Photoelectric hybrid calculation method and array for on-chip large-scale matrix multiplication

Abstract

According to the photoelectric hybrid calculation method and array for on-chip large-scale matrix multiplication disclosed by the invention, current output signals with products of photoelectric hybrid calculation units which are connected in parallel on different rows are summed by adopting an electric bus, so that on one hand, the requirement on the input power of a light source is lower; and 2, the requirements on the wavelengths of the input light in different rows are also reduced. Therefore, the photoelectric hybrid computing array provided by the invention reduces the requirement on the light source to a certain extent, and can quickly realize the integrated application of the light source and the photon computing chip.

Description

[0001] Divisional Application This application is a divisional application of the patent application

CN202310965549.3

Optoelectronic Hybrid Computing Method and Array for Large-Scale Matrix Multiplication Operations on Chip

[0002] The present invention relates to the field of photonic computing technology. Specifically, it particularly relates to an optoelectronic hybrid computing method and array for large-scale matrix multiplication operations on chip. Background Art

[0003] As the semiconductor industry gradually enters the post-Moore era, the development of integrated circuits continues to evolve in different directions. On the one hand, new semiconductor materials are developed, especially carbon nanotubes, two-dimensional semiconductor materials, etc., to continue the essence of Moore's Law and further reduce the size of devices or chips, namely "More Moore"; on the other hand, for specific application fields, new architectures and heterogeneous integrated chips are developed, such as neuromorphic chips, optoelectronic chips, quantum chips, etc., to achieve "More than Moore".

[0004] Among them, photonic chips based on silicon-based optoelectronic technology use materials and processes compatible with integrated circuits to integrate micron- and nanoscale photonic, electronic, and optoelectronic devices on the same silicon substrate to achieve functional integration and complementary advantages of microelectronic and optoelectronic devices, obtaining optoelectronic chips with superior performance. This is an effective way to solve the performance bottleneck and information congestion faced by traditional integrated circuits. Thanks to the mature application of optical fiber communication, photons, as information carriers, have more multiplexing dimensions than electrons, such as amplitude, phase, wavelength, mode, etc., and thus have greater bandwidth, faster speed, and lower energy consumption. Early silicon-based optoelectronic chips were developed to replace copper interconnection technology and solve the interconnection communication bottleneck between the processor cores and memories of microelectronic chips. Microprocessors and memory units are realized by microelectronic devices, while photonic devices mainly complete signal transceiver and information transmission. With the increasing maturity of silicon photonics technology and the great advantages of optical communication, people's attention to silicon photonics chips has gradually shifted from information transmission to information processing, including frontier application fields such as analog computing, quantum computing, and brain-like computing.

[0005] The existing technical paths for realizing photonic computing chips are as follows: Using Mach-Zehnder interferometers (MZIs) or micro-ring resonators (MMRs) to achieve photonic computing, and the computing methods are disclosed in Chinese patents CN115905792 and CN113392965. The computing arrays formed on this basis are disclosed in Chinese patents CN10407644 and CN116107037. However, these technologies generally have the following deficiencies: 1. Calculation array scale limitation: The basic calculation unit of the MZI-based photonic computing chip has a relatively large area. When fabricated on a silicon photonics platform, it is difficult to achieve an application-level calculation array scale with a conventional mask on a regular wafer area.

[0006] 2. Slow calculation speed: Mainly using the thermo-optic tuning principle based on silicon-based optoelectronic technology, and there will be a phenomenon of temperature drift (especially significant in the MMR structure), which affects the refractive index of the optical waveguide.

[0007] 3. High design & packaging difficulty: Each basic calculation unit of the MZI-based photonic computing chip has more than 10 electrodes, with a huge number of wiring, resulting in high design difficulty and packaging complexity.

[0008] The applicant previously proposed a novel photonic computing array, which adopts a crossbar switch matrix architecture; the crossbar switch matrix architecture is formed by the intersection of a group of mutually parallel input traveling waveguides and a group of mutually parallel output column waveguides. At the intersection of the traveling waveguides and the column waveguides, cross waveguides are used to achieve low-loss transmission of optical signals, and on the column waveguides, column waveguide coupling devices are used to sum the outputs of photonic computing units on different rows. Although the novel photonic computing array performs summation through light, in engineering applications, the cross waveguides and column waveguide coupling devices will bring relatively large losses, requiring a relatively high requirement for the entire input light source. From the perspective of the field of photonic computing, there are technical difficulties in currently integrated high-power light sources, and it is impossible to integrate the light source and the chip together in a short time. Therefore, some current photonic computing companies need to use external high-power light sources. Summary of the Invention

[0009] The object of the present invention is to overcome the deficiencies of the prior art and provide an optoelectronic hybrid calculation method and array for large-scale matrix multiplication operations on a chip.

