CN117632366A

CN117632366A - Method, device, equipment and medium for determining resource occupation state of simulation accelerator

Info

Publication number: CN117632366A
Application number: CN202311617870.9A
Authority: CN
Inventors: 胡东瑞; 吴佳欢; 陈保文
Original assignee: Hexin Technology Co ltd; Hexin Technology Suzhou Co ltd
Current assignee: Hexin Technology Co ltd; Hexin Technology Suzhou Co ltd
Priority date: 2023-11-29
Filing date: 2023-11-29
Publication date: 2024-03-01

Abstract

The application relates to a simulation accelerator resource occupation state determining method, a simulation accelerator resource occupation state determining device, computer equipment and a storage medium. The method comprises the following steps: in response to a resource state determination request for the simulation accelerator, acquiring first current data of the simulation accelerator in a first historical time interval and second current data of the simulation accelerator in a second historical time interval; based on the first current data, obtaining a resource occupation classification threshold value, and comparing the resource occupation classification threshold value with the second current data to obtain a predicted resource occupation state of a second historical time interval; acquiring a board occupation state of a second historical time interval, and acquiring a misjudgment rate parameter according to the board occupation state and the predicted resource occupation state; and under the condition that the misjudgment rate parameter meets the preset condition, acquiring the resource occupation state of the simulation accelerator at the current time based on the resource occupation classification threshold and the current data. The method can improve the accuracy of determining the resource occupation state of the simulation accelerator.

Description

Method, device, equipment and medium for determining resource occupation state of simulation accelerator

Technical Field

The present invention relates to the field of computer technologies, and in particular, to a method, an apparatus, a computer device, a storage medium, and a computer program product for determining a resource occupancy state of a simulation accelerator.

Background

With the development of computer technology, a Cadence Palladium simulation accelerator is applied to a large-scale SOC design, namely a system on a chip design, and is taken as an acceleration platform which is most widely applied in the SOC design, so that the computing resource of the simulation accelerator is precious for the SOC design, and the simulation accelerator needs to be fully utilized by SOC design users.

In the conventional technology, in order to ensure that the computing resources of the simulation accelerator are fully utilized, the resource occupation state of the simulation accelerator can be determined according to the board occupation state of the simulation accelerator, and then the load regulation and control are performed according to the resources.

However, in the actual production process, the simulation operation is often affected by the working habit of the SOC design user, and the simulation operation often occupies the accelerator board resources and cannot be fully utilized, for example, the automation script of the user often locks the accelerator board resources in the simulation process, but does not actually run the simulation calculation task. Therefore, the resource occupation state of the accelerator cannot be accurately represented directly according to the board card resource occupation state of the accelerator, and the accuracy of determining the resource occupation state of the existing simulation accelerator is low.

Disclosure of Invention

In view of the foregoing, it is desirable to provide a method, an apparatus, a computer device, a computer readable storage medium, and a computer program product for determining a simulated accelerator resource occupancy state that can improve the accuracy of determining a simulated accelerator resource occupancy state.

In a first aspect, the present application provides a method for determining a resource occupancy state of a simulation accelerator, including:

in response to a resource state determination request for a simulation accelerator, acquiring first current data of the simulation accelerator in a first historical time interval corresponding to a current time and second current data of the simulation accelerator in a second historical time interval corresponding to the current time; the current time is the triggering time of the request for determining the resource state, and the distance between the starting time of the first historical time interval and the current time is longer than the distance between the starting time of the second historical time interval and the current time;

clustering the first current data to obtain a resource occupation classification threshold, and comparing the resource occupation classification threshold with the second current data to obtain a predicted resource occupation state of the simulation accelerator in the second historical time interval;

Acquiring a board occupation state of the simulation accelerator in the second historical time interval, and acquiring a misjudgment rate parameter associated with the resource occupation classification threshold according to the board occupation state and the predicted resource occupation state;

and under the condition that the misjudgment rate parameter meets the update termination condition of a preset resource occupation classification threshold, obtaining the resource occupation state of the simulation accelerator at the current time based on the resource occupation classification threshold and the current data corresponding to the current time.

In one embodiment, before the obtaining the preset iteration time difference value when the misjudgment rate parameter does not meet the preset condition, the method further includes: acquiring a preset iteration time difference value under the condition that the misjudgment rate parameter does not meet the preset resource occupation classification threshold updating termination condition; and updating the first historical time interval by using the iteration time difference value, acquiring current data of the simulation accelerator in the updated first historical time interval as new first current data, and returning to execute the step of obtaining the resource occupation classification threshold value based on the first current data until the misjudgment rate parameter meets the update termination condition of the preset resource occupation classification threshold value.

In one embodiment, before the obtaining the preset iteration time difference value when the misjudgment rate parameter does not meet the preset resource occupation classification threshold updating termination condition, the method further includes: acquiring the current iteration round corresponding to the misjudgment rate parameter and a preset misjudgment rate parameter threshold; determining that the misjudgment rate parameter does not meet the update termination condition of the preset resource occupation classification threshold under the condition that the misjudgment rate parameter is larger than the misjudgment rate parameter threshold and the current iteration round does not reach the set iteration round; and under the condition that the misjudgment rate parameter meets a preset condition, obtaining the resource occupation state of the simulation accelerator at the current time based on the resource occupation classification threshold and the current data corresponding to the current time, wherein the method comprises the following steps: and obtaining the resource occupation state of the simulation accelerator at the current time based on the resource occupation classification threshold and the current data under the condition that the misjudgment rate parameter is smaller than or equal to the misjudgment rate parameter threshold or the current iteration round reaches the set iteration round.

In one embodiment, the number of the second current data is a plurality; the obtaining the misjudgment rate parameter associated with the resource occupation classification threshold according to the board occupation state and the predicted resource occupation state comprises the following steps: acquiring a board occupation state and a predicted resource occupation state corresponding to each second current data, and acquiring the number of the second current data; acquiring target second current data of which the corresponding board card occupation state represents that the board card of the simulation accelerator is unoccupied and the corresponding predicted resource occupation state represents that the simulation resource of the simulation accelerator is occupied from the second current data, and acquiring the number of the target second current data; and obtaining the misjudgment rate parameter according to the number of the second current data and the number of the target second current data.

In one embodiment, the comparing the resource occupancy classification threshold with the second current data to obtain the predicted resource occupancy state of the simulation accelerator in the second historical time interval includes: acquiring current second current data, and comparing the resource occupation classification threshold with the current second current data; determining that a predicted resource occupation state corresponding to the current second current data represents that simulation resources of the simulation accelerator are occupied under the condition that the current second current data is larger than or equal to the resource occupation classification threshold value; and under the condition that the current second current data is smaller than the resource occupation classification threshold value, determining that the predicted resource occupation state corresponding to the current second current data represents that simulation resources of the simulation accelerator are unoccupied.

