List of Contributors |
|
xv | |
List of Figures |
|
xix | |
List of Tables |
|
xxxv | |
List of Abbreviations |
|
xxxvii | |
Introduction |
|
1 | (10) |
|
|
|
|
Part I: Design, Application and Implementation |
|
|
1 Unified Inter- and Intra-chip Optical Interconnect Networks |
|
|
11 | (30) |
|
|
|
|
|
|
11 | (3) |
|
|
14 | (3) |
|
1.3 Architecture Overview |
|
|
17 | (2) |
|
1.4 Intra-chip Network Design |
|
|
19 | (8) |
|
1.4.1 Data Channel Design |
|
|
21 | (1) |
|
1.4.1.1 Optical network interface |
|
|
21 | (1) |
|
1.4.1.2 Optical transceiver |
|
|
22 | (1) |
|
1.4.1.3 Channel segmentation |
|
|
23 | (1) |
|
|
24 | (1) |
|
1.4.2.1 Channel partition |
|
|
25 | (1) |
|
1.4.2.2 Node agent design |
|
|
26 | (1) |
|
1.5 Inter-chip Network Design |
|
|
27 | (4) |
|
1.5.1 Inter-chip Data Channel |
|
|
28 | (1) |
|
|
29 | (2) |
|
|
31 | (6) |
|
1.6.1 Node Agent Power and Area |
|
|
32 | (2) |
|
|
34 | (1) |
|
|
35 | (2) |
|
|
37 | (1) |
|
|
37 | (4) |
|
2 Design and Optimization of Vertical Interconnections for Multilayer Optical Networks-on-Chip |
|
|
41 | (32) |
|
|
|
|
41 | (3) |
|
2.2 Materials for Photonic Integrated Circuit Fabrication |
|
|
44 | (2) |
|
2.3 Technological Platforms for Multilayer Photonics |
|
|
46 | (3) |
|
2.4 Overview of Vertical Interconnection Schemes |
|
|
49 | (3) |
|
2.5 Isolation between Layers in Vertical Stacked Structures |
|
|
52 | (3) |
|
2.6 Design Space Exploration of Inverse Tapered Couplers |
|
|
55 | (6) |
|
2.7 Design of MMI-based Vertical Links |
|
|
61 | (4) |
|
|
65 | (3) |
|
|
68 | (1) |
|
|
68 | (5) |
|
3 Optical Interconnection Networks: The Need for Low-Latency Controllers |
|
|
73 | (34) |
|
Felipe Gohring de Magalhaes |
|
|
|
|
|
|
|
74 | (1) |
|
|
75 | (5) |
|
|
76 | (2) |
|
|
78 | (1) |
|
3.2.3 Wavelength Division |
|
|
79 | (1) |
|
3.3 Low-Latency Controlling Solution for OINs |
|
|
80 | (7) |
|
|
87 | (10) |
|
|
97 | (5) |
|
|
102 | (1) |
|
|
102 | (5) |
|
4 Interconnects and Data System Throughput |
|
|
107 | (30) |
|
|
|
|
|
|
107 | (2) |
|
|
109 | (5) |
|
4.2.1 Electrical and Optical Interconnections |
|
|
110 | (1) |
|
4.2.1.1 Electrical interconnections |
|
|
111 | (1) |
|
4.2.1.2 Optical interconnections |
|
|
113 | (1) |
|
4.3 Photonic Transceivers |
|
|
114 | (3) |
|
|
117 | (5) |
|
|
117 | (1) |
|
4.4.2 Folded Clos (Leaf-Spine) Architecture |
|
|
117 | (2) |
|
4.4.3 Hierarchy of Interconnection |
|
|
119 | (1) |
|
4.4.4 Electrical Switching |
|
|
120 | (1) |
|
4.4.5 Electrical versus Optical |
|
|
120 | (2) |
|
4.5 Modeling of Optically Connected Data Centers |
|
|
122 | (4) |
|
4.5.1 Data Center Modeling |
|
|
123 | (1) |
|
4.5.2 Optics in Data Center Simulation |
|
|
124 | (2) |
|
|
126 | (2) |
|
|
128 | (1) |
|
|
128 | (1) |
|
|
129 | (8) |
Part II: Developing Design Automation Solutions and Enabling Design Exploration |
