Title | ||
---|---|---|
Edge Inference Engine for Deep & Random Sparse Neural Networks with 4-bit Cartesian-Product MAC Array and Pipelined Activation Aligner |
Abstract | ||
---|---|---|
A 4b-quantized convolutional neural network (CNN) inference engine for edge-AI is presented featuring a Cartesian-product MAC array and pipelined activation aligners targeting deep-/random-pruned models. A 40nm prototype with 32x32 MACs and 5Mb SRAM runs at 534 MHz, 1.07 TOPS, 352 mW at 1.1V, and attains 5.30 dense TOPS/W, 234 MHz at 0.8V. Sparse TOPS/W reaches 26.5 when running a randomly pruned ... |
Year | DOI | Venue |
---|---|---|
2021 | 10.1109/HCS52781.2021.9567328 | 2021 IEEE Hot Chips 33 Symposium (HCS) |
Keywords | DocType | ISBN |
Training,Neural networks,Random access memory,Prototypes,Inference algorithms,Convolutional neural networks,Engines | Conference | 978-1-6654-1397-8 |
Citations | PageRank | References |
1 | 0.37 | 0 |
Authors | ||
7 |
Name | Order | Citations | PageRank |
---|---|---|---|
Kota Ando | 1 | 24 | 6.81 |
Jaehoon Yu | 2 | 28 | 22.44 |
Kazutoshi Hirose | 3 | 1 | 1.39 |
Hiroki Nakahara | 4 | 155 | 37.34 |
Kazushi Kawamura | 5 | 3 | 2.58 |
Thiem Van Chu | 6 | 1 | 2.74 |
Masato Motomura | 7 | 8 | 3.65 |