In one embodiment, the number of the first current data is a plurality; the obtaining a resource occupation classification threshold based on the first current data includes: acquiring the median of a plurality of first current data, and taking the median as an initial resource occupation classification threshold; dividing the plurality of first current data into first sub-current data corresponding to the occupied simulation resources and second sub-current data corresponding to the unoccupied simulation resources based on the initial resource occupancy classification threshold; obtaining a first variance value according to the first sub-current data, obtaining a second variance value according to the second sub-current data, and carrying out summation processing on the first variance value and the second variance value to obtain a variance summation result; and taking the initial resource occupation classification threshold value as the resource occupation classification threshold value under the condition that the variance summation result is smaller than or equal to a preset variance threshold value.

In one embodiment, after the variance summing result is obtained, the method further includes: updating the initial resource occupation classification threshold by using the variance summation result under the condition that the variance summation result is larger than the variance threshold, so as to obtain an updated resource occupation classification threshold; and taking the updated resource occupation classification threshold value as a new initial resource occupation classification threshold value, and returning to execute the step of dividing the plurality of first current data into first sub-current data corresponding to occupied simulation resources and second sub-current data corresponding to unoccupied simulation resources based on the initial resource occupation classification threshold value until the variance summation result is smaller than or equal to the variance threshold value.

In one embodiment, the obtaining the first current data of the simulation accelerator in a first historical time interval corresponding to a current time and the second current data in a second historical time interval corresponding to the current time includes: acquiring initial first current data corresponding to the first historical time interval and initial second current data corresponding to the second historical time interval; acquiring current data with missing values in the initial first current data and current data with missing values in the initial second current data; writing the median value of the initial first current data into the current data with the missing value in the initial first current data to obtain the completed initial first current data, and writing the median value of the initial second current data into the current data with the missing value in the initial second current data to obtain the completed initial second current data; and carrying out smooth noise reduction treatment on the initial first current data after the completion and the initial second current data after the completion to obtain the first current data and the second current data.

In one embodiment, the acquiring the first current data of the simulation accelerator in a first historical time interval corresponding to a current time and before the second current data of the simulation accelerator in a second historical time interval corresponding to the current time further includes: acquiring a preset first time interval length matched with the first historical time interval and a preset second time interval length matched with the second historical time interval; determining the starting time of the second historical time interval according to the current time and the second time interval length, and taking the time interval between the starting time of the second historical time interval and the current time as the second historical time interval; determining the starting time of the first historical time interval according to the starting time of the second historical time interval and the length of the first time interval, and taking the time interval between the starting time of the first historical time interval and the starting time of the second historical time interval as the first historical time interval.

In a second aspect, the present application further provides a device for determining a resource occupancy state of a simulation accelerator, including:

The current data acquisition module is used for responding to a resource state determination request for the simulation accelerator and acquiring first current data of the simulation accelerator in a first historical time interval corresponding to the current time and second current data of the simulation accelerator in a second historical time interval corresponding to the current time; the current time is the triggering time of the request for determining the resource state, and the distance between the starting time of the first historical time interval and the current time is longer than the distance between the starting time of the second historical time interval and the current time;

the predicted occupation obtaining module is used for obtaining a resource occupation classification threshold based on the first current data, comparing the resource occupation classification threshold with the second current data and obtaining a predicted resource occupation state of the simulation accelerator in the second historical time interval;

the misjudgment parameter acquisition module is used for acquiring the board card occupation state of the simulation accelerator in the second historical time interval, and obtaining misjudgment rate parameters associated with the resource occupation classification threshold according to the board card occupation state and the predicted resource occupation state;

the resource occupation obtaining module is used for obtaining the resource occupation state of the simulation accelerator at the current time based on the resource occupation classification threshold and the current data corresponding to the current time under the condition that the misjudgment rate parameter meets the update termination condition of the preset resource occupation classification threshold.

In a third aspect, the present application also provides a computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:

obtaining a resource occupation classification threshold based on the first current data, and comparing the resource occupation classification threshold with the second current data to obtain a predicted resource occupation state of the simulation accelerator in the second historical time interval;

In a fourth aspect, the present application also provides a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of:

In a fifth aspect, the present application also provides a computer program product comprising a computer program which, when executed by a processor, performs the steps of:

The above-described simulation accelerator resource occupancy state determination method, apparatus, computer device, storage medium, and computer program product acquire first current data of a simulation accelerator in a first historical time interval corresponding to a current time and second current data of the simulation accelerator in a second historical time interval corresponding to the current time by responding to a resource state determination request for the simulation accelerator; the current time is the triggering time of the resource state determination request, and the distance between the starting time of the first historical time interval and the current time is longer than the distance between the starting time of the second historical time interval and the current time; based on the first current data, obtaining a resource occupation classification threshold value, and comparing the resource occupation classification threshold value with the second current data to obtain a predicted resource occupation state of the simulation accelerator in a second historical time interval; acquiring the board occupation state of the simulation accelerator in a second historical time interval, and obtaining a misjudgment rate parameter associated with a resource occupation classification threshold according to the board occupation state and the predicted resource occupation state; and under the condition that the misjudgment rate parameter meets the update termination condition of the preset resource occupation classification threshold, acquiring the resource occupation state of the simulation accelerator at the current time based on the resource occupation classification threshold and the current data corresponding to the current time. According to the method and the device, when a user triggers a resource state determining request for the simulation accelerator, first current data of the simulation accelerator corresponding to a first historical time interval and second current data of the simulation accelerator corresponding to a second historical time interval can be obtained, so that a resource occupation classifying threshold value can be obtained through the first current data, the second current data and the resource occupation classifying threshold value can be compared to obtain a predicted resource occupation state of the simulation accelerator in the second historical time interval, in addition, a misjudgment rate parameter can be obtained according to a board occupation state and a predicted resource occupation state of the simulation accelerator in the second historical time interval, if the misjudgment rate parameter meets a set resource occupation classifying threshold updating termination condition, the resource occupation classifying threshold value and the current data can be utilized to obtain a simulation resource occupation state of the simulation accelerator at the current time.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the related art, the drawings that are required to be used in the embodiments or the related technical descriptions will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and other drawings may be obtained according to the drawings without inventive effort for a person having ordinary skill in the art.

FIG. 1 is a flow diagram of a method for determining the occupancy state of a simulated accelerator resource in one embodiment;

FIG. 2 is a flow diagram of updating a resource occupancy classification threshold in one embodiment;

FIG. 3 is a flow chart of obtaining a false positive rate parameter in one embodiment;

FIG. 4 is a flow chart of obtaining a resource occupancy classification threshold in one embodiment;

FIG. 5 is a flow chart of acquiring first current data and second current data according to an embodiment;

FIG. 6 is a flow chart of a method for acquiring a historical time interval according to one embodiment;

FIG. 7 is a flow diagram of a method for classifier-based simulation accelerator resource occupancy prediction in one embodiment;

FIG. 8 is a block diagram of an embodiment of a simulated accelerator resource occupancy state determination device;

Fig. 9 is an internal structural diagram of a computer device in one embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.