|
|
5 Design Automation Beyond Its Electronic Roots: Toward a Synthesis Methodology for Wavelength-Routed Optical Networks-on-Chip |
|
|
137 | (34) |
|
|
|
|
|
|
|
|
|
|
137 | (2) |
|
5.2 Analogy with EDA Flows |
|
|
139 | (3) |
|
5.3 Wavelength-Selective Routing |
|
|
142 | (3) |
|
5.4 WRONoC Synthesis Methodology at a Glance |
|
|
145 | (1) |
|
5.5 Front-End Synthesis Methodology |
|
|
146 | (4) |
|
5.5.1 Wavelength Resolution |
|
|
147 | (1) |
|
|
148 | (1) |
|
5.5.3 Symbolic Wavelength Assignment |
|
|
149 | (1) |
|
5.5.4 Topology Connection |
|
|
150 | (1) |
|
5.6 Device Parameter Selection |
|
|
150 | (6) |
|
5.6.1 The Routing Fault Concern |
|
|
151 | (2) |
|
5.6.2 The Role of Parameter Uncertainty |
|
|
153 | (2) |
|
5.6.3 Problem Formulation: A Case Study |
|
|
155 | (1) |
|
5.7 Physical Mapping Flow |
|
|
156 | (3) |
|
|
159 | (7) |
|
5.8.1 Synthesis of Logical Topologies |
|
|
159 | (2) |
|
5.8.2 The Design Predictability Gap |
|
|
161 | (2) |
|
5.8.3 Bounds on Connectivity and Parallelism |
|
|
163 | (3) |
|
|
166 | (1) |
|
|
166 | (5) |
|
6 Application-Specific Mapping Optimizations for Photonic Networks-on-Chip |
|
|
171 | (20) |
|
|
|
|
|
171 | (1) |
|
6.2 Motivation: Application-Specific Mapping Optimization |
|
|
172 | (2) |
|
6.3 Architecture Description |
|
|
174 | (6) |
|
6.3.1 Architecture Overview |
|
|
175 | (1) |
|
6.3.2 The Photonic Switch Element Model |
|
|
176 | (2) |
|
|
178 | (2) |
|
|
180 | (5) |
|
6.4.1 Problem Formulation |
|
|
180 | (1) |
|
|
181 | (4) |
|
|
185 | (2) |
|
|
187 | (1) |
|
|
188 | (3) |
|
7 Integrated Photonics for Chip-Multiprocessor Architectures |
|
|
191 | (36) |
|
|
|
|
191 | (1) |
|
7.2 Tiled Architectures and Networks on Chip |
|
|
192 | (3) |
|
7.2.1 Limits of Electronic Wires and Benefits of Silicon Photonics from the Architectural Viewpoint |
|
|
193 | (2) |
|
|
195 | (2) |
|
7.4 Experimental Methodology |
|
|
197 | (2) |
|
|
197 | (1) |
|
|
198 | (1) |
|
7.5 Passive Optical Networks for CMPs |
|
|
199 | (7) |
|
7.5.1 Analyzed Architecture |
|
|
200 | (1) |
|
7.5.2 Arbitration Strategies and Physical Layout Implications |
|
|
201 | (2) |
|
|
203 | (3) |
|
7.6 Optical Dynamic Reconfigurable Networks for CMPs |
|
|
206 | (9) |
|
7.6.1 Background on Circuit Switched Optical NoCs |
|
|
206 | (1) |
|
7.6.2 Limitations of Existing Solutions |
|
|
207 | (3) |
|
7.6.3 Analyzed Architecture |
|
|
210 | (1) |
|
7.6.3.1 Logical operating scheme |
|
|
211 | (2) |
|
|
213 | (2) |
|
|
215 | (1) |
|
|
216 | (11) |
Part III: Challenges in Performance Analysis and Design Solutions |
|
|
8 Thermal Management of Silicon Photonic NoCs in Many-core Systems |
|
|
227 | (22) |
|
|
|
|
|
|
228 | (1) |
|
8.2 Thermal Sensitivity of Optical Devices in PNoCs |
|
|
229 | (3) |
|
