Performances of optical multi-chip-module interconnects: comparing guided-wave and free-space pathways

Baukens, Valerie; Verschaffelt, Guy; Tuteleers, Patrik; Vynck, Pedro; Ottevaere, Heidi; Kufner, M.; Kufner, Stefan; Veretennicoff, Irina; Bockstaele, Ronny; Hove, An Van; Dhoedt, Bart; Baets, Roel; Thienpont, Hugo

doi:10.1088/1464-4258/1/2/027

Cited by 10 publications

(6 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Notwithstanding these and other types of limiting factors it is important to notice that for this type of free-space interconnection modules all the different performance analyses converge to the same conclusions: interconnect densities of 10 3 /cm 2 at around 1 Gb/s bit rate can be achieved for passively cooled chips, which corresponds to an aggregate throughput capacity of 1Tb/s.cm 2 , while actively cooled systems could even perform aggregate bandwidths of several Tb/s.cm 2 over future inter-and intra-MCM interconnection distances [19][20][21][22]. The performances of both plastic optical fibre wave-guide and free-space optical pathway blocks have previously been compared [20]. As a general conclusion one can say that while both approaches have the potential for Tbits/s.cm 2 fire-hose data handling capacity the rigid free-space module and the more flexible guided-wave approach are complementary in that the first is better suited to bridge distances smaller than a few centimeters while the latter is more appropriate for interconnection lengths of several centimeters and more [20].…”

Section: Figure 3 Channel Density Versus Interconnection Lengthmentioning

confidence: 53%

“…This causes the channel density to drop with increasing interconnection length and limits the total throughput per chip area for an interconnection length beyond a couple of centimeters [20]. Notwithstanding these and other types of limiting factors it is important to notice that for this type of free-space interconnection modules all the different performance analyses converge to the same conclusions: interconnect densities of 10 3 /cm 2 at around 1 Gb/s bit rate can be achieved for passively cooled chips, which corresponds to an aggregate throughput capacity of 1Tb/s.cm 2 , while actively cooled systems could even perform aggregate bandwidths of several Tb/s.cm 2 over future inter-and intra-MCM interconnection distances [19][20][21][22]. The performances of both plastic optical fibre wave-guide and free-space optical pathway blocks have previously been compared [20].…”

Section: Figure 3 Channel Density Versus Interconnection Lengthmentioning

confidence: 99%

“…The performances of both plastic optical fibre wave-guide and free-space optical pathway blocks have previously been compared [20]. As a general conclusion one can say that while both approaches have the potential for Tbits/s.cm 2 fire-hose data handling capacity the rigid free-space module and the more flexible guided-wave approach are complementary in that the first is better suited to bridge distances smaller than a few centimeters while the latter is more appropriate for interconnection lengths of several centimeters and more [20].…”

Section: Figure 3 Channel Density Versus Interconnection Lengthmentioning

confidence: 99%

See 2 more Smart Citations

Plastic micro-optical modules for high-throughput interchip optical interconnections

Thienpont

Baukens

Ottevaere³

et al. 2000

SPIE Proceedings

View full text Add to dashboard Cite

We report on the design, the fabrication, the characterization and the demonstration of scalable multi-channel free-space interconnection components with the potential for Tb/s.cm 2 aggregate bit rate capacity over inter-chip interconnection distances. The demonstrator components are fabricated in a high quality optical plastic, PMMA, using an ion-based rapid prototyping technology that we call deep proton lithography. With the presently achieved Gigabit/s data rates for each of the individual 16 channels with a BER smaller than 10 -13 and with inter-channel cross-talk lower than -22dB the module aims at optically interconnecting 2-D opto-electronic VCSEL and receiver arrays, flip-chip mounted on CMOS circuitry. Furthermore, using ray-tracing software and radiometric simulation tools, we perform a sensitivity analysis for misalignment and fabrication errors on these plastic micro-optical modules and we study industrial fabrication and material issues related to the mass-replication of these components through injection-molding techniques. Finally we provide evidence that these components can be mass-fabricated in dedicated, highly-advanced optical plastics at low cost and with the required precision.