In one embodiment, as shown in fig. 1, a method for determining the resource occupancy state of a simulation accelerator is provided, and this embodiment is applied to a terminal for illustration by using the method, it is to be understood that the method may also be applied to a server, and may also be applied to a system including the terminal and the server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the steps of:

step S101, in response to a resource state determination request for a simulation accelerator, acquiring first current data of the simulation accelerator in a first historical time interval corresponding to the current time and second current data of the simulation accelerator in a second historical time interval corresponding to the current time; the current time is the triggering time of the resource state determination request, and the distance between the starting time of the first historical time interval and the current time is longer than the distance between the starting time of the second historical time interval and the current time.

The simulation accelerator refers to a large-scale SoC design, namely an acceleration platform which is most widely applied in system-on-a-chip design, and the resource state determination request is a request for determining the resource occupation state of the simulation accelerator, and the request can be triggered by a user to a terminal or actively triggered by a timing task in a mode of setting the timing task. The current time refers to a time for triggering a resource status determination request, the first historical time interval and the second historical time interval refer to historical time intervals associated with the current time, the first historical time interval and the second historical time interval can be determined according to the current time, the distance between the starting time of the first historical time interval and the current time is longer than the distance between the starting time of the second historical time interval and the current time, the first current data refers to current data of the first historical time interval, the second current data refers to current data of the second historical time interval, and the current data can be acquired through a current sensor of the simulation accelerator.

Specifically, when a resource state determination request for the simulation accelerator is received, the terminal may respond to the request, thereby obtaining first current data of the simulation accelerator in a first historical time interval and second current data in a second historical time interval.

Step S102, a resource occupation classification threshold is obtained based on the first current data, and the resource occupation classification threshold is compared with the second current data to obtain the predicted resource occupation state of the simulation accelerator in the second historical time interval.

The resource occupation classification threshold refers to a classification threshold for distinguishing the resource occupation state of the simulation accelerator, the classification threshold may be a certain current data threshold, and the terminal may determine the resource occupation state of the simulation accelerator according to the magnitude relation between the current data and the resource occupation classification threshold. The predicted resource occupancy state refers to a resource occupancy state of the simulation accelerator in a second historical time interval output by the terminal, and the terminal can be determined by comparing the second current data with a resource occupancy classification threshold.

Step S103, obtaining the board occupation state of the simulation accelerator in the second historical time interval, and obtaining the misjudgment rate parameter associated with the resource occupation classification threshold according to the board occupation state and the predicted resource occupation state.

The board card occupation state refers to a state of whether board card resources of the simulation accelerator are occupied or not, the state can be determined by a terminal by detecting whether the resources of the board card of the simulation accelerator are locked or not, and the misjudgment rate parameter is a parameter representing whether the resource occupation state of the simulation accelerator is misjudged or not.

Specifically, the terminal may further collect a board occupation state of the simulation accelerator in the second historical time interval, so as to obtain a misjudgment rate parameter associated with the resource occupation classification threshold according to the board occupation state and a predicted resource occupation state of the simulation accelerator in the second historical time interval.

Step S104, under the condition that the misjudgment rate parameter meets the update termination condition of the preset resource occupation classification threshold, acquiring the resource occupation state of the simulation accelerator at the current time based on the resource occupation classification threshold and the current data corresponding to the current time.

The current data refers to current data acquired by a current sensor of the simulation accelerator at the current time, and finally, if the misjudgment rate meets a preset condition, for example, when the misjudgment rate parameter is smaller, the terminal can obtain the resource occupation state of the simulation accelerator at the current time according to the resource occupation classification threshold value and the current data corresponding to the current time.

In the method for determining the resource occupation state of the simulation accelerator, the first current data of the simulation accelerator in a first historical time interval corresponding to the current time and the second current data of the simulation accelerator in a second historical time interval corresponding to the current time are obtained by responding to a resource state determination request of the simulation accelerator; the current time is the triggering time of the resource state determination request, and the distance between the starting time of the first historical time interval and the current time is longer than the distance between the starting time of the second historical time interval and the current time; based on the first current data, obtaining a resource occupation classification threshold value, and comparing the resource occupation classification threshold value with the second current data to obtain a predicted resource occupation state of the simulation accelerator in a second historical time interval; acquiring the board occupation state of the simulation accelerator in a second historical time interval, and obtaining a misjudgment rate parameter associated with a resource occupation classification threshold according to the board occupation state and the predicted resource occupation state; and under the condition that the misjudgment rate parameter meets the update termination condition of the preset resource occupation classification threshold, acquiring the resource occupation state of the simulation accelerator at the current time based on the resource occupation classification threshold and the current data corresponding to the current time. According to the method and the device, when a user triggers a resource state determining request for the simulation accelerator, first current data of the simulation accelerator corresponding to a first historical time interval and second current data of the simulation accelerator corresponding to a second historical time interval can be obtained, so that a resource occupation classifying threshold value can be obtained through the first current data, the second current data and the resource occupation classifying threshold value can be compared to obtain a predicted resource occupation state of the simulation accelerator in the second historical time interval, in addition, a misjudgment rate parameter can be obtained according to a board occupation state and a predicted resource occupation state of the simulation accelerator in the second historical time interval, if the misjudgment rate parameter meets a set resource occupation classifying threshold updating termination condition, the resource occupation classifying threshold value and the current data can be utilized to obtain a simulation resource occupation state of the simulation accelerator at the current time.

In one embodiment, as shown in fig. 2, after step S103, the method may further include:

step S201, obtaining a preset iteration time difference value under the condition that the misjudgment rate parameter does not meet the update termination condition of the preset resource occupation classification threshold.

The iteration time difference refers to a preset time difference for iteration, and in this embodiment, if the misjudgment rate parameter determined by the terminal does not meet the update termination condition of the preset resource occupation classification threshold, that is, if the misjudgment rate parameter indicates that the misjudgment rate of the resource occupation state is larger, the terminal may further perform iterative update processing on the resource occupation classification threshold for dividing the predicted resource occupation state. Therefore, the terminal can firstly acquire the preset iteration time difference value when the misjudgment rate parameter does not meet the preset condition.

Step S202, updating the first historical time interval by using the iteration time difference value, acquiring current data of the simulation accelerator in the updated first historical time interval as new first current data, and returning to execute step S102 until the misjudgment rate parameter meets the update termination condition of the preset resource occupation classification threshold.