8.3 Design Methods for Thermal Management in PNoCs |
|
|
232 | (7) |
|
8.3.1 Device-level Techniques |
|
|
232 | (2) |
|
8.3.2 Chip-level Techniques |
|
|
234 | (5) |
|
8.4 Runtime Methods for Thermal Management in PNoCs |
|
|
239 | (5) |
|
|
244 | (1) |
|
|
245 | (1) |
|
|
245 | (4) |
|
9 Thermal-Aware Design Method for On-Chip Laser-based Optical Interconnect |
|
|
249 | (32) |
|
|
|
|
|
|
|
|
249 | (2) |
|
|
251 | (2) |
|
|
253 | (5) |
|
9.3.1 Architecture Overview |
|
|
253 | (1) |
|
9.3.2 ONoC Interface and Thermal Sensitivity |
|
|
254 | (1) |
|
9.3.3 CMOS-Compatible On-Chip Lasers |
|
|
255 | (3) |
|
|
258 | (1) |
|
9.4 Proposed Design Methodology |
|
|
258 | (8) |
|
9.4.1 Design Methodology Overview |
|
|
258 | (1) |
|
|
259 | (2) |
|
|
261 | (1) |
|
|
261 | (1) |
|
9.4.3.2 Signal attenuation and crosstalk |
|
|
261 | (1) |
|
9.4.3.3 Transmission principles of MR |
|
|
263 | (3) |
|
|
266 | (5) |
|
9.5.1 System Specification |
|
|
266 | (1) |
|
|
266 | (1) |
|
|
267 | (2) |
|
9.5.2 SNR Analysis in the Considered Architecture |
|
|
269 | (2) |
|
9.5.3 Thermal Characteristics of On-Chip Laser |
|
|
271 | (1) |
|
|
271 | (5) |
|
9.6.1 Reduction of the ONI Gradient Temperature |
|
|
272 | (2) |
|
9.6.2 System Level Estimation of SNR |
|
|
274 | (2) |
|
|
276 | (1) |
|
|
276 | (1) |
|
|
276 | (5) |
|
10 Fault-tolerant Photonic Network-on-Chip |
|
|
281 | (38) |
|
|
|
|
281 | (7) |
|
|
282 | (2) |
|
|
284 | (1) |
|
10.1.2.1 PNoC signal strength |
|
|
284 | (1) |
|
10.1.2.2 Electrostatic discharge |
|
|
284 | (1) |
|
|
285 | (1) |
|
|
285 | (1) |
|
10.1.2.5 Process variability |
|
|
286 | (1) |
|
10.1.2.6 Temperature variation |
|
|
286 | (2) |
|
10.2 Fault-tolerant Photonic Network-on-Chip Architecture |
|
|
288 | (12) |
|
10.2.1 Microring Fault-resilient Photonic Router |
|
|
288 | (6) |
|
10.2.2 Light-weight Electronic Control Router |
|
|
294 | (1) |
|
10.2.3 Fault-tolerant Path-configuration and Routing |
|
|
294 | (6) |
|
|
300 | (7) |
|
10.3.1 Complexity Evaluation |
|
|
301 | (1) |
|
10.3.2 Latency and Bandwidth Evaluation |
|
|
302 | (3) |
|
|
305 | (2) |
|
|
307 | (3) |
|
10.5 Chapter Summary and Discussion |
|
|
310 | (1) |
|
|
311 | (8) |
|
11 Techniques for Energy Proportionality in Optical Interconnects |
|
|
319 | (36) |
|
|
|
|
320 | (15) |
|
11.1.1 Why Lasers Waste Power |
|
|
320 | (1) |
|
11.1.2 The Solution: Laser Power-Gating |
|
|
320 | (1) |
|
|
321 | (1) |
|
|
321 | (1) |
|
11.1.3.2 Nanophotonic interconnect topologies |
|
|
323 | (1) |
|
11.1.4 Laser Control Schemes |
|
|
323 | (1) |
|
11.1.4.1 Proactive laser control |
|
|
324 | (1) |
|
11.1.4.2 Controlling an off-chip laser source |
|
|
326 | (1) |
|
11.1.5 Experimental Methodology |
|
|
327 | (1) |
|
11.1.5.1 Interconnect modeling |
|
|
327 | (1) |
|
11.1.5.2 Modeling optical and electrical multicore NoC |
