Artificial Intelligence Hardware Design. Albert Chun-Chen Liu
sparse compression.Figure 8.11 Cambricon‐X buffer controller architecture.Figure 8.12 Cambricon‐X index module architecture.Figure 8.13 Cambricon‐X direct indexing architecture.Figure 8.14 Cambricon‐X step indexing architecture.Figure 8.15 Cambricon‐X timing performance comparison [4].Figure 8.16 Cambricon‐X energy efficiency comparison [4].Figure 8.17 SCNN convolution.Figure 8.18 SCNN convolution nested loop.Figure 8.19 PT‐IS‐CP‐dense dataflow.Figure 8.20 SCNN architecture.Figure 8.21 SCNN dataflow.Figure 8.22 SCNN weight compression.Figure 8.23 SCNN timing performance comparison [5].Figure 8.24 SCNN energy efficiency comparison [5].Figure 8.25 SeerNet architecture.Figure 8.26 SeerNet Q‐ReLU and Q‐max‐pooling.Figure 8.27 SeerNet quantization.Figure 8.28 SeerNet sparsity‐mask encoding.
9 Chapter 9Figure 9.1 2.5D interposer architecture.Figure 9.2 3D stacked architecture.Figure 9.3 3D‐IC PDN configuration (pyramid shape).Figure 9.4 PDN – Conventional PDN Manthan geometry.Figure 9.5 Novel PDN X topology.Figure 9.6 3D network bridge.Figure 9.7 Neural network layer multiple nodes connection.Figure 9.8 3D network switch.Figure 9.9 3D network bridge segmentation.Figure 9.10 Multiple‐channel bidirectional high‐speed link.Figure 9.11 Power switch configuration.Figure 9.12 3D neural processing power gating approach.Figure 9.13 3D neural processing clock gating approach.
Guide
6 Preface
11 Appendix A Neural Network Topology
12 Index
13 Wiley End User License Agreement
Pages
1 ii
2 iii
3 iv
4 xi
5 xiii
6 xiv
7 xv
8 xvii
9 xviii
10 xix
11 xx
12 xxi
13 xxii
14 1
15 2
16 3
17 4
18 5
19 6
20 7
21 8
22 9
23 10
24 11
25 12
26 13
27 14
28 15
29 16
30 17
31 18
32 19
33 20
34 21
35 22
36 23
37 24
38 25
39 26
40 27
41 28
42 29
43 30
44 31
45 32
46 33
47 34
48 35
49 36
50 37
51 38
52 39
53 40
54 41
55 42
56 43
57 44
58 45
59 46
60 47
61 48
62 49
63