After the iteration time difference is obtained, the terminal can update the first historical time interval by using the iteration time difference, for example, the iteration time difference can be 4 hours, and then the terminal can update the first historical time interval by extending the first historical time interval by 4 hours, thereby obtaining updated first historical time. And the terminal can acquire the updated current data of the first historical time interval through the current sensor to serve as new first current data, and then the updated resource occupation classification threshold value can be obtained by utilizing the new first current data again until the determined misjudgment rate meets the set resource occupation classification threshold value updating termination condition.

In this embodiment, if the misjudgment rate parameter does not meet the preset resource occupation classification threshold updating termination condition, the terminal may further update the first historical time interval and the first current data of the first historical time interval by acquiring a preset iteration time difference value, and implement iterative update of the misjudgment rate parameter by using the new first current data.

In addition, before step S201, the method may further include: acquiring a current iteration round corresponding to the false positive rate parameter and a preset false positive rate parameter threshold; under the condition that the misjudgment rate parameter is larger than the misjudgment rate parameter threshold and the current iteration round does not reach the set iteration round, determining that the misjudgment rate parameter does not meet the update termination condition of the preset resource occupation classification threshold; step S104 may further include: and under the condition that the misjudgment rate parameter is smaller than or equal to the misjudgment rate parameter threshold, or the current iteration round reaches the set iteration round, acquiring the resource occupation state of the simulation accelerator at the current time based on the resource occupation classification threshold and the current data.

The judgment that the false positive rate parameter meets the preset condition can be divided into two parts, and whether the false positive rate parameter is smaller than or equal to a certain false positive rate parameter threshold value or not and whether the current iteration round reaches the set maximum iteration round or not. In this embodiment, after obtaining the misjudgment rate parameter, the terminal may further obtain a current iteration round and a preset misjudgment rate parameter threshold, and then determine that the misjudgment rate parameter does not meet a preset resource occupation classification threshold updating termination condition if the misjudgment rate parameter is still greater than the misjudgment rate parameter threshold and the current iteration round does not reach the preset iteration round.

Similarly, if the misjudgment rate parameter is smaller than or equal to the misjudgment rate parameter threshold, or the current iteration round has reached the set iteration round, then the misjudgment rate parameter is indicated to meet the preset condition, and the like, and the terminal can obtain the resource occupation state of the simulation accelerator at the current time according to the resource occupation classification threshold of the current iteration round and the current data.

In this embodiment, the conditions for completing the iteration may include a misjudgment rate parameter condition and an iteration round condition, and when the misjudgment rate parameter is smaller than or equal to the misjudgment rate parameter threshold, or the current iteration round reaches the set iteration round, that is, the misjudgment rate parameter satisfies the preset condition, the efficiency of the iterative operation processing may be improved on the premise of ensuring the accuracy of the misjudgment rate parameter.

In one embodiment, the number of second current data is a plurality; as shown in fig. 3, step S103 may further include:

step S301, a board occupation state and a predicted resource occupation state corresponding to each second current data are obtained, and the number of the second current data is obtained.

In this embodiment, the number of the second current data may be plural, and the current data corresponding to different time points in the second historical time interval respectively, and the board occupation state corresponding to each second current data refers to the board occupation state corresponding to each time point in the second historical time interval. And, the terminal may also count the number of the second current data.

Step S302, obtaining target second current data of which the corresponding board card occupation state represents that the board card of the simulation accelerator is unoccupied and the corresponding predicted resource occupation state represents that the simulation resource of the simulation accelerator is occupied from the second current data, and obtaining the number of the target second current data.

The target second current data refers to second current data in which the corresponding board card occupation state represents that the board card of the simulation accelerator is unoccupied, and the predicted resource occupation state represents that the simulation resource of the simulation accelerator is occupied.

In general, the board occupation state and the predicted resource occupation state, which generally correspond to the second current data, may generally include the following four situations, which are respectively the board occupation state and represent the board unoccupied state, while the predicted resource occupation state is the situation that the simulation resource is unoccupied, in which case, it may be indicated that the board is unoccupied, and meanwhile, no simulation operation is occupied, which belongs to the normal situation. Secondly, the occupied state of the board can be represented, and the situation that the board is occupied can be predicted, wherein the resource occupied state is the situation that the simulation resource is occupied, and in the situation, the fact that the board resource is actually occupied by the simulation operation can be indicated, so that the situation is normal. Meanwhile, the board occupation state can also be used for representing the occupied state of the board, and the resource occupation state is predicted to be the situation that the simulation resource is unoccupied, in this case, the user may occupy the board resource, but does not perform actual simulation operation, so that the situation also belongs to the normal situation. And only the occupied state of the board card represents that the board card is not occupied, and the situation that the resource occupied state is occupied by the simulation resource is predicted at the same time, which belongs to an abnormal situation, so that the terminal can take the second current data of the identified situation as target second current data and count the number of the target second current data.

Step S303, obtaining the misjudgment rate parameter according to the number of the second current data and the number of the target second current data.

And finally, the terminal can take the ratio of the number of the target second current data and the number of the second current data as a misjudgment rate parameter.

In this embodiment, the terminal may obtain the misjudgment rate parameter by collecting the number of the second current data and the number of the second current data, where the board card of the board card occupation state indicates that the simulation accelerator is unoccupied, and predicting the number of the second current data, where the simulation resource of the resource occupation state indicates that the simulation accelerator is occupied.

Further, step S102 may further include: acquiring current second current data, and comparing the resource occupation classification threshold with the current second current data; under the condition that the current second current data is larger than or equal to a resource occupation classification threshold value, determining that the predicted resource occupation state corresponding to the current second current data represents that simulation resources of the simulation accelerator are occupied; and under the condition that the current second current data is smaller than the resource occupation classification threshold value, determining that the predicted resource occupation state corresponding to the current second current data represents that the simulation resource of the simulation accelerator is unoccupied.

In this embodiment, the terminal may determine the predicted resource occupancy state corresponding to each current second current data by comparing each current second current data with the resource occupancy classification threshold, respectively. Specifically, if the current second current data is greater than or equal to the resource occupancy classification threshold, the terminal may determine a predicted resource occupancy state corresponding to the current second current data, which characterizes that the simulation resources of the simulation accelerator are occupied, and if the current second current data is less than the resource occupancy classification threshold, the terminal may determine a predicted resource occupancy state corresponding to the current second current data, which characterizes that the simulation resources of the simulation accelerator are unoccupied. By the mode, the terminal can obtain the predicted resource occupation state corresponding to each piece of current second current data.

In this embodiment, the terminal may directly compare each current second current data with the resource occupation classification threshold, so as to obtain, according to the comparison result, a predicted resource occupation state corresponding to each current second current data, and in this manner, the efficiency of obtaining the predicted resource occupation state may be improved.