show abstract

Section: Figure 3 Channel Density Versus Interconnection Lengthmentioning

confidence: 53%

Section: Figure 3 Channel Density Versus Interconnection Lengthmentioning

confidence: 99%

Section: Figure 3 Channel Density Versus Interconnection Lengthmentioning

confidence: 99%

See 1 more Smart Citation

Plastic micro-optical modules for high-throughput interchip optical interconnections

Thienpont

Baukens

Ottevaere³

et al. 2000

SPIE Proceedings

View full text Add to dashboard Cite

show abstract

“…18 On the research side studies are in progress for the construction of optical backplanes 19,20 and on micro-optical components for interchip and intrachip communications, for example, at Vrije Universiteit Brussel. 21,22…”

Section: E Intrachip and Multichip-module Interconnectsmentioning

confidence: 99%

Architectural approach to the role of optics in monoprocessor and multiprocessor machines

et al. 2000

View full text Add to dashboard Cite

The relevance of introducing optical interconnects (OI's) in monoprocessors and multiprocessors is studied from an architectural point of view. We show that perhaps the major explanation for why optical technologies have nearly been unable to penetrate into computers is that OI's generally do not shorten the memory-access time, which is the most critical issue for today's stored-program machines. In monoprocessors the memory-access time is dominated by the electronic latency of the memory itself. Thus implementing OI's inside the memory hierarchy without changing the memory architecture cannot dramatically improve the global performance. In strongly coupled multiprocessors the node-bypass latency dominates. Therefore the higher the connectivity (possibly with optics), the shorter the path to another node, but the more expensive the network and the more complex the structure of electronic nodes. This relation leaves the choice of the best network open in terms of simplicity and latency reduction. The bottlenecks resulting from and the benefits of implementing OI's are discussed with respect to symmetric multiprocessors, rings, and distributed shared-memory supercomputers.

show abstract

“…There have been a number of studies on FSOI systems, its alignment, tolerancing, and packaging. [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18] In the present study, we approached the problem from a statistical point of view that was focused on the development of a probabilistic relationship between the system performance and tolerance variables in the form of a prediction equation. Such an equation could then serve as the basis for the design, evaluation, and optimization of the FSOI system.…”

Section: Introductionmentioning

confidence: 99%

Misalignment tolerance analysis of free-space optical interconnects via statistical methods

et al. 2002

View full text Add to dashboard Cite

System-level packaging is one of the critical issues that need to be addressed for free space optical interconnections (FSOI) to become useful in desktop systems. The performance of FSOI, e.g., in terms of system bit-error rate, is greatly affected by misalignments in the optical system. Therefore tolerancing, i.e., the ability to analyze and predict the effects of misalignments in the system, is of prime importance to system designers. We introduce an approach in which we study the effects of optical misalignments and other tolerance factors using statistical methods. We use Monte Carlo simulations, design of the experiments, and regression techniques to fit a polynomial equation that expresses the relationship between the system performance and the tolerance factors. This prediction model can be used for design, cost optimization, and quality control purposes. In addition, we perform a sensitivity analysis to determine those tolerance variables that have the greatest effect on system performance.

show abstract

Performances of optical multi-chip-module interconnects: comparing guided-wave and free-space pathways

Abstract: We simulate and compare optical transmission efficiencies, throughputs and interconnection lengths of free-space and POF-based guided-wave multi-chip-module optical interconnection demonstrators for different types of microcavity emitters.

Cited by 10 publications

References 7 publications

Plastic micro-optical modules for high-throughput interchip optical interconnections

Plastic micro-optical modules for high-throughput interchip optical interconnections

Architectural approach to the role of optics in monoprocessor and multiprocessor machines

Misalignment tolerance analysis of free-space optical interconnects via statistical methods

Contact Info

Product

Resources

About