A dynamic evaluation system for silicon-based optical interconnect chips based on deep learning

Abstract

The present invention discloses a silicon-based optical interconnect chip dynamic evaluation system based on deep learning, and relates to the technical field of chip evaluation. The system discloses a basic performance analysis module, an optical device screening module, an optical device evaluation module, and a dynamic evaluation module. The optical device screening module, the optical device evaluation module, and the dynamic evaluation module are set. After the optical device screening work is started, by constructing a simulation model of a silicon-based optical interconnect chip, the possible performance of each optical device of the silicon-based optical interconnect chip under the current performance parameters can be simulated, and the test evaluation requirements of each optical device are comprehensively analyzed. According to the dynamic screening of the test evaluation requirements and the test evaluation order of the optical devices, it is no longer necessary to test and evaluate each optical device, and the abnormal optical devices in the silicon-based optical interconnect chip can be quickly determined, thereby improving the test evaluation accuracy and efficiency of the optical devices.

Description

Technical Field

[0001] The present invention relates to the field of chip evaluation technology, and more specifically, to a deep learning-based dynamic evaluation system for silicon-based optical interconnect chips. Background Art

[0002] With the rapid development of information technology, silicon-based optical interconnect chips are becoming increasingly important as core components of high-performance computing, data centers, and high-speed communication systems. Silicon-based optical interconnect chips achieve efficient conversion and transmission between optical and electrical signals by integrating photonic devices and electronic devices, greatly improving data transmission rates and system bandwidth. However, with the continuous improvement of chip integration and increasingly stringent performance requirements, the design, manufacturing, and test evaluation of silicon-based optical interconnect chips face unprecedented challenges.

[0003] Traditional evaluation methods for silicon-based optical interconnect chips mainly rely on static testing and manual analysis. This method is not only time-consuming and labor-intensive, but also difficult to fully and accurately reflect the performance of the chip under actual working conditions. The performance parameters of silicon-based optical interconnect chips can only reflect the current performance of the silicon-based optical interconnect chips, and it is impossible to determine which optical devices in the silicon-based optical interconnect chips are abnormal. Therefore, it is often necessary to test each optical device one by one, which is not only inefficient, but also difficult to quickly locate the problem, thus affecting the product development cycle and market competitiveness.

[0004] Therefore, a deep learning-based dynamic evaluation system for silicon-based optical interconnect chips is developed. Summary of the invention

[0005] In view of the shortcomings of the prior art, the object of the present invention is to provide a dynamic evaluation system for silicon-based optical interconnect chips based on deep learning.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A deep learning-based dynamic evaluation system for silicon-based optical interconnect chips, including a basic performance analysis module, an optical device screening module, an optical device evaluation module, and a dynamic evaluation module;

[0008] The basic performance analysis module is used to periodically obtain the basic performance index of the silicon-based optical interconnect chip, and determine whether to start the optical device screening step based on the comparison result between the basic performance index and the basic performance boundary index;

[0009] When the optical device screening step is started, the optical device screening module constructs a silicon-based optical interconnect chip simulation model, inputs various basic performance parameters of the silicon-based optical interconnect chip into the silicon-based optical interconnect chip simulation model, and the silicon-based optical interconnect chip simulation model completes j performance simulation analyses, and obtains the device performance index of each optical device in each performance simulation analysis;

[0010] The optical device evaluation module is used to obtain a device evaluation test index of each optical device, and determine whether to mark the optical device as an evaluation test device based on a comparison result between the device evaluation test index and the device evaluation test threshold index;

[0011] The dynamic evaluation module is used to perform evaluation tests on each evaluation test device in turn.

[0012] Furthermore, the basic performance index of the silicon-based optical interconnect chip is obtained in the following manner: various basic performance parameters of the silicon-based optical interconnect chip in a cycle are collected, various basic performance parameters are combined into a basic performance parameter set in the form of a data set, a basic performance analysis model is obtained, the basic performance parameter set is input into the basic performance analysis model, and the basic performance analysis model outputs the basic performance index of the silicon-based optical interconnect chip.