In one embodiment, the number of first current data is a plurality; as shown in fig. 4, step S102 may further include:

in step S401, a median of the plurality of first current data is obtained, and the median is used as an initial resource occupation classification threshold.

The initial resource occupation classification threshold refers to an unchecked resource occupation classification threshold, in this embodiment, similar to the second current data, the number of the first current data may also be plural, which corresponds to different time points of the first history time interval, respectively.

Step S402, based on the initial resource occupation classification threshold, dividing the plurality of first current data into first sub-current data corresponding to the occupied simulation resource and second sub-current data corresponding to the unoccupied simulation resource.

The first sub-current data refers to first current data representing that the simulation resource is occupied, and the second sub-current data refers to first current data representing that the simulation resource is unoccupied.

Step S403, obtaining a first variance value according to the first sub-current data, obtaining a second variance value according to the second sub-current data, and summing the first variance value and the second variance value to obtain a variance summation result.

The first variance value refers to a variance value calculated by each first sub-current data, and the second variance value refers to a variance value calculated by each second sub-current data.

In step S404, when the variance summation result is less than or equal to the variance threshold set in advance, the initial resource occupation classification threshold is used as the resource occupation classification threshold.

Then, if the variance summation result is smaller than or equal to the preset variance threshold, that is, the variance summation result is smaller, the initial resource occupancy classification threshold may be regarded as the final resource occupancy classification threshold.

In this embodiment, whether the initial resource occupation classification threshold is accurate or not may be verified by setting the initial resource occupation classification threshold, dividing the first current data by the initial resource occupation classification threshold, and calculating the variance for the divided first current data.

Further, after step S403, the method may further include: under the condition that the variance summation result is larger than the variance threshold, updating the initial resource occupation classification threshold by utilizing the variance summation result to obtain an updated resource occupation classification threshold; and taking the updated resource occupation classification threshold value as a new initial resource occupation classification threshold value, and returning to the step S402 until the variance summation result is smaller than or equal to the variance threshold value.

And if the variance summation result is larger than the variance threshold, the terminal needs to update the initial resource occupation classification threshold by utilizing the variance summation result, and after the updated resource occupation classification threshold is obtained, the updated resource occupation classification threshold is used as a new initial resource occupation classification threshold to be subjected to iterative processing, so that the first current data is divided by utilizing the new initial resource occupation classification threshold again until the variance summation result is smaller than or equal to the variance threshold, and the initial resource occupation classification threshold iterated finally is used as a final resource occupation classification threshold.

In this embodiment, if the variance summation result is greater than the variance threshold, the terminal may iteratively update the initial resource occupation classification threshold by using the variance summation result, and by this way, the accuracy of acquiring the resource occupation classification threshold may be further improved.

In one embodiment, as shown in fig. 5, step S101 may further include:

in step S501, initial first current data corresponding to a first historical time interval and initial second current data corresponding to a second historical time interval are obtained through a current sensor associated with the simulation accelerator.

The initial first current data refers to current data corresponding to a first historical time interval directly obtained by a current sensor associated with the simulation accelerator, and the initial second current data refers to current data corresponding to a second historical time interval directly obtained by the current sensor. In this embodiment, the terminal may directly obtain, by emulating the current sensor associated with the accelerator, current data corresponding to the first historical time interval as initial first current data, and may also directly obtain, by using the current sensor, current data corresponding to the second historical time interval as initial second current data.

Step S502, current data with missing values in the initial first current data and current data with missing values in the initial second current data are obtained.

The missing value of the current data refers to the current data lacking the data value, in this embodiment, the initial first current data or the initial second current data acquired by the current sensor may have the current data lacking the data value, and in this embodiment, the terminal may determine the missing value of the current data from the initial first current data and the initial second current data.

Step S503, writing the median value of the initial first current data into the current data with the missing value in the initial first current data to obtain the completed initial first current data, and writing the median value of the initial second current data into the current data with the missing value in the initial second current data to obtain the completed initial second current data.

Then, in order to ensure the integrity of the data values in the initial first current data and the initial second current data, the terminal may also write the median value of the initial first current data and the median value of the initial second current data into the missing current data in the initial first current data and the initial second current data, so as to form the completed initial first current data and the completed initial second current data.

Step S504, smoothing and denoising the complemented initial first current data and the complemented initial second current data through a low-pass filter to obtain first current data and second current data.

And finally, after completing the complementation of the initial first current data and the initial second current data, the terminal can also perform low-pass filtering on the completed initial first current data and the completed initial second current data through a low-pass filter, so that smooth noise reduction processing is realized, and final first current data and second current data are obtained.

In this embodiment, after obtaining the initial first current data and the initial second current data obtained by the current sensor, the terminal may further perform missing value filling and smoothing noise reduction processing on the initial first current data and the initial second current data to obtain final first current data and second current data, which may further improve the integrity and accuracy of the first current data and the second current data.

In one embodiment, as shown in fig. 6, before step S101, the method may further include:

step S601, a preset first time interval length matched with the first historical time interval and a preset second time interval length matched with the second historical time interval are obtained.

The first time interval length refers to a time interval length of a first historical time interval, and the second time interval length refers to a time interval length of a second historical time interval. For example, the second time interval length may be set to 4 hours, and then the first time interval length may be set to 44 hours.

Step S602, determining the start time of the second historical time interval according to the current time and the second time interval length, and taking the time interval between the start time of the second historical time interval and the current time as the second historical time interval.

The starting time of the second historical time interval refers to the starting endpoint time corresponding to the second historical time interval, in this embodiment, the second historical time interval refers to the starting time of the second historical time interval, and the time interval from the starting time of the second historical time interval to the current time is taken as the second historical time interval. For example, the second time interval is 4 hours, and after the current time is obtained, the terminal may advance the current time by 4 hours as the start time of the second historical time interval, so that the first 4 hours of the current time is taken as the second historical time interval.

Step S603, determining a start time of the first historical time interval according to the start time of the second historical time interval and the first time interval length, and taking a time interval between the start time of the first historical time interval and the start time of the second historical time interval as the first historical time interval.

The start time of the first historical time interval refers to the start endpoint time corresponding to the first historical time interval, in this embodiment, the first historical time interval refers to the time interval from the start time of the first historical time interval to the start time of the second historical time interval, and after determining the start time and the first time interval length of the second historical time interval, the terminal can calculate the start time of the first historical time interval according to the start time and the first time interval length of the second historical time interval, so that a time interval between the start time of the first historical time interval and the start time of the second historical time interval is used as the first historical time interval. For example, the second time interval is 44 hours, and after obtaining the start time of the second historical time interval, the terminal may advance the current time by 44 hours as the start time of the first historical time interval, so that a time interval from the first 48 hours of the current time to the first 4 hours of the current time is used as the first historical time interval.