[0010] The object of the present invention is achieved by the following technical solutions: An optoelectronic hybrid calculation method for large-scale matrix multiplication operations on a chip, comprising the following steps: S1. Provide an optoelectronic hybrid calculation unit, the optoelectronic hybrid calculation unit comprising a photonic calculation unit and an optoelectronic conversion device connected to each other, the photonic calculation unit generating an optical output signal with encoded information; the optoelectronic conversion device converts the optical output signal into an electrical current signal, and the magnitude of the electrical current signal is proportional to the power of the optical output signal; S2. Provide an electrical bus, sum the corresponding number of electrical current signals after conversion through a plurality of parallel optoelectronic conversion devices to generate an electrical output signal, and the sum after the addition operation of the electrical current signals is encoded in the electrical output signal.

[0011] Preferably, in step S1, "the photon computing unit generates an optical output signal with encoded information" specifically includes: S11. Provide a write signal and encode the multiplication value into the write signal; S12. Use the write signal to map the multiplication value to the state of the photon computing unit. The photon computing unit includes an optical waveguide and a modulation element optically coupled to the optical waveguide. The modulation element modifies the transmission, reflection, refraction, or absorption characteristics of the optical waveguide according to its own state. Among them, the own state of the modulation element can be adjusted by the write signal; the state is manifested as the absorption coefficient α or the refractive index n of the optical waveguide for light; S13. Encode the multiplicand value into the optical input signal of the photon computing unit; the optical input signal generates the optical output signal after passing through the photon computing unit, where the product of the multiplication value and the multiplicand value is encoded in the optical output signal.

[0012] Preferably, the modulation element is an electro-optic modulator based on the optical absorption effect. The electro-optic modulator uses an electrical signal as an external excitation, and changes its free carrier concentration by injecting current or applying voltage in its doped region to change the absorption coefficient α of the optical waveguide containing free carriers for light, so as to realize multiplication operation on the optical signal passing through the optical waveguide.

[0013] Preferably, the modulation element is a phase change material deposited on the optical waveguide; the phase change material can be selected to change its own state by using an optical signal as an external excitation (i.e., the write signal), and the own state is manifested as modifying the absorption coefficient α of the optical waveguide containing the phase change material for light.

[0014] Alternatively, in step S1, "the photon computing unit generates an optical output signal with encoded information" is implemented by using a Mach-Zehnder interferometer (MZI) or a microring structure (MMR).

[0015] Preferably, the electrical bus is a metal interconnection layer on the photon computing chip.

[0016] The present invention also discloses an optoelectronic hybrid computing array for large-scale matrix multiplication operations on a chip, including: A group of photon computing units, the photon computing unit includes an optical waveguide and a modulation element optically coupled to the optical waveguide. The modulation element modifies the transmission, reflection, refraction, or absorption characteristics of the optical waveguide according to its own state. Among them, the own state of the modulation element can be adjusted by the write signal; the state is manifested as the absorption coefficient α or the refractive index n of the optical waveguide for light; the optical input signal generates the optical output signal after passing through the photon computing unit, where the product of the multiplication value and the multiplicand value is encoded in the optical output signal; A group of optoelectronic conversion devices are respectively and matchingly connected to the output side of the optical waveguide of the photon computing unit, converting the optical output signal into an electric current signal, and the magnitude of the electric current signal is proportional to the power of the optical output signal; Crossbar switch matrix architecture; the crossbar switch matrix architecture is formed by the intersection of a group of mutually parallel input optical waveguides and a group of mutually parallel electric buses; the optical input signal is transmitted through the input optical waveguides; At each intersection of an input optical waveguide and an electric bus in the crossbar switch matrix architecture, there is a photon computing unit. The input side of the optical waveguide of each photon computing unit is coupled to an adjacent input optical waveguide through a waveguide coupling device. The waveguide coupling device evanescently couples a part of the optical power from the adjacent input optical waveguide, and the coupled optical power depends on the length of the part of the waveguide coupling device that is placed adjacent to the input optical waveguide and extends parallel to the adjacent input optical waveguide; A number of parallel optoelectronic conversion devices are connected to each electric bus. The electric bus sums up the corresponding number of electric current signals after conversion through the number of parallel optoelectronic conversion devices, generating an electric output signal, and the sum after the addition operation of the electric current signals is encoded in the electric output signal.