[0013] Furthermore, the construction process of the silicon-based optical interconnect chip simulation model is as follows: obtain the optical path diagram of the silicon-based optical interconnect chip, select simulation software, create a silicon-based optical interconnect chip model in the simulation software based on the optical path diagram of the silicon-based optical interconnect chip, add optical devices based on the optical path diagram, and determine the position and connection relationship of the optical devices, obtain the design parameters of the optical devices, assign corresponding design parameters to the optical devices in the silicon-based optical interconnect chip model, and then construct a silicon-based optical interconnect chip simulation model. The silicon-based optical interconnect chip simulation model can simulate the operation of the silicon-based optical interconnect chip.

[0014] Furthermore, the device performance index of each optical device in a performance simulation analysis is obtained in the following manner: after the silicon-based optical interconnect chip simulation model completes a performance simulation analysis, the characteristic parameters of each optical device are obtained, the characteristic parameter analysis model corresponding to each optical device is obtained, the characteristic parameters of each optical device are respectively input into the corresponding characteristic parameter analysis model, and the device performance index of each optical device is obtained as output.

[0015] Furthermore, the device evaluation test index of the optical device is obtained by the following method: obtain the device performance index Sjp of an optical device in each performance simulation analysis, j=1, 2, ..., J, J is the total number of performance simulation analyses, j is the sequence number of the corresponding performance simulation analyses, set the device performance coefficient to ep, p=1, 2, ..., P, e1<e2<e3<...<eP, set the device performance standard index, when the device performance index is less than the device performance standard index, increase the number of performance disappointments by one, mark the number of performance disappointments as Hvz, and use the formula The device evaluation test index Ewk of the optical device is obtained, wherein f1 is the performance disappointment coefficient.

[0016] Furthermore, a device evaluation test threshold index is set. When the device evaluation test index of the optical device is greater than the device evaluation test threshold index, the optical device is marked as an evaluation test device. When the device evaluation test index of the optical device is less than or equal to the device evaluation test threshold index, no further processing is performed.

[0017] Furthermore, each evaluation test device is evaluated and tested in turn, specifically: all evaluation test devices are sorted in order from large to small according to the value of the device evaluation test index, and each evaluation test device is evaluated and tested in order according to the sorting order.

[0018] Compared with the prior art, the present invention has the following beneficial effects:

[0019] 1. Set up a basic performance analysis module, combine the basic performance parameter collection of silicon-based optical interconnect chips with deep learning technology, conduct regular comprehensive evaluation of the performance of silicon-based optical interconnect chips, accurately evaluate the performance of silicon-based optical interconnect chips, and start optical device screening in time when the performance of silicon-based optical interconnect chips is poor;

[0020] 2. Set up an optical device screening module, an optical device evaluation module and a dynamic evaluation module. After starting the optical device screening work, by building a simulation model of a silicon-based optical interconnect chip, the possible performance of each optical device in the silicon-based optical interconnect chip under the current performance parameters can be simulated, and the test and evaluation requirements of each optical device can be comprehensively analyzed. According to the dynamic screening and sorting of the test and evaluation order of the optical devices based on the test and evaluation requirements, it is no longer necessary to test and evaluate each optical device. The abnormal optical devices in the silicon-based optical interconnect chip can be quickly identified, thereby improving the test and evaluation accuracy and efficiency of the optical devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 This is a system module diagram of a deep learning-based dynamic evaluation system for silicon-based optical interconnect chips;

[0022] Figure 2This is a system operation flow chart of a deep learning-based dynamic evaluation system for silicon-based optical interconnect chips;

[0023] Figure 3 A flow chart for obtaining basic performance indexes of silicon-based optical interconnect chips;

[0024] Figure 4 A flow chart for obtaining a device performance index for an optical device. DETAILED DESCRIPTION

[0025] Reference Figure 1-Figure 4 , a deep learning-based dynamic evaluation system for silicon-based optical interconnect chips, including a basic performance analysis module, an optical device screening module, an optical device evaluation module, and a dynamic evaluation module.