In this embodiment, the terminal may further determine the first historical time interval and the second historical time interval based on the set first time interval length and the second time interval length, and by this way, the acquisition efficiency of the first historical time interval and the second historical time interval may be further improved.

In one embodiment, a method for predicting the resource occupation of a simulation accelerator based on a classifier is also provided, the method can analyze and process related parameters of the running load of the accelerator, calculate the actual utilization rate of resources, and generate a critical value for reflecting the actual load change and the logic duty ratio of the board card current by generating critical value for the simulation accelerator current sensor data, thereby providing the actual running load with reference value for operation and maintenance managers and facilitating planning and coordination of limited and expensive calculation resources.

As shown in fig. 7, the method is a process flow of the overall current sensor data of the single board card. The input of the system is the source data of the self-sampling service, and the output is the threshold value obtained by the system through processing of a machine learning method. The system performs the above analysis flow on each board card in the simulation accelerator, taking a logic board with the number XX as an example, and the detailed flow is introduced as follows:

Step 1-3: and (5) preprocessing data.

Step 1: the default selects two time intervals to obtain data, the value is a preset initial value of training iteration, wherein t ₂ T is 48 hours before the current time ₁ T is 4 hours before the current time ₀ For the current time, the time difference order t of the subsequent iteration loop _step For 4 hours.

Step 2: dividing data into a training data set and a verification data set according to time intervals, wherein a longer interval in the data is the training data set, namely t ₂ -t ₁ The data in this time interval is identified as D as a training set of machine learning classifiers _xx Train{t ₂ -t ₁ Simultaneously defining the latest historical data interval as a verification data set, namely t ₁ -t ₀ The data in this time interval is identified as D as a check set for evaluating the classifier estimation results _xx Valid{t ₁ -t ₀ }。

Step 3: the validity of the selected data set is checked, because the input data belongs to a time sequence and is limited by the problems of stability of a sampling process, asynchronism of the system time of a main control server of the simulation accelerator and the like, the obtained data has occasional breakpoints, dislocation and other anomalies, the input data is sensor original data, and the accuracy of an algorithm is influenced by peak fluctuation caused by instantaneous current and the like.

Thus, for the two-point problem above, the validity check of step 3 will be performed by the following sequence:

(1) Removing data points missing logical occupancy states or sensor values;

(2) Filling the missing data points of both the previous using the median of the complete data set;

(3) And smoothing and denoising the complete data set by adopting low-pass filtering.

Step 4: and constructing a classifier.

And the correction processing of the estimated threshold value is realized through the adaptive iteration of the limited loop. Wherein less than the threshold is idle and greater than or equal to the threshold is occupied, by selecting the initial value as the median of the training data set toThe variance of the initial iteration is minimized. The algorithm will then adjust the iteration of the estimated threshold by calculating the Manhattan distance until convergence, generating a decision threshold r _xx 。

The algorithm then makes a threshold decision on the speculative value V for the validation dataset _xx {t ₁ -t ₀ The calculation is as follows:

e＝D _xx Valid{t ₁ -t ₀ }-r _xx

when e<0，V _xx {t ₁ -t ₀ 0, indicating that the simulation accelerator is in an idle state, when e is greater than or equal to 0, V _xx {t ₁ -t ₀ And 0, indicating that the simulation accelerator is in an occupied state.

Step 5: truth table and second iteration.

Will verify the data set V _xx {t ₁ -t ₀ The sequence of logical occupancies and speculative actual occupancies in a filter truth table, which may be as shown in table 1 below:

TABLE 1 Board occupancy truth table

The corresponding relation is as follows:

when the truth table returns 0 to the current data point, judging that the board card is free from occupation of operation;

when 1 is returned to the current data point, judging that the current data point is occupied, and the computing resource is idle;

when the current data point returns to 2, the current data point is judged to be actually occupied, and the board card is busy;

and when returning to the X state, misjudgment is performed.

For the probability of returning to the X state in the truth table filtering in the time sequence, generating a fault rate parameter err _xx ：

err _xx ＝X(V _xx {t ₁ -t ₀ })/L{t ₁ -t ₀ }*100％

Setting a threshold value for the misjudgment rate parameter, and adjusting the current learning operation when the misjudgment rate parameter is larger than the misjudgment rate threshold valueHistorical data acquisition section, let t in step 1 ₂ ＝t ₂ -t _step 。

And (3) entering a secondary iteration process outside the current construction model, and repeating until the misjudgment rate is lower than a preset value or the maximum iteration times are reached. When the error judgment rate is lower than the error judgment rate, the system adopts and returns the learned judgment threshold value n _xx When the current learning operation reaches the maximum iteration number, the system will give up learning to avoid being limited to the non-convergent loop, and adopts the last learned or preset decision threshold n _xx 。

In this embodiment, the actual utilization rate of the computing resources of the accelerator board card can be effectively reflected through the dynamic load determination threshold value and the truth table generated by the algorithm, so that the workflow of the simulation operation can be more conveniently and reasonably adjusted.

It should be understood that, although the steps in the flowcharts related to the embodiments described above are sequentially shown as indicated by arrows, these steps are not necessarily sequentially performed in the order indicated by the arrows. The steps are not strictly limited to the order of execution unless explicitly recited herein, and the steps may be executed in other orders. Moreover, at least some of the steps in the flowcharts described in the above embodiments may include a plurality of steps or a plurality of stages, which are not necessarily performed at the same time, but may be performed at different times, and the order of the steps or stages is not necessarily performed sequentially, but may be performed alternately or alternately with at least some of the other steps or stages.

Based on the same inventive concept, the embodiment of the application also provides a simulation accelerator resource occupation state determining device for realizing the above related simulation accelerator resource occupation state determining method. The implementation of the solution provided by the device is similar to the implementation described in the above method, so the specific limitation in the embodiments of the device for determining the resource occupancy state of one or more simulation accelerators provided below may be referred to the limitation of the method for determining the resource occupancy state of a simulation accelerator hereinabove, and will not be repeated here.

In one embodiment, as shown in fig. 8, there is provided a simulation accelerator resource occupancy state determining apparatus, including: a current data acquisition module 801, a predicted occupation acquisition module 802, a misjudgment parameter acquisition module 803, and a resource occupation acquisition module 804, wherein:

a current data obtaining module 801, configured to obtain, in response to a resource status determination request for a simulation accelerator, first current data of the simulation accelerator in a first historical time interval corresponding to a current time, and second current data of the simulation accelerator in a second historical time interval corresponding to the current time; the current time is the triggering time of the resource state determination request, and the distance between the starting time of the first historical time interval and the current time is longer than the distance between the starting time of the second historical time interval and the current time;

the predicted occupation obtaining module 802 is configured to obtain a resource occupation classification threshold based on the first current data, and compare the resource occupation classification threshold with the second current data to obtain a predicted resource occupation state of the simulation accelerator in a second historical time interval;

the misjudgment parameter obtaining module 803 is configured to obtain a board occupation state of the simulation accelerator in a second historical time interval, and obtain a misjudgment rate parameter associated with a resource occupation classification threshold according to the board occupation state and a predicted resource occupation state;

The resource occupation obtaining module 804 is configured to obtain, when the misjudgment rate parameter meets a preset resource occupation classification threshold update termination condition, a resource occupation state of the simulation accelerator at the current time based on the resource occupation classification threshold and current data corresponding to the current time.