[0017] Preferably, the input optical waveguides and the electric buses are perpendicular to each other.

[0018] Preferably, the optical wavelengths of the optical input signals of the input optical waveguides in different rows are different or the same.

[0019] Preferably, under the action of the waveguide coupling device, the optical power of the optical input signal is evenly distributed to each photon computing unit in the same row.

[0020] Preferably, the number of columns of the input optical waveguides and the electric buses is greater than or equal to 2.

[0021] Preferably, the electric bus is a metal interconnect layer on the photon computing chip.

[0022] The beneficial effects of the present invention are mainly reflected in: (1) The area of the basic computing unit is small, and the mask template under the conventional wafer area can make the scale of the computing array reach the application level; the modulation speed is fast, and the modulation speed can be increased to the nanosecond level through electric pulses, thereby improving the computing performance of the photon computing chip, and at the same time, there will be no phenomenon of temperature drift; each basic computing unit only has 2 electrodes, the number of wirings is small, and the design difficulty and packaging complexity are low; (2) The optoelectronic hybrid computing array proposed by the present invention uses optoelectronic conversion devices and an electrical bus to sum the optical output signals with products of the photon computing units on different rows. First, the requirement for the input power of the light source is lower, and second, the requirement for the wavelengths of the input light on different rows is also reduced. Therefore, the optoelectronic hybrid computing array of the present invention reduces the requirement for the light source to a certain extent and can more quickly realize the integrated application of the light source and the photon computing chip. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The technical solution of the present invention will be further described below with reference to the drawings: Figure 1 : Schematic flow chart of the optoelectronic hybrid computing method of the present invention; Figure 2 : Schematic diagram of a preferred embodiment of the optoelectronic hybrid computing array of the present invention; Figure 3 : Schematic diagram of a 3×3 optoelectronic hybrid computing array of a preferred embodiment of the present invention; Figure 4 : Schematic diagram of a second embodiment of the optoelectronic hybrid computing array of the present invention; Figure 5 : Schematic diagram of a third embodiment of the optoelectronic hybrid computing array of the present invention; Figure 6 : Schematic diagram of the structure of the photon computing unit of a preferred embodiment of the present invention; Figure 7 : Figure 6 Schematic diagram of the working principle of performing multiplication calculation on the photon computing unit of. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0024] The present invention will be described in detail below with reference to the specific embodiments shown in the drawings. However, these embodiments are not limited to the present invention, and any structural, method, or functional transformation made by those of ordinary skill in the art based on these embodiments is included in the protection scope of the present invention.

[0025] The present invention will be described in detail below with reference to the drawings and in conjunction with the embodiments.

[0026] An optoelectronic hybrid computing array provided by the present invention for on-chip large-scale matrix multiplication operation has a specific structure as Figure 2 shown in the preferred embodiment, including an optoelectronic hybrid computing unit, a waveguide coupling device 5, an input optical waveguide 31, and an electrical bus 32. The optoelectronic hybrid computing unit includes a photon computing unit 1 and an optoelectronic conversion device 2.

[0027] The present invention performs multiplication operations using the photon computing unit 1, which includes an optical waveguide 11 and a modulation element 12 optically coupled to the optical waveguide. The modulation element modifies the transmission, reflection, refraction, or absorption characteristics of the optical waveguide according to its own state, where the own state of the modulation element can be adjusted by a write signal; the state is manifested as the absorption coefficient α or the refractive index n of the optical waveguide for light. The optical input signal generates the optical output signal after passing through the photon computing unit, where the product of the multiplier value and the multiplicand value is encoded in the optical output signal.

[0028] In the present invention, the modulation element 12 includes three forms.