[0026] Basic performance analysis module: collect various basic performance parameters of silicon-based optical interconnect chips in real time (basic performance parameters include transmission rate, optical loss, optical power, modulation efficiency, bit error rate, etc.), periodically obtain the basic performance index of silicon-based optical interconnect chips, set the basic performance boundary index (the basic performance boundary index is the threshold set by the system), when the basic performance index of the silicon-based optical interconnect chip is greater than the basic performance boundary index, no further processing is performed, when the basic performance index of the silicon-based optical interconnect chip is less than or equal to the basic performance boundary index, start the optical device screening step.

[0027] The basic performance index of the silicon-based optical interconnect chip is obtained in the following way: collecting various basic performance parameters of the silicon-based optical interconnect chip in one cycle, combining the various basic performance parameters into a basic performance parameter set in the form of a data set, obtaining a basic performance analysis model, inputting the basic performance parameter set into the basic performance analysis model, and the basic performance analysis model outputs the basic performance index of the silicon-based optical interconnect chip.

[0028] The basic performance analysis model is constructed as follows: collect s basic performance parameter sets, build a deep learning model, use the medical feature comparison group as the training data of the deep learning model, assign a basic performance index to each training data, and the value range of the basic performance index is (0.1~1.9). The closer the value of the basic performance index is to 1.9, the better the performance of the silicon-based optical interconnect chip is, and the closer the value of the basic performance index is to 0.1, the worse the performance of the silicon-based optical interconnect chip is. The training data is divided into training set, validation set and test set according to the set ratio of 4:2:1, and the training set, validation set and test set are trained. After the training is completed, the basic performance analysis model is constructed.

[0029] A basic performance analysis module is set up to combine the basic performance parameter collection and deep learning technology of silicon-based optical interconnect chips to conduct regular comprehensive evaluations on the performance of silicon-based optical interconnect chips, accurately evaluate the performance of silicon-based optical interconnect chips, and start optical device screening in time when the performance of silicon-based optical interconnect chips is poor.

[0030] Optical device screening module: When the optical device screening step is started, a silicon-based optical interconnect chip simulation model is constructed, and various basic performance parameters of the silicon-based optical interconnect chip are input into the silicon-based optical interconnect chip simulation model. The silicon-based optical interconnect chip simulation model completes j performance simulation analyses, and obtains the device performance index of each optical device in each performance simulation analysis.

[0031] The construction process of the silicon-based optical interconnect chip simulation model is as follows: obtain the optical path diagram of the silicon-based optical interconnect chip, select simulation software (Lumerical series simulation software can be selected), create a silicon-based optical interconnect chip model in the simulation software based on the optical path diagram of the silicon-based optical interconnect chip, add optical devices based on the optical path diagram, and determine the position and connection relationship of the optical devices. Obtain the design parameters of the optical devices (design parameters include size, shape, material properties, etc.), assign corresponding design parameters to the optical devices in the silicon-based optical interconnect chip model, and then construct a silicon-based optical interconnect chip simulation model. The silicon-based optical interconnect chip simulation model can simulate the operation of the silicon-based optical interconnect chip.

[0032] Input various basic performance parameters of silicon-based optical interconnect chips into the silicon-based optical interconnect chip simulation model. The silicon-based optical interconnect chip simulation model will adjust the characteristic parameters of the optical devices in the silicon-based optical interconnect chip simulation model according to the various basic performance parameters. The characteristic parameters include optical properties (such as refractive index, loss, etc.), electrical properties (such as resistance, capacitance, etc.), and thermal properties (such as thermal conductivity, thermal expansion coefficient, etc.). The characteristic parameters of these optical devices are often difficult to directly collect from the silicon-based optical interconnect chip due to the complex structure of the silicon-based optical interconnect chip. After the characteristic parameters of the optical devices are adjusted, various basic parameters of the silicon-based optical interconnect chip simulation model will be adjusted. This performance parameter has an impact. After the characteristic parameter adjustment of the silicon-based optical interconnect chip is completed, the silicon-based optical interconnect chip simulation model is simulated and run. When the various basic performance parameters of the silicon-based optical interconnect chip simulation model are completely consistent with the various basic performance parameters of the silicon-based optical interconnect chip, a performance simulation analysis is completed. The adjustment of optical devices in all completed performance simulation analyses is inconsistent, but the various basic performance parameters reflected in all completed performance simulation analyses are consistent, which means that a variety of different problems in the silicon-based optical interconnect chip will cause the various basic performance parameters of the silicon-based optical interconnect chip to show the same performance.