In one embodiment, the misjudgment parameter obtaining module 803 is further configured to obtain a preset iteration time difference value when the misjudgment rate parameter does not meet the preset resource occupation classification threshold updating termination condition; updating the first historical time interval by using the iteration time difference value, acquiring current data of the simulation accelerator in the updated first historical time interval as new first current data, and returning to execute the step of obtaining the resource occupation classification threshold value based on the first current data until the misjudgment rate parameter meets the update termination condition of the preset resource occupation classification threshold value.

In one embodiment, the misjudgment parameter obtaining module 803 is further configured to obtain a current iteration round corresponding to the misjudgment rate parameter and a preset misjudgment rate parameter threshold; under the condition that the misjudgment rate parameter is larger than the misjudgment rate parameter threshold and the current iteration round does not reach the set iteration round, determining that the misjudgment rate parameter does not meet the update termination condition of the preset resource occupation classification threshold; the resource occupation obtaining module 804 is further configured to obtain, based on the resource occupation classification threshold and the current data, a resource occupation state of the simulation accelerator at the current time when the misjudgment rate parameter is less than or equal to the misjudgment rate parameter threshold, or the current iteration round reaches the set iteration round.

In one embodiment, the number of second current data is a plurality; the misjudgment parameter obtaining module 803 is further configured to obtain a board occupation state and a predicted resource occupation state corresponding to each second current data, and obtain the number of the second current data; acquiring target second current data of which the corresponding board card occupation state represents that the board card of the simulation accelerator is unoccupied and the corresponding predicted resource occupation state represents that the simulation resource of the simulation accelerator is occupied from the second current data, and acquiring the number of the target second current data; and obtaining the misjudgment rate parameter according to the number of the second current data and the number of the target second current data.

In one embodiment, the predicted occupancy obtaining module 802 is further configured to obtain current second current data, and compare the resource occupancy classification threshold with the current second current data; under the condition that the current second current data is larger than or equal to a resource occupation classification threshold value, determining that the predicted resource occupation state corresponding to the current second current data represents that simulation resources of the simulation accelerator are occupied; and under the condition that the current second current data is smaller than the resource occupation classification threshold value, determining that the predicted resource occupation state corresponding to the current second current data represents that the simulation resource of the simulation accelerator is unoccupied.

In one embodiment, the number of first current data is a plurality; the predicted occupation obtaining module 802 is further configured to obtain a median of the plurality of first current data, and take the median as an initial resource occupation classification threshold; dividing the plurality of first current data into first sub-current data corresponding to occupied simulation resources and second sub-current data corresponding to unoccupied simulation resources based on an initial resource occupancy classification threshold; obtaining a first variance value according to each first sub-current data, obtaining a second variance value according to each second sub-current data, and carrying out summation processing on the first variance value and the second variance value to obtain a variance summation result; and taking the initial resource occupation classification threshold as the resource occupation classification threshold under the condition that the variance summation result is smaller than or equal to a preset variance threshold.

In one embodiment, the predicted occupation obtaining module 802 is further configured to update the initial resource occupation classification threshold with the variance summation result if the variance summation result is greater than the variance threshold, to obtain an updated resource occupation classification threshold; and taking the updated resource occupation classification threshold value as a new initial resource occupation classification threshold value, and returning to execute the step of dividing the plurality of first current data into first sub-current data corresponding to occupied simulation resources and second sub-current data corresponding to unoccupied simulation resources based on the initial resource occupation classification threshold value until the variance summation result is smaller than or equal to the variance threshold value.

In one embodiment, the current data obtaining module 801 is further configured to obtain, by using a current sensor associated with the simulation accelerator, initial first current data corresponding to a first historical time interval, and initial second current data corresponding to a second historical time interval; acquiring current data of missing values in the initial first current data and current data of missing values in the initial second current data; writing the median value of the initial first current data into the current data with the missing value in the initial first current data to obtain the initial first current data after being complemented, and writing the median value of the initial second current data into the current data with the missing value in the initial second current data to obtain the initial second current data after being complemented; and carrying out smooth noise reduction treatment on the initial first current data after completion and the initial second current data after completion through a low-pass filter to obtain the first current data and the second current data.

In one embodiment, the current data obtaining module 801 is further configured to obtain a preset first time interval length that matches the first historical time interval, and a second time interval length that matches the second historical time interval; determining the starting time of a second historical time interval according to the current time and the second time interval length, and taking the time interval between the starting time of the second historical time interval and the current time as the second historical time interval; determining the starting time of the first historical time interval according to the starting time of the second historical time interval and the length of the first time interval, and taking the time interval between the starting time of the first historical time interval and the starting time of the second historical time interval as the first historical time interval.

The above-mentioned each module in the simulation accelerator resource occupation state determination device may be implemented in whole or in part by software, hardware, and combinations thereof. The above modules may be embedded in hardware or may be independent of a processor in the computer device, or may be stored in software in a memory in the computer device, so that the processor may call and execute operations corresponding to the above modules.

In an exemplary embodiment, a computer device, which may be a terminal, is provided, and an internal structure thereof may be as shown in fig. 9. The computer device includes a processor, a memory, an input/output interface, a communication interface, a display unit, and an input means. The processor, the memory and the input/output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input/output interface. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input/output interface of the computer device is used to exchange information between the processor and the external device. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless mode can be realized through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. The computer program, when executed by a processor, implements a method for determining the resource occupancy state of a simulation accelerator. The display unit of the computer device is used for forming a visual picture, and can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, can also be a key, a track ball or a touch pad arranged on the shell of the computer equipment, and can also be an external keyboard, a touch pad or a mouse and the like.

It will be appreciated by those skilled in the art that the structure shown in fig. 9 is merely a block diagram of a portion of the structure associated with the present application and is not limiting of the computer device to which the present application applies, and that a particular computer device may include more or fewer components than shown, or may combine some of the components, or have a different arrangement of components.

In an embodiment, there is also provided a computer device comprising a memory and a processor, the memory having stored therein a computer program, the processor implementing the steps of the method embodiments described above when the computer program is executed.

In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored which, when executed by a processor, carries out the steps of the method embodiments described above.

In an embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, implements the steps of the method embodiments described above.

