KR20210066545A - Electronic device, method, and computer readable medium for simulation of semiconductor device - Google Patents

Abstract

Various embodiments of the present invention relate to a first electronic device for a first deep neural network (DNN) included in a set of DNNs having a hierarchical structure, which comprises: at least one processor; at least one memory configured for storage of instructions; and a communication circuit operatively coupled with the at least one processor, wherein the at least one processor, when the instructions are executed, is configured to: receive characteristics information on a result of simulation of a first semiconductor device performed by a semiconductor device simulator connected to a second electronic device from the second electronic device for a second DNN which is included in the set of the DNNs, and is ranked lower than the first DNN; perform machine learning, based on the characteristics information; receive a signal of requesting provision of an initial solution in order to perform simulation of the second semiconductor device by using the semiconductor device simulator from the second electronic device; and in respond to reception of the signal, transmit the refined initial solution to the second electronic device based on a result of the machine learning. Accordingly, the resource efficiency of a semiconductor device simulator can be reinforced.

Description

ELECTRONIC DEVICE, METHOD, AND COMPUTER READABLE MEDIUM FOR SIMULATION OF SEMICONDUCTOR DEVICE

Various embodiments to be described below relate to an electronic device, a method, and a computer readable medium for simulation of a semiconductor device.

A deep neural network (DNN) is a neural network with two or more layers with a certain level of complexity. DNNs can use sophisticated mathematical modeling to process data in complex ways.

For a simulation for predicting electrical characteristics of a semiconductor device, a semiconductor device simulator is used. The semiconductor device simulator always performs the respective simulations from an equilibrium state despite performing simulations for similar semiconductor devices. Accordingly, a method for enhancing the resource efficiency of the semiconductor device simulator may be required.

The technical problems to be achieved in this document are not limited to the technical problems mentioned above, and other technical problems not mentioned can be clearly understood by those of ordinary skill in the art to which the present invention belongs from the description below. There will be.

According to various embodiments, a first electronic device for a first DNN included in a set of deep neural networks (DNNs) having a hierarchical structure includes at least one processor, at least one memory configured to store instructions, and communication circuitry operatively coupled with the at least one processor, wherein the at least one processor, when executing the instructions, is configured to: Characteristic information on a simulation result of a first semiconductor device performed by a semiconductor device simulator connected to the second electronic device from a second electronic device for a second DNN that is included in the set and is a lower DNN of the first DNN a signal requesting to receive, based on the feature information, perform machine learning, and provide an initial solution for performing a simulation of a second semiconductor device using the semiconductor device simulator from the second electronic device receive, and in response to the reception of the signal, transmit an initial solution refined based on a result of the machine learning to the second electronic device.

A method for operating a first electronic device for a first DNN included in a set of deep neural networks (DNNs) having a hierarchical structure, according to various embodiments, includes: Characteristics of a simulation result of a first semiconductor device performed by a semiconductor device simulator connected to the second electronic device from a second electronic device for a second DNN included in the set of DNNs and which is a lower DNN of the first DNN to provide an initial solution for receiving information, performing machine learning based on the feature information, and performing simulation of a second semiconductor device using the semiconductor device simulator from the second electronic device The method may include receiving a signal for requesting that the signal is received, and transmitting, to the second electronic device, an initial solution refined based on a result of the machine learning in response to the reception of the signal.

A non-transitory computer readable storage medium according to various embodiments is included in a set of deep neural networks (DNNs) having a communication circuit and a hierarchical structure. When executed by one or more processors of a first electronic device for a first DNN, from a second electronic device for a second DNN that is included in the set of DNNs and is a lower DNN of the first DNN, the second electronic device Receive feature information on the simulation result of the first semiconductor device performed by the semiconductor device simulator connected to the device, perform machine learning based on the feature information, and execute the semiconductor device simulator from the second electronic device receiving a signal requesting to provide an initial solution for performing a simulation of a second semiconductor device using the second semiconductor device, and in response to the reception of the signal, a refined initial solution based on a result of the machine learning One or more programs including instructions for causing the first electronic device to be transmitted to the second electronic device may be stored.