[0033] The device performance index of each optical device in a performance simulation analysis is obtained in the following way: after the silicon-based optical interconnect chip simulation model completes a performance simulation analysis, the characteristic parameters of each optical device are obtained, the characteristic parameter analysis model corresponding to each optical device is obtained, the characteristic parameters of each optical device are respectively input into the corresponding characteristic parameter analysis model, and the device performance index of each optical device is obtained as output.

[0034] Each optical device corresponds to a characteristic parameter analysis model, and all characteristic parameter analysis models are constructed based on the deep learning model. The difference between different characteristic parameter analysis models lies only in the difference in training data. In this embodiment, taking the optical waveguide as an example, the construction process of the characteristic parameter analysis model of the optical waveguide is disclosed: the characteristic parameters of b optical waveguides are collected (if the characteristic parameter analysis model of the optical beam splitter is constructed, the characteristic parameters of b optical beam splitters are collected), the deep learning model is constructed, the characteristic parameters of the optical waveguide are used as the training data of the deep learning model, and a device performance index is assigned to each training data. The value range of the device performance index is (4.0~8.0). The larger the value of the device performance index, the more normal the performance of the optical waveguide is (if the characteristic parameter analysis model of the optical beam splitter is constructed, the larger the value of the device performance index, the more normal the performance of the optical beam splitter is), and the smaller the value of the device performance index, the more abnormal the performance of the optical waveguide is. The training data is divided into a training set and a validation set according to a set ratio of 5:2, and the training set and the validation set are trained. After the training is completed, the characteristic parameter analysis model of the optical waveguide is constructed.

[0035] Optical device evaluation module: obtain the device evaluation test index of each optical device, set the device evaluation test threshold index (the device evaluation test threshold index is the threshold set by the system), when the device evaluation test index of the optical device is greater than the device evaluation test threshold index, mark the optical device as an evaluation test device, when the device evaluation test index of the optical device is less than or equal to the device evaluation test threshold index, no further processing is performed.

[0036] The device evaluation test index of the optical device is obtained in the following way: obtain the device performance index Sjp of an optical device in each performance simulation analysis, j=1, 2, ..., J, J is the total number of performance simulation analyses, j is the sequence number of the corresponding performance simulation analysis, set the device performance coefficient to ep, p=1, 2, ..., P, e1<e2<e3<...<eP, each device performance coefficient corresponds to a device performance index within a range, the value range of the device performance index includes (0, Sj1], (Sj1, Sj2], ..., (SjP-1, SjP], when Sjp∈(0, Sj1], the device performance coefficient is e1, set the device performance standard index (the device performance standard index is the threshold set by the system), when the device performance index is greater than or equal to the device performance standard index, no further processing is performed, when the device performance index is less than the device performance standard index, the number of performance disappointments is increased by one, and the number of performance disappointments is marked as Hvz, and the formula is used. The device evaluation test index Ewk of the optical device is obtained, wherein f1 is the performance disappointment coefficient, and the value of f1 is 0.89.

[0037] Dynamic evaluation module: sort all evaluation test devices in descending order according to the value of the device evaluation test index, and perform evaluation tests on each evaluation test device in the sorted order (the evaluation test methods are all conventional methods and will not be elaborated here).

[0038] An optical device screening module, an optical device evaluation module and a dynamic evaluation module are set up. After starting the optical device screening work, by constructing a simulation model of a silicon-based optical interconnect chip, the possible performance of each optical device in the silicon-based optical interconnect chip under the current performance parameters can be simulated, and the test and evaluation requirements of each optical device can be comprehensively analyzed. The test and evaluation order of optical devices can be dynamically screened and sorted according to the test and evaluation requirements. It is no longer necessary to test and evaluate each optical device, and abnormal optical devices in the silicon-based optical interconnect chip can be quickly identified, thereby improving the test and evaluation accuracy and efficiency of optical devices.

[0039] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0040] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0041] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0042] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0043] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0044] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0045] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks.

