Optical interconnect for high speed processors

Details Optical interconnect for high speed processors Optical interconnect for high speed processors

1999-10-29

2001-06-05

Negash, Kinfe-Michael (Department: 2733)

Optical: systems and elements

Deflection using a moving element

Using a periodically moving element

C359S199200, C359S199200, C359S199200

Reexamination Certificate

active

06243180

ABSTRACT:

BACKGROUND OF THE INVENTION
This invention relates to an optical free space interconnect for ultra-high speed single instruction multiple data processors.
DESCRIPTION OF THE RELEVANT ART
As the geometries of VLSI grow smaller and denser electronic interconnects and heat dissipation have been recognized as bottlenecks of advanced electronic computing systems. F. E. Kiamilev and et al., PERFORMANCE COMPARISON BETWEEN OPTOELECTRONIC AND VLSI MULTISTAGE INTERCONNECTION NETWORKS, J. Lightwave Tech., vol. 9, pp. 1674-1692, 1991. M. R. Feldman, S. C. Esener, C. C. Guest, and S. H. Lee, COMPARISON BETWEEN OPTICAL AND ELECTRIC INTERCONNECTS BASED ON POWER AND SPEED CONSIDERATIONS, Appl, Opt., vol. 27, pp. 1742-1751, 1988. Furthermore, as systems are operated at higher and higher speeds, the latency induced by electronic connections becomes a limiting factor. Although some new techniques, such as three dimensional multi-chip modules, have been developed to provide short connection distances and less latency, the basic limitation of the pin-out problem in electronic connections cannot be fully removed. L. D. Hutchson and P. Haugen, OPTICAL INTERCONNECTS REPLACE HARDWIRE, IEEE Spectrum., pp. 30-35, 1987.
Optical interconnections, because of their three-dimensioinal (3-D) processing capabilities and matched impedance characteristic, have been considered as the best alternative to electronic interconnections. Optical implementations of chip-to-chip and backplane-to-backplane interconnections have been reported. See, e.g. J. W. Goodman, OPTICAL INTERCONNECTIONS FOR VLSI SYSTEMS, Proc. IEEE, vol. 72, p. 850, 1984. Optical 3-D multi-stage interconnection networks have been investigated and realized. A. A. Sawchuk Proc. SPIE, vol. 813, p. 547, 1987. Recently, a new free space optical interconnect based on ring topologies was proposed by Y. Li, B. Ha, T. W. Wang, A. Katz, X. J. Lu, and E. Kanterakis Appl, Opt., vol. 31, p. 5548, 1992.
Arranging the input array on a ring, this novel architecture is capable of interconnecting many processors with identical latency and minimal complexity, and cost. This architecture is best suitable for the implementation of Single Instruction Multiple Data (SIMD) stream machines. Most topologies developed for rectangular array, such as Nearest Neighbor (NN), Plus Minus 2I (PM2I), and Hypercube can be implemented using this ring topology. A simplified architecture of the ring topology architecture is depicted in FIG.
1
. The optical interconnect has an input ring array
42
and an output ring array
45
. The optical interconnect uses a first plurality of beamsplitters
23
,
24
,
25
,
26
, connected to the input ring array
42
. The first plurality of beamsplitters
23
,
24
,
25
,
26
is connected through a plurality of optical switches
27
,
28
,
29
,
30
, a plurality of dove prisms
56
,
57
,
58
,
59
to a second plurality of beamsplitters
61
,
62
,
63
,
54
. A multiple channel system is used to perform a certain interconnection, i.e. a set of permutations. The number of channels depends on the topology to be realized. For example, a realize the NN type of interconnect shown in
FIG. 2
, a four channel system is needed.
SUMMARY OF THE INVENTION
A general object of the invention is an optical interconnect for connecting multiple data processors.
Another object of the invention is an optical interconnect having high speed.
According to the present invention, as embodied and broadly described herein, an optical interconnect is provided for use with a controller and a plurality of single-instruction-multiple-data (SIMD) processors. The input to the optical interconnect uses a modulator system which may include a plurality of laser diodes, a first plurality of optical switching devices, a fiber combiner, an optical fiber, a fiber splitter and a second plurality of optical-switching devices. Each of the plurality of laser diodes transmits light at a different wavelength. Each of the first plurality of optical-switching devices is connected, respectively, to a laser diode. The fiber combiner is connected to the first plurality of optical-switching devices. An optical fiber is connected to the fiber combiner for carrying the output light from the fiber combiner. The fiber splitter is connected to the optical fiber. The fiber splitter divides the output light into a plurality of equal length paths, with each path corresponding to each of the plurality of SIMD processors. The second plurality of optical-switching devices is connected to a plurality of outputs, respectively, of the fiber splitter, for modulating the output light.
At the input of the optical interconnect, a plurality of optical fibers are connected to the second plurality of optical-switching devices, respectively. The plurality of optical fibers form an input ring at the input plane of the optical interconnect.
The optical interconnect includes a plurality of wavelength selective holographic-optical elements, a plurality of dove prisms, and a plurality of beamsplitters. The wavelength selecting holographic-optical elements is referred to hereinafter as a holographic-optical elements. The plurality of holographic-optical elements are coupled to the input ring. Each holographic-optical element reflects only one light wavelength and is transparent to other light wavelengths. A plurality of dove prisms is coupled to the plurality of holographic-optical elements. The first plurality of optical-switching devices select, at a given time, an optical channel. The input light propagates through the dove prism of the selected channel. Each dove prism performs a given, fixed, permutation, i.e., interconnection between processors, by rotating the image formed by the input ring array. The set of permutations obtained by all channels constitute the overall set of desired interconnections for the architecture to be realized. A plurality of beamsplitters is coupled to the plurality of dove prisms, respectively, for reflecting the selected optical channel to an output-ring detector array. The output-ring detector array is connected to the plurality of SIMD processors.
Additional objects and advantages of the invention are set forth in part in the description which follows, and in part are obvious from the description, or may be learned by practice of the invention. The objects and advantages of the invention also may be realized and attained by means of the instrumentalities and combinations particularly pointed out in the appended claims.