An electronic device, a method, and a computer-readable medium according to various embodiments of the present disclosure collect characteristic information about a simulation result of a semiconductor device from another electronic device connected to a semiconductor device simulator and perform a machine operation based on the collected characteristic information By providing the initial solution to the other electronic device based on the learning result, the time consumed by the semiconductor device simulator to acquire electrical characteristics of the semiconductor device may be reduced.

Effects obtainable in the present disclosure are not limited to the above-mentioned effects, and other effects not mentioned may be clearly understood by those of ordinary skill in the art to which the present disclosure belongs from the description below. will be.

1 shows an example of a set of deep neural networks (DNNs) having a hierarchical structure, in accordance with various embodiments.
2 is a simplified block diagram of a first electronic device for a first DNN included in a set of DNNs having a hierarchical structure according to various embodiments.
3 illustrates an example of an operation of a first electronic device and an operation of a second electronic device for providing an initial solution to a semiconductor device simulator according to various embodiments of the present disclosure.
4 illustrates an example of an electrical characteristic obtained by providing an initial solution to a semiconductor device simulator according to various embodiments.
5 is a flowchart illustrating a method of providing an initial solution to a semiconductor device simulator according to various embodiments of the present disclosure;

Terms used in the present disclosure are used only to describe specific embodiments, and may not be intended to limit the scope of other embodiments. The singular expression may include the plural expression unless the context clearly dictates otherwise. Terms used herein, including technical or scientific terms, may have the same meaning as commonly understood by one of ordinary skill in the art described in the present disclosure. Among the terms used in the present disclosure, terms defined in a general dictionary may be interpreted with the same or similar meaning as the meaning in the context of the related art, and unless explicitly defined in the present disclosure, ideal or excessively formal meanings is not interpreted as In some cases, even terms defined in the present disclosure cannot be construed to exclude embodiments of the present disclosure.

In various embodiments of the present disclosure described below, a hardware approach method will be described as an example. However, since various embodiments of the present disclosure include technology using both hardware and software, various embodiments of the present disclosure do not exclude a software-based approach.

1 shows an example of a set of deep neural networks (DNNs) having a hierarchical structure, in accordance with various embodiments.

Referring to FIG. 1 , a set 10 of DNNs includes DNN 90 , DNN 91-1 through DNN 91-n (n is any natural number greater than 1), and DNN 92-1 through DNN 92-1. DNN(92-m), where m is any natural number greater than 1 and greater than n.

In various embodiments, the set 10 of DNNs may have a hierarchical structure. In various embodiments, the DNN 90 may be a DNN having the highest level in the set 10 of DNNs having a hierarchical structure. In various embodiments, each of DNNs 91-1 through DNNs 91-n may be a lower DNN of DNN 90 . For example, each of the DNNs 91-1 to 91-n may be a DNN immediately below the DNN 90 . In various embodiments, each of DNN 92-1 through DNN 92-m may be a lower DNN than DNN 91-1 through DNN 91-n. For example, the DNN 92-1, the DNN 92-2, and the DNN 92-3 may be DNNs immediately below the DNN 91-1.

In various embodiments, some of the DNNs included in the set of DNNs 10 (eg, DNNs 92-1 to DNNs 92-m) perform simulation to obtain electrical characteristics of the semiconductor device. may be connected to at least one semiconductor device simulator.

For example, each of the DNNs 92-1 to 92-m provides an initial solution to the at least one semiconductor device simulator to obtain electrical characteristics of the semiconductor device using the semiconductor device simulator. can do. The initial solution is the result of machine learning performed by each of the DNNs 92-1 to 92-m and other parts included in the set of DNNs 10 (eg, DNN 90 and DNN 91 ). -1) to at least some of the DNNs 91-n) may be determined based on the results of machine learning performed. However, it is not limited thereto.

For example, each of the DNNs 92-1 to 92-m may obtain information on a result of a simulation performed on the semiconductor device using the initial solution from the at least one semiconductor device simulator. can For example, each of the DNNs 92-1 to 92-m may extract feature information from a result of the simulation. The extraction of the feature information may be refined or updated by machine learning of each of the DNNs 92-1 to 92-m, according to embodiments.