[0029] As Figure 2 shown in the preferred embodiment, the modulation element 12(a) is an electro-optic modulator based on the optical absorption effect. The electro-optic modulator uses an electrical signal as an external excitation, injects current or applies a voltage in its doped region to change its free carrier concentration, and changes the absorption coefficient α of the optical waveguide containing free carriers for light, so that the optical signal passing through the optical waveguide is absorbed by free carriers to achieve multiplication operations.

[0030] The specific multiplication operation process is as Figure 6 、 Figure 7 shown and includes the following steps: S11. Provide a write signal and encode the multiplier value into the write signal 108; S12. Use the write signal 108 to map the multiplier value to the state of the doped region of the modulation element 12, and the state is manifested as the absorption coefficient α of the optical waveguide for light at different free carrier concentrations; S13. Encode the multiplicand value into the optical input signal 106 of the photon computing unit; S14. When the optical input signal 106 passes through the optical absorption region of the modulation element 12 through the optical waveguide 11, an optical output signal 107 is generated, where the product of the multiplier value and the multiplicand value is encoded in the optical power of the optical output signal.

[0031] Among them, the write signal 108 for modulating the free carrier concentration in the modulation element 12 is provided by the electrical signal generator 105. Two electrical interconnection devices in the metal layer of the electrical signal generator 105 transmit the generated electrical signal to the doped regions 103a, 103b of the modulation element 12 through the first contact electrode 104a and the second contact electrode 104b respectively. The electrical signal causes the free carriers to move directionally under the action of the electric field, and the parameter change of the electrical signal can change the free carrier concentration in the doped region, thereby changing the light absorption coefficient of the optical waveguide containing free carriers. The electro-optic modulator based on the light absorption effect includes a doped region based on semiconductor doping technologies such as ion implantation or high-temperature diffusion, and a pair of contact electrodes that form an ohmic contact or a Schottky contact with the doped region.

[0032] The electrical signal generator 105 uses an external signal, that is, the write signal 108, to map the multiplication value b to the light absorption coefficient α of the optical waveguide containing free carriers. The modulation element 12 injects current or applies voltage to change the free carrier concentration (charge or hole) in the doped region, thereby changing the light absorption coefficient α of the optical waveguide containing free carriers. The energy Pwrite of the external electrical signal has a mapping in the form of an elementary function or a rational function with the light absorption coefficient α. The optical input signal 106 decays inversely with the light absorption coefficient α of the optical waveguide containing free carriers, thereby generating a multiplication output signal 107 corresponding to the external electrical signal write signal 108 and the optical input signal 106. The output signal Pout of the optical waveguide 11 = α × Pin is the result of mapping the multiplication value b to α and mapping the multiplicand value a to Pin.

[0033] Emerging photonic devices based on phase change materials have been widely studied in recent years. Phase change materials have the characteristics of fast writing and reading speeds (in the nanosecond range), high cycle numbers (>10 12 ), low power consumption, etc., can be compatible with existing CMOS processes, and have low technical implementation difficulty and industrial cost. The optical and electrical properties of phase change materials are quite different between the crystalline state and the amorphous state. Phase change can be induced by various methods such as heat, light, and electricity, and has stable properties. The modulation of light by phase change materials does not require the maintenance of a static bias voltage and can maintain a certain state unchanged at room temperature. Therefore, the theoretical static control energy consumption of phase change materials is zero, which greatly reduces the energy consumption of the system and improves the stability of the system. These characteristics make phase change materials have the potential to become the basic functional materials for photonic computing.

[0034] Such as Figure 4In the second embodiment shown, the modulation element 12(b) uses a phase change material 12' deposited on an optical waveguide; the phase change material 12' uses an optical signal as an external excitation (i.e., a write signal) to change its own state, and the own state is manifested as modifying the light absorption coefficient α of the optical waveguide containing the phase change material 12'. Selecting a phase change material as the modulation element has the advantage that the phase change material is non-volatile and can remain in this state after modulation and power-off, which is extremely suitable for the application scenario of artificial intelligence for inference. Due to the non-volatile characteristic of the phase change material, compared with the technical path of existing photonic computing chips, there is no power consumption for maintenance, and the computing power energy consumption ratio is extremely high.

