IMCA: An Efficient In-Memory Convolution Accelerator

Traditional convolutional neural network (CNN) architectures suffer from two bottlenecks: computational complexity and memory access cost. In this study, an efficient in-memory convolution accelerator (IMCA) is proposed based on associative in-memory processing to alleviate these two problems direct...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on very large scale integration (VLSI) systems 2021-03, Vol.29 (3), p.447-460
Main Authors: Yantir, Hasan Erdem, Eltawil, Ahmed M., Salama, Khaled N.
Format: Article
Language:English
Subjects:
Artificial intelligence
Artificial intelligence (AI)
Artificial neural networks
Associative memory
associative processors
Computer architecture
Convolution
convolution accelerator
convolutional neural networks (CNNs)
in-memory computing
Memory management
Multiplication
non-von Neumann architectures
Registers
Table lookup
Task analysis
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Traditional convolutional neural network (CNN) architectures suffer from two bottlenecks: computational complexity and memory access cost. In this study, an efficient in-memory convolution accelerator (IMCA) is proposed based on associative in-memory processing to alleviate these two problems directly. In the IMCA, the convolution operations are directly performed inside the memory as in-place operations. The proposed memory computational structure allows for a significant improvement in computational metrics, namely, TOPS/W. Furthermore, due to its unconventional computation style, the IMCA can take advantage of many potential opportunities, such as constant multiplication, bit-level sparsity, and dynamic approximate computing, which, while supported by traditional architectures, require extra overhead to exploit, thus reducing any potential gains. The proposed accelerator architecture exhibits a significant efficiency in terms of area and performance, achieving around 0.65 GOPS and 1.64 TOPS/W at 16-bit fixed-point precision with an area less than 0.25 mm 2 .
ISSN:1063-8210
1557-9999
DOI:10.1109/TVLSI.2020.3047641