In various embodiments, some other of the DNNs included in the set of DNNs 10 (eg, DNN 91-1 through DNN 91-n) are DNNs included in the set 10 of DNNs. may be connected to some of the above. For example, the DNN 91-1 is a DNN above the DNNs 92-1 to 92-3, and may be connected to each of the DNNs 92-1 to DNN 92-3.

For example, the DNN 91-1 may generate an initial solution to be provided to the at least one semiconductor device simulator based on a request of each of the DNNs 92-1 to 92-3. ) to DNN (92-3) can be provided to each.

For example, the DNN 91-1 may perform machine learning for refining the initial solution provided to each of the DNNs 92-1 to 92-3, the at least one The characteristic information may be received from each of the DNNs 92-1 to 92-3 connected to the semiconductor device simulator. The DNN 91-1 may refine the initial solution by performing machine learning based on the received feature information.

In various embodiments, a remaining portion of the DNNs included in the set 10 of DNNs (eg, the DNN 90 ) may be connected to the other portion included in the set 10 of DNNs. For example, the DNN 90 may be connected to the DNN 91-1 and the DNN 91-2, respectively.

For example, the DNN 90 may generate an initial solution to be provided to the at least one semiconductor device simulator based on the respective requests of the DNN 91-1 and the DNN 91-2. It may be provided to each of the DNNs 91-2. The initial solution provided to each of the DNNs 91-1 and 91-2 may be provided to each of the DNNs 92-1 to 92-m.

In various embodiments, the hierarchical structure of the set 10 of DNNs may be configured to correspond to the structure of an organization using the at least one semiconductor device simulator. For example, users using the semiconductor device simulator connected to the DNN 92-1 to the DNN 92-3 below the DNN 91-1 may belong to the first laboratory in the organization, and the DNN ( 91-2), users using other semiconductor device simulators connected to the lower DNNs 92-4 to DNN 92-k may belong to a second laboratory in the organization that is distinct from the first laboratory. For example, the DNN 91-1 and the DNN 90 above the DNN 91-2 may be configured for a first group including the first laboratory and the second laboratory.

1 illustrates a set 10 of DNNs having a hierarchical structure having three levels, but this is for convenience of description. For example, the number of levels of the hierarchical structure of the set 10 of DNNs may be changed according to at least one of the number of users or the structure of an organization to which users using the at least one semiconductor device simulator belong.

For convenience of description below, if necessary, the DNN 91-1 is referred to as the DNN 91, and any one of the DNN 92-1 to the DNN 92-3 may be referred to as the DNN 92. have.

2 is a simplified block diagram of a first electronic device for a first DNN included in a set of DNNs having a hierarchical structure according to various embodiments. This simplified block diagram may represent functional configurations of the first electronic device 100 for the DNN 91 shown in FIG. 1 . These functional components may be included in the first electronic device 100 to perform the operations described through the descriptions of FIGS. 3 to 5 .

The artificial neural network may refer to hardware, software, or a combination thereof for providing a generalized output with respect to an input.

For example, an artificial neural network is used to simulate a convolutional neural network (CNN), a Markov chain, a deep neural network (DNN), or a binarized neural network (BNN), etc. It may operate based on an application for and a processor for executing the application.

Referring to FIG. 2 , the first electronic device 100 is a device capable of performing machine learning through training, and may include a device for learning using a model composed of an artificial neural network. For example, the first electronic device 100 may input, output, construct and store information used for data mining, data analysis, and machine learning algorithms (eg, deep learning algorithms). can be configured.

The first electronic device 100 may transmit/receive data to and from an external electronic device (not shown) through a communication circuit (not shown), and may derive a result value by analyzing or learning data received from the external electronic device . The first electronic device 100 may distribute and process the calculation of the external electronic device.

The first electronic device 100 may be implemented as a server. The first electronic device 100 may be configured in plurality to form a neural network device set. Each of the first electronic devices 100 may process the calculation by distributing it, and may derive a result value through data analysis and learning based on the distributed processed data. The first electronic device 100 may transmit a result obtained by using a machine learning algorithm or the like to an external electronic device or another neural network device.

According to various embodiments, the first electronic device 100 may include an input unit 110 , a processor 120 , a memory 130 , and a learning processor 140 .