[0035] Specifically, by transmitting an optical signal as a write signal through the optical waveguide, for example, using a specific high-power optical signal and attaching a specific time, the own state of the phase change material can be changed, that is, it is manifested as modifying the light absorption coefficient α of the optical waveguide containing the phase change material. The optoelectronic hybrid computing array based on the phase change material reduces the requirements for the light source to a certain extent and can achieve the integrated application of the light source and the photonic computing chip faster.

[0036] As Figure 5 In the third embodiment shown, the modulation element (c) is a modulator based on the phase change material. When modulated with an optical signal, there is no doped region. When modulated with an electrical signal, a doped region needs to be included. After adopting this embodiment, an optical signal or an electrical signal can be selectively used as the write signal. The multiplication calculation process using an electrical signal for modulation is similar to the preferred embodiment, and the multiplication operation process using an optical signal for modulation is similar to the second embodiment, so it will not be elaborated here.

[0037] Of course, the multiplication operation of the present invention can also be implemented by using a Mach-Zehnder interferometer (MZI) or a microring structure (MMR) of the prior art.

[0038] The optoelectronic conversion device 2 is matched and connected to the output side of the optical waveguide of the photonic computing unit 1 to convert the optical output signal into an electrical current signal, and the magnitude of the electrical current signal is proportional to the power of the optical output signal. Therefore, the number and position of the optoelectronic conversion devices 2 are set in one-to-one correspondence with the photonic computing unit 1.

[0039] The optoelectronic hybrid computing array of the present invention adopts a crossbar switch matrix architecture; it is formed by crossing a group of input optical waveguides 31 that can be doped and a group of the electrical buses 32. The optical input signal is transmitted through the input optical waveguides 31. The input optical waveguides 31 that can be doped are arranged in parallel rows, and the electrical buses 32 are arranged in parallel columns. Preferably, the input optical waveguides that can be doped are perpendicular to the electrical buses. In the present invention, the number of columns of the input optical waveguides 31 that can be doped and the electrical buses 32 is greater than or equal to 2.

[0040] At each intersection of an input optical waveguide 31 and an electrical bus 32 in the crossbar switch matrix architecture, there is a photon computing unit 1. The input side of the optical waveguide 11 of each photon computing unit 1 is coupled to the adjacent input optical waveguide 31 through a waveguide coupling device 5. The waveguide coupling device 5 evanescently couples a part of the optical power from the adjacent input optical waveguide 31, and the coupled optical power depends on the length of the part of the waveguide coupling device 5 that is placed adjacent to the input optical waveguide and extends parallel to the adjacent input optical waveguide.

[0041] The other side of the optical waveguide 11 is then coupled to the adjacent electrical bus 32 through the optoelectronic conversion device 2. Preferably, the electrical bus 32 adopts a metal interconnect layer on the photon computing chip.

[0042] A number of parallel optoelectronic conversion devices 2 are connected to each electrical bus 32. The electrical bus 32 sums the corresponding number of current signals after being converted by the number of parallel optoelectronic conversion devices 2, generates an electrical output signal, and the sum after the addition operation of the current signals is encoded in the electrical output signal.

[0043] The feature of the present invention is to use a crossbar switch matrix (Crossbar) architecture to implement large-scale matrix multiplication and addition operations. That is: after the multiplication operation by the photon computing unit, the optical output signal with the product is converted into a current signal through the optoelectronic conversion device, and then the sum is carried out on the electrical bus immediately.

[0044] In a preferred embodiment, the wavelengths of the optical input signals of the input traveling waveguides 31 in different rows are different, in order to avoid the interference phenomenon during the addition operation of light and affect the accuracy of photon computing. Since the present invention adopts an optoelectronic hybrid computing architecture based on electrical buses, the wavelengths of the optical input signals of the input traveling waveguides 31 in different rows can be the same, reducing the requirements for the wavelength of the light source.

[0045] Specifically as Figure 1 and Figure 2 shown, the present invention provides a method for performing matrix multiplication and addition operations in the optical domain. Briefly speaking, it includes the following steps: S1. Provide an optoelectronic hybrid computing unit, wherein the optoelectronic hybrid computing unit comprises a photon computing unit and an optoelectronic conversion device connected to each other, wherein the photon computing unit generates an optical output signal with coded information; and the optoelectronic conversion device converts the optical output signal into an electric current signal, wherein the magnitude of the electric current signal is proportional to the power of the optical output signal; S2. Provide an electrical bus, sum a corresponding number of current signals converted through a plurality of parallel-connected photoelectric conversion devices, and generate an electrical output signal, wherein the sum of the current signals after addition is encoded in the electrical output signal.

