Sharing and placement of on-chip laser sources in silicon-photonic NoCs

Chen, Chao; Zhang, Tiansheng; Contu, Pietro; Klamkin, Jonathan; Coskun, Ayse K.; Joshi, Ajay

doi:10.1109/nocs.2014.7008766

Cited by 28 publications

(28 citation statements)

References 36 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Given that all laser sources output the same wavelengths, there is further potential in our design for laser source sharing. Recent studies showed that sharing a laser source over up to 16 waveguides is beneficial in terms of laser power [31], before splitter loss starts to neutralize the benefits again. Moreover, placing laser sources on the edges of the chipwhere temperature fluctuations are less frequent [32]decreases the impact of temperature on the laser efficiency and, in turn, laser power, and does not constitute a major imposition on path loss at current waveguide loss parameters of 0.1dB/mm [21].…”

Section: Vlsi Layout Of the Data Networkmentioning

confidence: 99%

Efficient Sharing of Optical Resources in Low-Power Optical Networks-on-Chip

Werner¹,

Navaridas²,

Luján³

2017

J. Opt. Commun. Netw.

View full text Add to dashboard Cite

Abstract-With the ever-growing core counts in modern computing systems, NoCs consume an increasing part of the power budget due to bandwidth and power density limitations of electrical interconnects. To maintain performance and power scaling, alternative technologies are required, with silicon photonics being a promising candidate thanks to high-bandwidth, low-energy data transmission. In order to get the best of silicon photonics, sophisticated network designs are required to minimize static power overheads. In this paper, we propose Amon, a low-power ONoC that decreases number of µRings, wavelengths and path losses to reduce power consumption. Amon performs destination checking prior to data transmission on an underlying control network, allowing the sharing of optical bandwidth. Compared to a wide range of state-of-the-art optical, hybrid, and electrical NoCs, Amon improves Throughputper-Watt by at least 23% (up to 70%), while reducing power without latency overheads on both synthetic and realistic applications. For aggressive optical technology parameters, Amon considerably outperforms all alternative NoCs in terms of power, outlining its increasing superiority as technology matures.

show abstract

Section: Vlsi Layout Of the Data Networkmentioning

confidence: 99%

Efficient Sharing of Optical Resources in Low-Power Optical Networks-on-Chip

Werner¹,

Navaridas²,

Luján³

2017

J. Opt. Commun. Netw.

View full text Add to dashboard Cite

show abstract

“…6 b). Another benefit of tuning the receiver is allowing the sharing of onchip lasers [27]. In summary, for optical links using on-chip DFB lasers, we could tune either the laser output power or the receiver sensitivity.…”

Section: Adaptive Tuning Approachmentioning

confidence: 99%

“…For a multi-receiver link, its hardware cost can be amortized by sharing the BER monitor, as the multiple receivers on one link cannot receive signal simultaneously. In this way, for a 64-cluster crossbar with 64 WDM channels, the total area cost of the BER monitors is only 1.0% of the chip area (366.1 mm 2 in [27]). The tuning power overhead is avoided at runtime as the BER monitor circuit is switched off after the tuning process is complete.…”

Section: Adaptive Tuning Approachmentioning

confidence: 99%

“…Other than specified for parameter sweeps, the number of clusters is set to 16; the maximum temptation variation is set to 17 • C [13]; the laser type is on-chip DFB laser. The length of the longest communication path is set to 4 cm to accommodate the chip area (366.1 mm 2 in [27]). The waveguide loss is assumed to be 0.74 dB/cm [32].…”

Section: A Experimental Setupmentioning

confidence: 99%

See 1 more Smart Citation

Variation-aware adaptive tuning for nanophotonic interconnects

Wu¹,

Chen²,

Li³

et al. 2015

2015 IEEE/ACM International Conference on Computer-Aided Design (ICCAD)

View full text Add to dashboard Cite

Short-reach nanophotonic interconnects are promising to solve the communication bottleneck in data centers and chip-level scenarios. However, the nanophotonic interconnects are sensitive to process and thermal variations, especially for the microring structures, resulting in significant variation of an optical link's bit error rate (BER). In this paper, we propose a power-efficient adaptive tuning approach for nanophotonic interconnects to address the variation issues. During the adaptive tuning process, each nanophotonic interconnect is adaptively allocated just enough power to meet the BER requirement. The proposed adaptive tuning approach could reduce the photonic receiver power by 8% -34% than the worst-case based fixed design while achieving the same BER. Our evaluation results show that the adaptive tuning approach scales well with the process variation, the thermal variation and the number of communication nodes, and can accommodate different types of NoC architectures and lasers.

show abstract

“…The signal then needs to be fed to a photodetector. An optical station typically corresponds to a set of cores (2)(3)(4)(5)(6)(7)(8). The individual cores are connected by electrical links to the optical stations.…”

Section: Introductionmentioning

confidence: 99%

Optimal Power Efficient Photonic SWMR Buses

Peter

Sarangi

2015

2015 Workshop on Exploiting Silicon Photonics for Energy-Efficient High Performance Computing

View full text Add to dashboard Cite

In this paper we solve the problem of minimizing the power consumption in SWMR (single writer, multi reader) optical buses for both the ring and tree based configurations. To minimize the power consumption it is necessary to design the beam splitters in a SWMR bus optimally. A prior approach by Binzhang et al. [5] proposes to use an exponential time NLP based solution for solving a related problem, and there are other approaches for solving this problem sub-optimally. In comparison we propose an optimal linear time algorithm for both ring and tree based buses. Additionally, we take into account the fact that the power loss in a splitter is dependent on the split ratio. Subsequently, we propose a hardware version of our algorithm that takes 4 cycles for a 64-station ring, and 32 cycles for a 64-station balanced tree to compute an optimal configuration. With the help of exhaustive Synopsys RSoft simulations, we show that our SWMR bus is 87% more power efficient than a popularly used alternative approach proposed by Kurian et al. [9]. For large number of stations, an optimally designed tree based bus is a further 25% more power efficient than a ring.

show abstract

Sharing and placement of on-chip laser sources in silicon-photonic NoCs

Cited by 28 publications

References 36 publications

Efficient Sharing of Optical Resources in Low-Power Optical Networks-on-Chip

Efficient Sharing of Optical Resources in Low-Power Optical Networks-on-Chip

Variation-aware adaptive tuning for nanophotonic interconnects

Optimal Power Efficient Photonic SWMR Buses

Contact Info

Product

Resources

About