According to various embodiments, the input unit 110 may obtain input data for deriving an output value through artificial neural network model learning. The input unit 110 may obtain raw input data.

The processor 120 or the learning processor 140 may preprocess the raw input data to generate training data that can be input to the artificial neural network model learning. The preprocessing may be to extract a feature point from the input data. As described above, the input unit 110 may receive data through a communication circuit (not shown) to obtain input data or pre-process the data.

According to various embodiments, the processor 120 may collect usage history information from the first electronic device 100 and store it in the memory 130 . The processor 120 may determine the best combination for executing a specific function through the stored usage history information and predictive modeling. The processor 120 may receive image information, audio information, data, information about an initial solution, characteristic information, or user input information from the input unit 110 .

According to various embodiments, the processor 120 collects information in real time, processes or classifies the information, and stores the processed information in the memory 130 , a database of the memory 130 , or the learning processor 140 . can

According to various embodiments, when the operation of the first electronic device 100 is determined based on data analysis and machine learning algorithm, the processor 120 executes the determined operation. can control According to various embodiments, the processor 120 may control the first electronic device 100 according to a control command to perform a determined operation.

When a specific operation is performed, the processor 120 analyzes historical information indicating the execution of the specific operation through data analysis and machine learning algorithms and techniques, and updates previously learned information based on the analyzed information. can The processor 120 may improve the accuracy of data analysis and machine learning algorithms and performance based on the updated information together with the learning processor 140 .

According to various embodiments, the memory 130 may store input data obtained from the input unit 110 , learned data, or a learning history. The memory 130 may store the artificial neural network model 131 .

According to various embodiments, the artificial neural network model 131 may be stored in a space allocated to the memory 130 . The space allocated to the memory 130 stores the artificial neural network model 131 during learning or learned through the learning processor 140, and when the artificial neural network model 131 is updated through learning, the updated artificial neural network model ( 131) can be stored. The space allocated to the memory 130 may store the learned model by dividing it into a plurality of versions according to a learning time point or learning progress.

According to various embodiments, the memory 130 may include a database capable of storing and classifying input data obtained from the input unit 110 and the learned data.

According to various embodiments, the learning processor 140 learns the artificial neural network model 131 by directly acquiring data obtained by preprocessing the input data obtained by the processor 120 through the input unit 110 , or the memory 130 . The artificial neural network model 131 may be trained by acquiring preprocessed input data stored in the database of . For example, the learning processor 140 may acquire optimized parameters of the neural network model 131 by repeatedly learning the artificial neural network model 131 using various learning techniques.

According to various embodiments, the trained model may update the artificial neural network model 131 in the database. The learning processor 140 may be integrated into the neural network learning apparatus 100 or implemented in the memory 130 . For example, the learning processor 140 may be implemented using the memory 130 .

According to various embodiments, the learning processor 140 generally identifies, indexes, categorizes, manipulates, stores, retrieves, and outputs data for use in supervised or unsupervised learning, data mining, predictive analytics, or other devices. to store data in one or more databases. Here, the database may be implemented using the memory 130 , a memory maintained in a cloud computing environment, or another remote memory location accessible by the terminal through a communication method such as a network. The information stored in the learning processor 140 may be utilized by the processor 120 using any of a variety of different types of data analysis algorithms and machine learning algorithms. For example, examples of such algorithms include k-recent adjacency systems, fuzzy logic (eg likelihood theory), neural networks, Boltzmann machines, vector quantization, pulse neural networks, support vector machines, maximum margin classifiers, hill climbing, derivation Logical Systems Bayesian Networks, Perritnets (e.g. Finite State Machine, Milli Machine, Moore Finite State Machine), Classifier Trees (e.g. Perceptron Tree, Support Vector Tree, Markov Tree, Decision Tree Forest, Random Forest), Read Model and systems, artificial fusion, sensor fusion, image fusion, reinforcement learning, augmented reality, pattern recognition, automated planning, and more.

Hereinafter, for convenience of description, the processor 120 and the learning processor 140 may be referred to as at least one processor.