[0046] This method realizes the calculation of the m×n-order matrix P×U=A: The specific steps include: (1) Encode the weight matrix U into the write signal, for example, input code U11 into the photon calculation unit in the first row and first column, input code U12 into the photon calculation unit in the first row and second column, input code U21 into the photon calculation unit in the second row and first column, and so on; (2) using a write signal to map the multiplication value to the state of the modulation element 12 or 12' in each photon computing unit, wherein the state is represented by the absorption coefficient α of the optical waveguide 11 to light; (3) Encoding the input data matrix P into an optical input signal, and the optical power of the optical input signal is evenly distributed to each unit on the same row of input optical waveguides 31 under the action of the waveguide coupling device 5; for example, the input code P1 is input into the input optical waveguide of the first row, the input code P2 is input into the input optical waveguide of the second row, and so on; (4) After the optical input signal passes through the optical waveguide 11 and the modulation element of the photon computing unit, an optical output signal is generated, wherein the product of the multiplier value and the multiplicand value is encoded in the optical output signal. The optical power of the optical output signal is converted into a current signal by a photoelectric conversion device and then fed into the electrical bus for summation. The magnitude of the current signal is proportional to the power of the optical output signal. The response of the optical output signal converted into a current signal is β. In practical applications, the value of β depends on the process level of the silicon photonic chip manufacturer.

[0047] For example, the power of the optical output signal of the first column is: The common factor in the above formula is β / n. And so on.

[0048] Figure 3 A schematic diagram of a 3*3 optoelectronic hybrid computing array is disclosed to realize the calculation of a third-order matrix P×U=A, Wherein: 101-1, 101-2, 101-3: Input optical waveguides; 102-1, 102-2, 102-3: Electrical buses; P1, P2, P3: Input (read) signals; A1, A2, A3: Electrical output signals; U11, U12, U13, U21, U22, U23, U31, U32, U33: Photonic computing units for multiplication operations; r11, r12, r13, r21, r22, r23, r31, r32, r33: Waveguide coupling devices; d11, d12, d13, d21, d22, d23, d31, d32, d33: Optoelectronic conversion devices; According to the method of the present invention, the optical power output of the first column can be realized as: The common factor in the above formula is β / 3.

[0049] The present invention adopts an optoelectronic hybrid computing array based on electrical bus summation. Compared with a photonic computing array based on optical bus summation, it has no losses caused by cross waveguides and column waveguide coupling devices themselves. At the same time, due to the absence of a cross-transmission structure in the optical path, the requirements for the wavelength of the input light are also lower, and the input channels can be of the same wavelength or different wavelengths.

[0050] The current injection or voltage application in the present invention is mainly in the form of electrical pulses. The duration of the electrical pulses is at the nanosecond level. At the same time, the duration of the electrical pulses determines the modulation speed of the optoelectronic hybrid computing array, and thus determines the computing performance of the optoelectronic hybrid computing array; and it will not cause the phenomenon of temperature drift.

[0051] The series of detailed descriptions listed above are only specific descriptions of the feasible implementation manners of the present invention, and they are not used to limit the protection scope of the present invention. Any equivalent implementation manners or changes made without departing from the technical spirit of the present invention should be included within the protection scope of the present invention.