[0046] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A deep learning-based dynamic evaluation system for silicon-based optical interconnect chips, characterized in that: Including basic performance analysis module, optical device screening module, optical device evaluation module, dynamic evaluation module; The basic performance analysis module is used to periodically obtain the basic performance index of the silicon-based optical interconnect chip, and determine whether to start the optical device screening step based on the comparison result between the basic performance index and the basic performance boundary index; The basic performance index of the silicon-based optical interconnect chip is obtained by the following method: collecting various basic performance parameters of the silicon-based optical interconnect chip in a cycle, combining various basic performance parameters into a basic performance parameter set in the form of a data set, obtaining a basic performance analysis model, inputting the basic performance parameter set into the basic performance analysis model, and the basic performance analysis model outputs the basic performance index of the silicon-based optical interconnect chip; When the optical device screening step is started, the optical device screening module constructs a silicon-based optical interconnect chip simulation model, inputs various basic performance parameters of the silicon-based optical interconnect chip into the silicon-based optical interconnect chip simulation model, and the silicon-based optical interconnect chip simulation model completes j performance simulation analyses, and obtains the device performance index of each optical device in each performance simulation analysis; The device performance index of each optical device in a performance simulation analysis is obtained in the following manner: after the silicon-based optical interconnect chip simulation model completes a performance simulation analysis, characteristic parameters of each optical device are obtained, a characteristic parameter analysis model corresponding to each optical device is obtained, the characteristic parameters of each optical device are respectively input into the corresponding characteristic parameter analysis model, and the device performance index of each optical device is obtained as output; The optical device evaluation module is used to obtain a device evaluation test index of each optical device, and determine whether to mark the optical device as an evaluation test device based on a comparison result between the device evaluation test index and the device evaluation test threshold index; The device evaluation test index of the optical device is obtained in the following way: obtain the device performance index Sjp of an optical device in each performance simulation analysis, j=1, 2, ..., J, J is the total number of performance simulation analyses, j is the sequence number of the corresponding performance simulation analyses, set the device performance coefficient to ep, p=1, 2, ..., P, e1<e2<e3<...<eP, each device performance coefficient corresponds to a device performance index within a range, and the value range of the device performance index includes (0, Sj1], (Sj1, Sj2], ..., (SjP-1, SjP], set the device performance standard index, when the device performance index is less than the device performance standard index, increase the number of performance disappointments by one, mark the number of performance disappointments as Hvz, and use the formula The device evaluation test index Ewk of the optical device is obtained, wherein f1 is the performance disappointment coefficient; The dynamic evaluation module is used to perform evaluation tests on each evaluation test device in turn.

2. According to claim 1, a deep learning-based dynamic evaluation system for silicon-based optical interconnect chips is characterized in that: The construction process of the silicon-based optical interconnect chip simulation model is as follows: obtain the optical path diagram of the silicon-based optical interconnect chip, select simulation software, create a silicon-based optical interconnect chip model in the simulation software based on the optical path diagram of the silicon-based optical interconnect chip, add optical devices based on the optical path diagram, and determine the position and connection relationship of the optical devices, obtain the design parameters of the optical devices, assign corresponding design parameters to the optical devices in the silicon-based optical interconnect chip model, and then construct a silicon-based optical interconnect chip simulation model. The silicon-based optical interconnect chip simulation model can simulate the operation of the silicon-based optical interconnect chip.

3. According to the deep learning-based dynamic evaluation system for silicon-based optical interconnect chips of claim 1, it is characterized in that: A device evaluation test threshold index is set. When the device evaluation test index of the optical device is greater than the device evaluation test threshold index, the optical device is marked as an evaluation test device. When the device evaluation test index of the optical device is less than or equal to the device evaluation test threshold index, no further processing is performed.

4. According to the deep learning-based dynamic evaluation system for silicon-based optical interconnect chips of claim 1, it is characterized in that: The evaluation test is performed on each evaluation test device in sequence. Specifically, all the evaluation test devices are sorted in descending order according to the value of the device evaluation test index, and the evaluation test is performed on each evaluation test device in sequence according to the sorting order.

Images