In various embodiments, the at least one processor is included in the set of DNNs 10 and includes, from a second electronic device for a second DNN 92 that is a lower DNN of the first DNN 91 , the second 2 It is possible to receive feature information about a simulation result of the first semiconductor device performed by the semiconductor device simulator connected to the electronic device. For example, the second electronic device may provide the initial solution provided from the first electronic device 100 to the semiconductor device simulator. The semiconductor device simulator may perform a simulation of the first semiconductor device based on the initial solution provided from the first electronic device 100 and provide the simulation result to the second electronic device. In various embodiments, the simulation result may include information on the electrical characteristics of the first semiconductor device. For example, the information on the electrical characteristics of the first semiconductor device may include a change in the magnitude of the drain current of the first semiconductor device according to the change in the magnitude of the gate voltage of the first semiconductor device. For example, the information on the electrical characteristics of the first semiconductor device may include data on an active mode section of the first semiconductor device and data on a saturation mode section of the first semiconductor device. However, it is not limited thereto. In various embodiments, the second electronic device may acquire characteristic information about the first semiconductor device based on a result of the simulation. For example, the second electronic device may perform machine learning on a result of the simulation, and obtain the feature information as a result of the performed machine learning. The second electronic device may provide the acquired characteristic information to the first electronic device 100 .

In various embodiments, the at least one processor may receive the feature information from the second electronic device. In various embodiments, the at least one processor may perform machine learning based on the feature information. According to embodiments, the at least one processor may perform validation of the feature information based on receiving the feature information.

In various embodiments, the at least one processor may receive, from the second electronic device, a signal requesting to provide an initial solution for performing a simulation of a second semiconductor device using the semiconductor device simulator. . The at least one processor may transmit an initial solution refined based on a result of the machine learning to the second electronic device in response to the reception of the signal. The time consumed for the simulation of the second semiconductor device performed using the refined initial solution may be shorter than the time consumed for the simulation of the first semiconductor device. For example, the refined initial solution may be configured to directly start the simulation of the second semiconductor device using an electrostatic potential, rather than starting the simulation of the second semiconductor device from a potential of 0 (V). Since it is configured, the time consumed for the simulation of the second semiconductor device performed using the refined initial solution may be shorter than the time consumed for the simulation of the first semiconductor device.

In various embodiments, the at least one processor may transmit information changed according to the result of the machine learning to a third electronic device for the third DNN 90 that is higher than the first DNN 91 . Through the transmission, the third electronic device may perform machine learning for an initial solution and resource management of semiconductor device simulators in terms of the entire system.

In various embodiments, the at least one processor simultaneously receives from a plurality of electronic devices including the second electronic device (eg, electronic devices for the DNN 92-1 to DNN 92-3). When receiving signals requesting to provide an initial solution, initial solutions for simulations of semiconductor devices associated with the first semiconductor device on which the simulation has been performed may first be provided. For example, the at least one processor may be configured to distinguish an initial solution for simulation of the second semiconductor device, which is expected to have similar electrical properties to those of the first semiconductor device, from the electrical properties of the first semiconductor device. An initial solution for simulation of a third semiconductor device predicted to have electrical characteristics of Through the initial solution providing process provided based on the correlation according to the simulation history of the semiconductor device, the system performing the simulation of the semiconductor device may enhance resource efficiency.

3 illustrates an example of operations of DNNs for providing an initial solution to a semiconductor device simulator according to various embodiments. The DNNs may be a DNN 91-k and a DNN 92-1.

Referring to FIG. 3 , the environment 300 may include a DNN 91 - k , a DNN 92 - 1 , and a semiconductor device simulator 310 .

Each of the DNN 91-k and the DNN 92-1 may proceed with a training process for providing an initial solution until it is identified that the reliability of the initial solution to be provided to the semiconductor device simulator 310 is equal to or greater than the reference reliability. have. For example, the DNN 92 - 1 may provide a first initial solution to the semiconductor device simulator 310 for simulation of device #(N) in the training processor. The semiconductor device simulator 310 may provide a simulation result using the first initial solution to the DNN 92 - 1 . The DNN 92 - 1 may extract feature information from the simulation result and perform a feasibility study on the first initial solution based on the feature information. For example, the DNN 92-1 may perform a feasibility study on the first initial solution by identifying whether the reliability of the first initial solution is equal to or greater than the reference reliability based on the feature information. . The DNN 92 - 1 may provide the DNN 91 - k with a result of the feasibility study for the first initial solution.