|
|
328 | (1) |
|
11.1.5.3 Laser power modeling |
|
|
329 | (1) |
|
11.1.5.4 Resonant ring heater modeling |
|
|
331 | (1) |
|
11.1.5.5 The overheads of laser control |
|
|
331 | (1) |
|
11.1.6 Experimental Results |
|
|
332 | (1) |
|
11.1.7 Case Study 1: Radix-16 R-SWMR |
|
|
333 | (1) |
|
11.1.8 Case Study 2: Firefly |
|
|
333 | (1) |
|
|
334 | (1) |
|
11.2 Minimizing Ring Trimming Power |
|
|
335 | (8) |
|
|
335 | (1) |
|
11.2.2 Solution: Photonic Die Insulation with Parka |
|
|
336 | (1) |
|
11.2.3 Experimental Methodology |
|
|
337 | (1) |
|
11.2.3.1 Modeling the Ring-Heater Power Consumption |
|
|
337 | (1) |
|
11.2.3.2 Modeling cooling solutions |
|
|
328 | (1) |
|
11.2.4 Experimental Results |
|
|
328 | (1) |
|
11.2.4.1 Impact on ring-heating power consumption |
|
|
338 | (1) |
|
11.2.4.2 Impact on a realistic multicore |
|
|
341 | (2) |
|
|
343 | (1) |
|
|
343 | (1) |
|
|
344 | (1) |
|
|
344 | (11) |
Part IV: On the Impact of Fabrication Non-Uniformity |
|
|
12 Impact of Fabrication Non-Uniformity on Silicon Photonic Networks-on-Chip |
|
|
355 | (30) |
|
|
|
|
|
|
356 | (2) |
|
12.2 Optical Waveguides and MR-Based Devices |
|
|
358 | (7) |
|
12.2.1 Strip Waveguides under Process Variations |
|
|
359 | (2) |
|
12.2.2 MR-Based Add-Drop Filters and Switches under Process Variations |
|
|
361 | (1) |
|
12.2.2.1 Resonant wavelength shift in MRs |
|
|
362 | (1) |
|
12.2.3 Optical Spectra of MRs under Process Variations |
|
|
363 | (2) |
|
12.3 Optical Networks-on-Chip under Process Variations |
|
|
365 | (4) |
|
12.3.1 Process Variations in Optical Switches |
|
|
365 | (2) |
|
12.3.2 Process Variations at the System-Level in ONoCs |
|
|
367 | (2) |
|
12.4 Quantitative Simulation Results and Evaluations |
|
|
369 | (8) |
|
12.4.1 Simulation Results at the Component- and Device-level |
|
|
370 | (5) |
|
12.4.2 Simulation Results at the System-level |
|
|
375 | (2) |
|
12.5 Chip Fabrication and Measurement Results |
|
|
377 | (4) |
|
|
381 | (1) |
|
|
381 | (4) |
|
13 Enhancing Process Variation Resilience in Photonic NoC Architectures |
|
|
385 | (22) |
|
Sai Vineel Reddy Chittamuru |
|
|
|
|
|
385 | (2) |
|
|
387 | (2) |
|
13.3 Analytical Model for PV-Aware Crosstalk Analysis |
|
|
389 | (6) |
|
13.3.1 Impact of Localized Trimming on Crosstalk |
|
|
389 | (2) |
|
13.3.2 Crosstalk Modeling for Corona PNoC |
|
|
391 | (3) |
|
13.3.3 Modeling PV of MR Devices in Corona PNoC |
|
|
394 | (1) |
|
13.4 Double-Bit Crosstalk Mitigation (DBCTM) |
|
|
395 | (3) |
|
|
395 | (2) |
|
13.4.2 DBCTM Sensitivity Analysis with Corona PNoC |
|
|
397 | (1) |
|
|
398 | (6) |
|
13.5.1 Experimental setup |
|
|
398 | (1) |
|
13.5.2 Experimental Results with Corona PNoC |
|
|
399 | (5) |
|
|
404 | (1) |
|
|
404 | (3) |
Index |
|
407 | (4) |
About the Editors |
|
411 | |