A Survey of Silicon Photonics for Energy-Efficient Manycore Computing

Pasricha, Sudeep; Nikdast, Mahdi

doi:10.1109/mdat.2020.2982628

Cited by 43 publications

(22 citation statements)

References 129 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Although APDs with electronic CMOS circuitry are used in short-reach interconnections, APDs without additional amplification stages are very compelling because of promising energy savings [2,4,5]. We further elaborate on that by analyzing the energy consumption of Si-Ge-Si photodetectors.…”

Section: Energy Efficiencymentioning

confidence: 99%

See 1 more Smart Citation

Silicon-Germanium Avalanche Receivers With fJ/bit Energy Consumption

Benedikovič¹,

Virot

Aubin

et al. 2022

IEEE J. Select. Topics Quantum Electron.

View full text Add to dashboard Cite

Fast, low-noise and sensitive avalanche photoreceivers are needed for surging short-reach photonic applications. Limitations concerning bandwidth, throughput and energy consumption should be overcome. In this work, we comprehensively study the performance opportunities provided by avalanche p-i-n photodetectors with lateral silicongermanium-silicon heterojunctions. Our aim is to circumvent the need for chip-bonded electronic amplifiers. In particular, we demonstrate that avalanche photodetectors based on silicongermanium-silicon heterostructures yield reliable opto-electrical performances, with high gain-bandwidth products up to 480 GHz and low effective ionization ratios down to 0.15. Moreover, they improve power sensitivities for high-speed optical signals and have a low energy dissipation of only a few fJ per received information bit. These results pave the way for high-performing receivers for energy-aware data links, in next-generation shortdistance data communications. IndexTerms-Group-IV semiconductors; Integrated photonics; Avalanche photodetectors; Energy consumption; Short-reach communications; CMOS technology I. INTRODUCTIONHE relentless demands of data-hungry devices as in data centers [1], high-performance computers and servers [2] or big data clouds and storages [3] places exponential requirements on electrical wire signaling. Data overflow in metal wires is a critical bottleneck that hampers the take-off of such devices. The benefits of optical interconnects extend from long-distance to short-distance systems. Components used in long-haul systems are too costly and complex to be adopted in short-reach systems, however. Short-reach architectures comprise opto-electrical and electro-optical Manuscript received MAY 31, 2021.

show abstract

Section: Energy Efficiencymentioning

confidence: 99%

“…HE relentless demands of data-hungry devices as in data centers [1], high-performance computers and servers [2] or big data clouds and storages [3] places exponential requirements on electrical wire signaling. Data overflow in metal wires is a critical bottleneck that hampers the take-off of such devices.…”

Section: Introductionmentioning

confidence: 99%

Silicon-Germanium Avalanche Receivers With fJ/bit Energy Consumption

Benedikovič¹,

Virot

Aubin

et al. 2022

IEEE J. Select. Topics Quantum Electron.

View full text Add to dashboard Cite

show abstract

“…In ( 5), to represent a column vector of activations (A) and to represent a row vector of weights (W). The resulting vector is a summation of dot products of vector elements (6). Much like with CONV layers, these can be decomposed into lower dimensional dot products.…”

Section: C1 Decomposing Vector Operations In Conv/fc Layersmentioning

confidence: 99%

“…Table II shows the optoelectronic parameters considered for this simulation-based analysis. We considered photonic signal losses due to various factors: signal propagation (1 dB/cm [6]), splitter loss (0.13 dB [27]), combiner loss (0.9 dB [28]), MR through loss (0.02 dB [29]), MR modulation loss (0.72 dB [30]), microdisk loss (1.22 dB [31]), EO tuning loss (6 dB/cm [20]), and TO tuning loss (1 dB/cm [17]). We also considered the 1-to-56-Gb/s ADC/DAC-based transceivers from recent work [37].…”

Section: A Simulation Setupmentioning

confidence: 99%

“…Silicon photonics is a promising technology to enable ultra-high bandwidth, lowlatency, and energy-efficient communication solutions [5]. CMOScompatible photonic interconnects have already replaced metallic ones for light-speed data transmission at almost every level of computing, and are now actively being considered for chip-scale integration [6].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

CrossLight: A Cross-Layer Optimized Silicon Photonic Neural Network Accelerator

Sunny

Mirza

Nikdast

et al. 2021

2021 58th ACM/IEEE Design Automation Conference (DAC)

Self Cite

View full text Add to dashboard Cite

Domain-specific neural network accelerators have seen growing interest in recent years due to their improved energy efficiency and inference performance compared to CPUs and GPUs. In this paper, we propose a novel cross-layer optimized neural network accelerator called CrossLight that leverages silicon photonics. CrossLight includes device-level engineering for resilience to process variations and thermal crosstalk, circuit-level tuning enhancements for inference latency reduction, and architecture-level optimization to enable higher resolution, better energy-efficiency, and improved throughput. On average, CrossLight offers 9.5× lower energy-per-bit and 15.9× higher performance-per-watt at 16-bit resolution than state-of-the-art photonic deep learning accelerators.

show abstract

Silicon Photonics for Future Computing Systems

Shafiee

Pasricha

Nikdast

2022

Wiley Encyclopedia of Electrical and Electronics Engineering

View full text Add to dashboard Cite

The primary goal of this article is to provide an overview of silicon photonics technology and its applications in the design and improvement of current and future computing systems. We start by reviewing silicon photonics technology and introducing some of its benefits and challenges as well as providing some background on it. Next, we introduce some fundamental silicon photonic components in the design of silicon photonic integrated circuits (PICs) and optical interconnect for computing systems as well as their operating principles and applications. These components can be active, such as photodetectors and optical modulators, or passive, such as silicon‐on‐insulator (SOI) waveguides. Subsequently, we discuss the application of silicon photonics to improve the communication and computation infrastructure in future computing systems, while reviewing the state‐of‐the‐art and some design and implementation challenges. Finally, we discuss several research opportunities to push forward the application of silicon photonics in the design of future computing systems.

show abstract

A Survey of Silicon Photonics for Energy-Efficient Manycore Computing

Cited by 43 publications

References 129 publications

Silicon-Germanium Avalanche Receivers With fJ/bit Energy Consumption

Silicon-Germanium Avalanche Receivers With fJ/bit Energy Consumption

CrossLight: A Cross-Layer Optimized Silicon Photonic Neural Network Accelerator

Silicon Photonics for Future Computing Systems

Contact Info

Product

Resources

About