The DNN 91 - k may obtain a result of the feasibility study for the first initial solution from the DNN 92 - 1 . For example, the result of the validation may include data for indicating whether the reliability of the first initial solution is equal to or greater than the reference reliability. For example, the feasibility study result may include information on a difference between the first initial solution and the simulation result (or the feature information). However, it is not limited thereto.

The DNN 91 - k may initiate an initial solution prediction process based on identifying that the reliability of the first initial solution is greater than or equal to the reference reliability. The initial solution prediction process may refer to a process in which an initial solution provided through the DNN 92 - 1 is used for simulation of an actual semiconductor device. The training process may be continuously performed independently of the initial solution prediction process. The DNN 91 - k may perform machine learning to increase the reliability of the initial solution based on identifying that the reliability of the first initial solution is less than the reference reliability.

In the initial solution prediction process, the semiconductor device simulator 310 may request an initial solution for the simulation of device #(N+1) from the DNN 92 - 1 . The DNN 92 - 1 may request the initial solution from the DNN 91 - k in response to the request. The DNN 91 - k may provide the DNN 92 - 1 with a second initial solution as the initial solution in response to the request for the initial solution from the DNN 92 - 1 . The second initial solution may be an initial solution refined based on a result of the machine learning. The DNN 92 - 1 may provide the second initial solution to the semiconductor device simulator 310 . The semiconductor device simulator 310 may perform a simulation on device #(N+1) using the second initial solution, and provide the simulation result to the DNN 92 - 1 . The DNN 92 - 1 may extract feature information based on the simulation result and perform a feasibility study on the second initial solution. In order to increase the reliability of the initial solution provided to the semiconductor device simulator 310 , the DNN 92 - 1 uses at least one of the extracted feature information or the result of a validation of the second initial solution to the DNN 91 - 1 . k) can be provided.

As described above, the DNN 91-k according to various embodiments performs machine learning based on information received from the DNN 92-1, and generates a refined initial solution based on the performed machine learning in the semiconductor. By providing the device to the device simulator 310 , it is possible to reduce the simulation time for the semiconductor device performed by the semiconductor device simulator 310 . Since the refined initial solution is configured to use an electrostatic potential, it may be helpful to develop a model that considers quantum effects based on only the desired voltage condition for the simulation.

4 illustrates an example of an electrical characteristic obtained by providing an initial solution to a semiconductor device simulator according to various embodiments.

Referring to FIG. 4 , the horizontal axis of the graph 400 may represent the gate voltage of the semiconductor device, and the vertical axis of the graph 400 may represent the drain current of the semiconductor device. The curves of the graph 400 are obtained by using an electrostatic potential for simulation of a semiconductor device, and may represent a change in the drain current according to a change in the gate voltage. Meanwhile, the electrostatic potential may be determined by an initial solution obtained based on a result of machine learning performed using the set 10 of DNNs shown in FIG. 1 . The curves may be obtained by simulation of a semiconductor device omitting (or bypassing) the process of gradually changing VDD from 0 (V).

As can be seen through the curves of the graph 400, the result of the simulation using the electrostatic potential is configured to be able to identify that the section 410 is a section of the saturation mode and the section 420 is a section of the active mode. Therefore, the electrical characteristics of the semiconductor device can be exhibited.

5 is a flowchart illustrating a method of providing an initial solution to a semiconductor device simulator according to various embodiments of the present disclosure; The operations illustrated by the flowchart may be executed by the first electronic device 100 illustrated in FIG. 2 or at least one processor of the first electronic device 100 illustrated in FIG. 2 .

Referring to FIG. 5 , in operation 510 , the at least one processor includes a second electron for a second DNN 92-1 that is included in the set of DNNs 10 and is a lower DNN of the first DNN 91 . From the device, feature information on the simulation result of the first semiconductor device performed by the semiconductor device simulator (eg, the semiconductor device simulator 310 ) connected to the second electronic device may be received. The characteristic information may be transmitted from the second electronic device to the first electronic device 100 based on the authorization of the first electronic device 100 .

