Modular array computer with optical intercell communications pathways

Herein, related art may be discussed to put the invention in context. Related art labeled “prior art” is admitted prior art; related art not labeled “prior art” is not admitted prior art.

Computer purchases for large enterprises can be challenging as computing needs are difficult to predict and tend to change, e.g., grow, over time. As computing power tends to become less expensive overtime, purchasing enough hardware for long-term future needs is usually not economical. Expandable systems are widely available, but are still costly as provisions for expansion, e.g., slots and electrical connectivity typically must be included in the hardware with the initial acquisition. The present invention addresses the problem allowing an expandable system without requiring a pre-existing interconnect structure for modular components. Other features of the invention are apparent in the detailed description below with reference to the following drawings.

The following drawings are of embodiments of the invention and not of the invention itself.

FIG. 1 is a schematic diagram of a modular computing array in accordance with an embodiment of the invention.

FIG. 2 is a schematic illustration of an optical pathway of the array of FIG. 1 .

FIG. 3 is a flow chart of a method in accordance with an embodiment of the invention.

The present invention provides for arrays of arbitrary numbers of cells in which adjacent cells communicate optically. In FIG. 1 , a 3×3 section of a larger two-dimensional square array AP 1 is shown with cells C 11 -C 33 arranged within respective square slots of a sheet metal frame 11 . Cells C 11 -C 33 are not in contact with each other, and are electrically isolated from each other. Each cell C 11 has a functional component 13 and an array interface component 15 .

The functional component can be a PC board with integrated circuits and other elements mounted thereon; for expository purposes a processor 17 is shown in FIG. 1 . The exact nature of the components depends on the function: e.g., computation, disk storage, main memory, routing, input/output interfacing. The array interface components 15 are essentially identical. Each includes a sheet-metal sheath 21 , flexible PC board 23 , optical transmitters (semiconductor lasers) 25 and optical detectors 27 mounted thereon. The flexible PC board 23 is folded to define the square cross section apparent in FIG. 1 . Sheath 21 holds the shape of each flexible PC board 23 . Each array communications component 15 has a “tongue” 31 with an interface pad 33 for interfacing with a complementary interface pad 35 of the functional component 13 . Each sheath 21 has leaf springs 37 for guiding insertion of a cell into frame 11 . Apertures 39 in sheet metal frame 11 allow optical communications between adjacent cells.

Each communications component 15 has four radial sides ST, SR, SB, SL, a front end EF, and a rear end ER. The ends EF and ER are open for ventilation: the tongues and functional components protrude from rear end ER for access. Each of the four radial sides ST, SR, SB, and SL includes optical communications elements. As shown each side has a four-by-four array of photo-transmitters 25 and a four-by-four array of detectors 27 . The side SR facing the reader at the bottom of FIG. 1 has the detector array toward front end EF and the transmitter array toward the respective rear end ER, as does the adjacent side ST above. The side SB shown at the very bottom of FIG. 1 has a complementary arrangement, as does the hidden radial side SL. All cells in array AP 1 have two adjacent sides with detectors in front and two adjacent sides with emitters in front. This permits detectors to oppose emitters and vice versa throughout array AP 1 .

The emitters and detectors are photoelectric elements surface mounted onto the flexible PC boards 23 . Mounted on each emitter 25 and each detector 27 is a respective plastic light pipe 45 , 47 , as shown in FIG. 2 . An opposing emitter-detector pair defines an optical pathway 50 that extends through a respective aperture 39 through sheet metal frame 11 . The light pipes for an optical intercell pathway 50 are separated by an air gap 49 , which can be less than 1 millimeter (mm) apart, although longer distances can be used as well. In an alternative embodiment, there is no air gap between light pipes. In general, the spacing between the emitter and detector of a pair should be less than 10 centimeters for efficient packing of the array and to minimize the sensitivity, power, and transparency requirements for the optical elements.

Since the optical paths are short, the light pipes can be plastic (e.g., instead of glass). The light pipes can be molded. For example, the sixteen light pipes for a detector or emitter array can be molded together and mounted monolithically over the corresponding photoelectric array.

Control of the emitters and access to the detectors is had through electrical connections in the flexible PC boards, which are in turn accessed by the functional component of a cell through the mated connection pads. The emitters can be Vertical-Cavity Surface-Emitting Lasers (VCSELs) for 15.5 gigabit-per-second conductivity with milliamps of drive current. Alternatively, other types of semiconductor lasers or light-emitting diodes can be used. The frame and shrouds obviously have apertures so that the detectors can receive what the emitters transmit.

A method ME 1 in accordance with an embodiment of the invention is flow-charted in FIG. 3 . At method segment M 1 , computing cells are arranged in an array. At method segment M 2 , adjacent cells in the array communicate with each other via optical pathways. Method segment M 2 includes method subsegments, beginning with generating an electrical signal at method segment M 21 , emitting an optical signal at method subsegment M 22 , guiding the signal using an optical guide at method subsegment M 23 , having the signal cross an air gap at method subsegment M 24 , guiding the light using another optical guide at method subsegment M 25 , converting the optical signal to an electrical signal at method segment M 26 , and interpreting the electrical signal at method segment M 27 . In method ME 1 , the distance traveled by the optical signal is less than 1 centimeter, but in other embodiments the distance can be up to 10 centimeters.

While the illustrated embodiment is a two-dimensional square array with nominally identical optical arrangements, the invention provides for many alternatives. In the case of a square array, “square” refers only to the nature of the array, e.g., an arrangement in rows and columns. The cells need not collectively define a square, as some of the array positions may not be available or may not be populated. The invention provides for other two-dimensional array geometries, e.g., hexagonal and triangular arrays. In the case of a triangular, there would be different optical arrangements for different cells. In general, any self-packing cell geometry can be employed. The invention provides for one-dimensional arrays as a special case of two-dimensional arrays.

The invention provides for three-dimensional arrays. For example, two two-dimensional arrays can be arranged front-to-front where optical arrays adorn the fronts cells such as those in FIG. 1 . By using a frame for power delivery, cubic and other three-dimensional arrays are also provided for. These and other variations upon and modifications to the illustrated embodiment are provided for by the present invention, the scope of which is defined in the following claims.