Claims

1. An optoelectronic hybrid computing method for large-scale matrix multiplication operations on a chip. The optoelectronic hybrid computing method is applied to an optoelectronic hybrid computing array, and the optoelectronic hybrid computing array adopts a crossbar matrix architecture, which is formed by crossing a group of dopable input optical waveguides and a group of electrical buses. The optoelectronic hybrid computing method is used to calculate an m×n order matrix, and is characterized in that: It includes the following steps: S1. Provide an optoelectronic hybrid computing unit, which includes a photon computing unit and an optoelectronic conversion device connected to each other. The photon computing unit generates an optical output signal with encoded information; the optoelectronic conversion device converts the optical output signal into an electrical current signal, and the magnitude of the electrical current signal is proportional to the power of the optical output signal. Correspondingly, the electrical bus is a metal interconnection layer on the photon computing chip; Among them, the specific content of "the photon computing unit generates an optical output signal with encoded information" in step S1 includes: S11. Provide a write signal and encode the multiplication value into the write signal; S12. Use the write signal to map the multiplication value to the state of the photon computing unit. The photon computing unit includes an optical waveguide and a modulation element optically coupled to the optical waveguide. The modulation element modifies the transmission, reflection, refraction or absorption characteristics of the optical waveguide according to its own state. Among them, the own state of the modulation element can be adjusted by the write signal; the state is manifested as the absorption coefficient or refractive index of the optical waveguide to light; S13. Encode the value to be multiplied into the optical input signal of the photon computing unit; the optical input signal generates the optical output signal after passing through the photon computing unit, and the product of the multiplication value and the value to be multiplied is encoded in the optical output signal; S2. Provide an electrical bus, sum the corresponding number of electrical current signals converted after passing through several parallel optoelectronic conversion devices to generate an electrical output signal, and the sum after the addition operation of the electrical current signals is encoded in the electrical output signal; Among them, the summation process includes: ; Among them, A1 is the power of the optical output signal in the first column, β / n is the common factor, P1, P2... P m is the input coding in each row of the input optical waveguide, U 11 , U 21 ... U m1 is the input coding in each row of the photon computing units in the first column.

2. The optoelectronic hybrid computing method according to claim 1, wherein: In step S1, "the photon computing unit generates an optical output signal with encoded information" is implemented by a Mach-Zehnder interferometer (MZI) or a microring structure (MMR).

3. An optoelectronic hybrid computing array for large-scale matrix multiplication operations on a chip, characterized in that: It includes a group of photon computing units. The photon computing unit includes an optical waveguide and a modulation element optically coupled to the optical waveguide. The modulation element modifies the transmission, reflection, refraction or absorption characteristics of the optical waveguide according to its own state. Among them, the own state of the modulation element can be adjusted by the write signal; the state is manifested as the absorption coefficient or refractive index of the optical waveguide to light; the optical input signal generates the optical output signal after passing through the photon computing unit, and the product of the multiplication value and the value to be multiplied is encoded in the optical output signal; A group of optoelectronic conversion devices, which are connected in one-to-one matching with the output side of the optical waveguide of the photon computing unit, convert the optical output signal into an electrical current signal, and the magnitude of the electrical current signal is proportional to the power of the optical output signal; Crossbar switch matrix architecture; the crossbar switch matrix architecture is formed by the intersection of a set of mutually parallel input optical waveguides and a set of mutually parallel electrical buses; the optical input signals are transmitted through the input optical waveguides, and the electrical buses are metal interconnect layers on the photonic computing chip; At each intersection of an input optical waveguide and an electrical bus in the crossbar switch matrix architecture, there is a photonic computing unit. The input side of the optical waveguide of each photonic computing unit is coupled to an adjacent input optical waveguide through a waveguide coupling device. The waveguide coupling device evanescently couples a part of the optical power from the adjacent input optical waveguide, where the coupled optical power depends on the length of the part of the waveguide coupling device that is placed adjacent to and extends parallel to the adjacent input optical waveguide; Several parallel optoelectronic conversion devices are connected to each electrical bus. The electrical bus sums the corresponding number of current signals converted after passing through the several parallel optoelectronic conversion devices to generate an electrical output signal. The sum after the addition operation of the current signals is encoded in the electrical output signal; Among them, the summation process includes: ; Wherein, A1 is the power of the optical output signal of the first column, β / n is a common factor, P1, P2... P m are the input encodings in the input optical waveguides of each row, and U 11 , U 21 ... U m1 are the input encodings in the photon calculation units of each row in the first column.

4. The optoelectronic hybrid computing array according to claim 3, wherein: The input optical waveguides and the electrical buses are perpendicular to each other.

5. The optoelectronic hybrid computing array according to claim 3, wherein: The optical wavelengths of the optical input signals of the input optical waveguides in different rows are different or the same.

6. The optoelectronic hybrid computing array according to claim 3, wherein: Under the action of the waveguide coupling device, the optical power of the optical input signal is evenly distributed to each photonic computing unit in the same row.

7. The optoelectronic hybrid computing array according to claim 3, wherein: The number of columns of the input optical waveguides and the electrical buses is greater than or equal to 2.

Images