In operation 520, the at least one processor may perform machine learning based on the feature information. For example, the at least one processor may perform the machine learning based on the feature information to obtain an initial solution to be provided for a simulation of a semiconductor device to be performed later by the semiconductor device simulator.

In operation 530, the at least one processor may receive a signal requesting to provide an initial solution for performing a simulation of a second semiconductor device using the semiconductor device simulator from the second electronic device. For example, the signal may be transmitted from the second electronic device to the first electronic device 100 based on the application of the first electronic device 100 . In various embodiments, the second semiconductor device is a semiconductor device distinct from the first semiconductor device, but may be a semiconductor device having electrical characteristics similar to those of the first semiconductor device.

In operation 540, in response to the reception of the signal, the at least one processor may transmit an initial solution refined based on a result of the machine learning to the second electronic device. The second electronic device may perform a simulation of the second semiconductor device by providing the refined initial solution to the semiconductor device simulator, and transmit characteristic information of the simulation result to the first electronic device 100 . have.

Meanwhile, the at least one processor includes information changed according to the result of the machine learning in the set of DNNs 10 and is a third electronic device for a third DNN 90 that is a higher DNN of the first DNN 91 . can be sent to

As described above, the first electronic device 100 according to various embodiments includes not only the second electronic device for a DNN lower than the DNN associated with the first electronic device 100 but also the first electronic device ( 100) through communication with the third electronic device for a DNN higher than the DNN related to 100), the efficiency of resources used for simulation of a semiconductor device using the semiconductor device simulator may be enhanced.

As described above, according to various embodiments, a first electronic device for a first DNN included in a set of deep neural networks (DNNs) having a hierarchical structure is at least a processor, at least one memory configured to store instructions, and a communication circuit operatively coupled with the at least one processor, wherein the at least one processor is configured to: When executed, simulation of a first semiconductor device performed by a semiconductor device simulator connected to the second electronic device from a second electronic device for a second DNN included in the set of DNNs and which is a lower DNN of the first DNN Receive feature information on the result, perform machine learning based on the feature information, and provide an initial solution for performing simulation of a second semiconductor device using the semiconductor device simulator from the second electronic device and receive a signal requesting to do so, and in response to the reception of the signal, transmit an initial solution refined based on a result of the machine learning to the second electronic device.

In various embodiments, when the at least one processor executes the instructions, a result of the machine learning is provided to a third electronic device for a third DNN that is included in the set of DNNs and is a higher DNN of the first DNN. It may be further configured to transmit the changed information according to the

In various embodiments, the set of DNNs may be allocated to a plurality of users using the semiconductor device simulator, include the second DNN, and include a plurality of DNNs lower than the first DNN. .

In various embodiments, the at least one processor may be further configured to, when executing the instructions, perform a feasibility study of the characteristic information based on receiving the characteristic information.

In various embodiments, the initial solution refined based on the result of the machine learning may be configured so that the semiconductor device simulator directly uses the electrostatic potential for simulation of the second semiconductor device. In various embodiments, the electrostatic potential may be used to measure the magnitude of the drain current of the second semiconductor device. In various embodiments, the change in the magnitude of the drain current of the second semiconductor device may include obtaining a period of an active mode of the second semiconductor device and a saturation mode of the second semiconductor device. It can be used to obtain a section of .

In various embodiments, the hierarchical structure of the set of DNNs may correspond to a hierarchical structure of a group of users using a plurality of semiconductor device simulators including the semiconductor device simulator.

In various embodiments, the characteristic information may be transmitted from the second electronic device to the first electronic device based on the authorization of the first electronic device.

In various embodiments, the signal requesting to provide the initial solution may be transmitted from the second electronic device to the first electronic device based on the authorization of the first electronic device.

Methods according to the embodiments described in the claims or specifications of the present disclosure may be implemented in the form of hardware, software, or a combination of hardware and software.

When implemented in software, a computer-readable storage medium storing one or more programs (software modules) may be provided. One or more programs stored in the computer-readable storage medium are configured to be executable by one or more processors in an electronic device (device). One or more programs include instructions for causing an electronic device to execute methods according to embodiments described in a claim or specification of the present disclosure.

