
|
[6] |
|
[7] |
|
[8] |
|
[9] |
|
[10] |
|
[11] |
|
[12] |
|
[13] |
|
[14] |
|
[15] |
|
[16] |
|
[17] |
|
[18] |
|
[19] |
|
[20] |
|
[21] |
|
[22] |
|
[23] |
|
[24] |
|
[25] |
|
[26] |
|
[27] |
|
[28] |
|
[29] |
|
[30] |
|
[31] |
|
[32] |
|
[33] |
|
[34] |
|
[35] |
|
[36] |
|
[37] |
|
[38] |
|
[39] |
|
/
〈 |
|
〉 |