REFERENCES:
patent: 4222115 (1980-09-01), Cooper et al.
patent: 4386414 (1983-05-01), Case
patent: 4765753 (1988-08-01), Schmidt
patent: 4779235 (1988-10-01), Kubota
patent: 4986640 (1991-01-01), Athale
patent: 5101501 (1992-03-01), Gilhousen et al.
patent: 5128959 (1992-07-01), Bruckert
patent: 5140466 (1992-08-01), Parker
patent: 5153757 (1992-10-01), Cloonan et al.
patent: 5159473 (1992-10-01), Feldman
patent: 5164913 (1992-11-01), Guilfoyle et al.
patent: 5241610 (1993-08-01), Labiche et al.
patent: 5291324 (1994-03-01), Hinterlong
patent: 5450224 (1995-09-01), Johansson
patent: 5488735 (1996-01-01), Tanabe et al.
patent: 5870216 (1999-02-01), Brock et al.
patent: 5943150 (1999-08-01), Deri et al.
Li et al., “Ring-Array Processor Distribution Topology for Optical Interconnects”, Applied Optics, vol. 31, No. 26, (1992) pp. 5548-5558.
F.E. Kiamiley et al., “Performance Comparison Between Optoelectronic and VLSI Multistage Interconnection Networks”, J. Lightwave Tech., vol. 9, No. 12 (Dec. 1991) pp. 1674-1692.
L.D. Hutcheson and P. Haugen, “Optical Interconnects Replace Hardwire”, IEEE Spectrum (Mar. 1987), pp. 30-35.
J.W. Goodman, et al., “Optical Interconnections For VLSI Systems”, Proc. IEEE, vol. 72, No. 7 (Jul. 1984), pp. 850-866.
L. Thylen, “Integrated Optics in LiNbO3: Recent Developments in Devices for Telecommunications”, IEEE, J. Lightwave Tech., vol. 6, No. 6 (Jun. 1988), pp. 847-861.
N. Kagelnik, “Coupled Wave Theory for Thick Hologram Gratings”, Bell System Tech. Journal, vol. 48, No. 9 (Nov. 1969), pp. 2909-2947.
M.R. Feldman, S. C. Esener, C.C. Guest, and

Affiliated with

Also associated with

Rating

Optical interconnect for high speed processors does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Optical interconnect for high speed processors, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Optical interconnect for high speed processors will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2500164