Such programs (software modules, software) include random access memory, non-volatile memory including flash memory, read only memory (ROM), electrically erasable programmable ROM (EEPROM: Electrically Erasable Programmable Read Only Memory), magnetic disc storage device, Compact Disc-ROM (CD-ROM), Digital Versatile Discs (DVDs), or any other form of It may be stored in an optical storage device or a magnetic cassette. Alternatively, it may be stored in a memory composed of a combination of some or all thereof. In addition, each configuration memory may be included in plurality.

In addition, the program is transmitted through a communication network consisting of a communication network such as the Internet, Intranet, Local Area Network (LAN), Wide LAN (WLAN), or Storage Area Network (SAN), or a combination thereof. It may be stored on an attachable storage device that can be accessed. Such a storage device may be connected to a device implementing an embodiment of the present disclosure through an external port. In addition, a separate storage device on the communication network may be connected to the device implementing the embodiment of the present disclosure.

In the specific embodiments of the present disclosure described above, components included in the disclosure are expressed in the singular or plural according to the specific embodiments presented. However, the singular or plural expression is appropriately selected for the situation presented for convenience of description, and the present disclosure is not limited to the singular or plural component, and even if the component is expressed in plural, it is composed of the singular or singular. Even an expressed component may be composed of a plurality of components.

Meanwhile, although specific embodiments have been described in the detailed description of the present disclosure, various modifications are possible without departing from the scope of the present disclosure. Therefore, the scope of the present disclosure should not be limited to the described embodiments and should be defined by the claims described below as well as the claims and equivalents.

Claims (10)

A first electronic device for a first DNN included in a set of deep neural networks (DNNs) having a hierarchical structure, the first electronic device comprising:
at least one processor;
at least one memory configured to store instructions; and
a communication circuit operatively coupled with the at least one processor;
When the at least one processor executes the instructions,
From a second electronic device for a second DNN included in the set of DNNs and which is a lower DNN of the first DNN, a simulation result of a first semiconductor device performed by a semiconductor device simulator connected to the second electronic device receive characteristic information;
Based on the feature information, machine learning is performed,
receiving, from the second electronic device, a signal requesting to provide an initial solution for performing a simulation of a second semiconductor device using the semiconductor device simulator;
a first electronic device configured to, in response to receiving the signal, transmit an initial solution refined based on a result of the machine learning to the second electronic device.
The method according to claim 1, wherein the at least one processor, when executing the instructions,
The first electronic device further configured to transmit information changed according to a result of the machine learning to a third electronic device for a third DNN included in the set of DNNs and which is a DNN higher than the first DNN.
The method according to claim 1, wherein the set of DNNs,
A first electronic device allocated to each of a plurality of users using the semiconductor device simulator, including the second DNN, and including a plurality of DNNs lower than the first DNN.
The method according to claim 1, wherein the at least one processor, when executing the instructions,
The first electronic device further configured to perform a feasibility study of the feature information based on receiving the feature information.
The method according to claim 1, The initial solution refined based on the result of the machine learning,
a first electronic device, wherein the semiconductor device simulator is configured to directly use an electrostatic potential for simulation of the second semiconductor device.
The method according to claim 5, wherein the electrostatic potential is,
A first electronic device used to measure a magnitude of a drain current of the second semiconductor device.
The method according to claim 6, The change in the magnitude of the drain current of the second semiconductor device,
A first electronic device used to acquire an active mode section of the second semiconductor device and a saturation mode section of the second semiconductor device.
The method according to claim 1, wherein the hierarchical structure of the set of DNNs comprises:
A first electronic device corresponding to a hierarchical structure of a group of users using a plurality of semiconductor device simulators including the semiconductor device simulator.
The method according to claim 1, The characteristic information,
An electronic device transmitted from the second electronic device to the first electronic device based on the authorization of the first electronic device.
The method according to claim 1, wherein the signal requesting to provide the initial solution comprises:
An electronic device transmitted from the second electronic device to the first electronic device based on the authorization of the first electronic device.

Images