It should be noted that, the user information (including, but not limited to, user equipment information, user personal information, etc.) and the data (including, but not limited to, data for analysis, stored data, presented data, etc.) referred to in the present application are information and data authorized by the user or sufficiently authorized by each party, and the collection, use, and processing of the related data are required to meet the related regulations.

Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, database, or other medium used in the various embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high density embedded nonvolatile Memory, resistive random access Memory (ReRAM), magnetic random access Memory (Magnetoresistive Random Access Memory, MRAM), ferroelectric Memory (Ferroelectric Random Access Memory, FRAM), phase change Memory (Phase Change Memory, PCM), graphene Memory, and the like. Volatile memory can include random access memory (Random Access Memory, RAM) or external cache memory, and the like. By way of illustration, and not limitation, RAM can be in the form of a variety of forms, such as static random access memory (Static Random Access Memory, SRAM) or dynamic random access memory (Dynamic Random Access Memory, DRAM), and the like. The databases referred to in the various embodiments provided herein may include at least one of relational databases and non-relational databases. The non-relational database may include, but is not limited to, a blockchain-based distributed database, and the like. The processors referred to in the embodiments provided herein may be general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic units, quantum computing-based data processing logic units, etc., without being limited thereto.

The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.

The above examples only represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the present application. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application shall be subject to the appended claims.

Claims

1. A method for determining the occupancy state of a simulation accelerator resource, the method comprising:

2. The method according to claim 1, further comprising, after said obtaining the misjudgment rate parameter associated with the resource occupancy classification threshold:

acquiring a preset iteration time difference value under the condition that the misjudgment rate parameter does not meet the preset resource occupation classification threshold updating termination condition;

and updating the first historical time interval by using the iteration time difference value, acquiring current data of the simulation accelerator in the updated first historical time interval as new first current data, and returning to execute the step of obtaining the resource occupation classification threshold value based on the first current data until the misjudgment rate parameter meets the update termination condition of the preset resource occupation classification threshold value.

3. The method according to claim 2, wherein, before obtaining the preset iteration time difference value, if the misjudgment rate parameter does not meet the preset resource occupation classification threshold updating termination condition, the method further comprises:

acquiring the current iteration round corresponding to the misjudgment rate parameter and a preset misjudgment rate parameter threshold;

determining that the misjudgment rate parameter does not meet the update termination condition of the preset resource occupation classification threshold under the condition that the misjudgment rate parameter is larger than the misjudgment rate parameter threshold and the current iteration round does not reach the set iteration round;

and under the condition that the misjudgment rate parameter meets a preset condition, obtaining the resource occupation state of the simulation accelerator at the current time based on the resource occupation classification threshold and the current data corresponding to the current time, wherein the method comprises the following steps:

and obtaining the resource occupation state of the simulation accelerator at the current time based on the resource occupation classification threshold and the current data under the condition that the misjudgment rate parameter is smaller than or equal to the misjudgment rate parameter threshold or the current iteration round reaches the set iteration round.

4. The method of claim 1, wherein the second current data is a plurality of numbers;

the obtaining the misjudgment rate parameter associated with the resource occupation classification threshold according to the board occupation state and the predicted resource occupation state comprises the following steps:

acquiring a board occupation state and a predicted resource occupation state corresponding to each second current data, and acquiring the number of the second current data;

acquiring target second current data of which the corresponding board card occupation state represents that the board card of the simulation accelerator is unoccupied and the corresponding predicted resource occupation state represents that the simulation resource of the simulation accelerator is occupied from the second current data, and acquiring the number of the target second current data;

and obtaining the misjudgment rate parameter according to the number of the second current data and the number of the target second current data.

5. The method of claim 4, wherein comparing the resource occupancy classification threshold with the second current data to obtain the predicted resource occupancy state of the simulation accelerator during the second historical time interval comprises:

Acquiring current second current data, and comparing the resource occupation classification threshold with the current second current data;

determining that a predicted resource occupation state corresponding to the current second current data represents that simulation resources of the simulation accelerator are occupied under the condition that the current second current data is larger than or equal to the resource occupation classification threshold value;

and under the condition that the current second current data is smaller than the resource occupation classification threshold value, determining that the predicted resource occupation state corresponding to the current second current data represents that simulation resources of the simulation accelerator are unoccupied.

6. The method of claim 1, wherein the first current data is a plurality of numbers; the obtaining a resource occupation classification threshold based on the first current data includes:

acquiring the median of a plurality of first current data, and taking the median as an initial resource occupation classification threshold;

dividing the plurality of first current data into first sub-current data corresponding to the occupied simulation resources and second sub-current data corresponding to the unoccupied simulation resources based on the initial resource occupancy classification threshold;

Obtaining a first variance value according to the first sub-current data, obtaining a second variance value according to the second sub-current data, and carrying out summation processing on the first variance value and the second variance value to obtain a variance summation result;

and taking the initial resource occupation classification threshold value as the resource occupation classification threshold value under the condition that the variance summation result is smaller than or equal to a preset variance threshold value.

7. The method of claim 6, wherein after obtaining the variance summation result, further comprising:

updating the initial resource occupation classification threshold by using the variance summation result under the condition that the variance summation result is larger than the variance threshold, so as to obtain an updated resource occupation classification threshold;

and taking the updated resource occupation classification threshold value as a new initial resource occupation classification threshold value, and returning to execute the step of dividing the plurality of first current data into first sub-current data corresponding to occupied simulation resources and second sub-current data corresponding to unoccupied simulation resources based on the initial resource occupation classification threshold value until the variance summation result is smaller than or equal to the variance threshold value.

8. The method of claim 1, wherein the obtaining first current data for the simulation accelerator during a first historical time interval corresponding to a current time and second current data for the simulation accelerator during a second historical time interval corresponding to the current time comprises:

acquiring initial first current data corresponding to the first historical time interval and initial second current data corresponding to the second historical time interval;

acquiring current data with missing values in the initial first current data and current data with missing values in the initial second current data;

writing the median value of the initial first current data into the current data with the missing value in the initial first current data to obtain the completed initial first current data, and writing the median value of the initial second current data into the current data with the missing value in the initial second current data to obtain the completed initial second current data;

and carrying out smooth noise reduction treatment on the initial first current data after the completion and the initial second current data after the completion to obtain the first current data and the second current data.

9. The method of any of claims 1 to 8, wherein the obtaining the first current data of the simulation accelerator for a first historical time interval corresponding to a current time and before the second current data for a second historical time interval corresponding to the current time further comprises:

acquiring a preset first time interval length matched with the first historical time interval and a preset second time interval length matched with the second historical time interval;

determining the starting time of the second historical time interval according to the current time and the second time interval length, and taking the time interval between the starting time of the second historical time interval and the current time as the second historical time interval;

determining the starting time of the first historical time interval according to the starting time of the second historical time interval and the length of the first time interval, and taking the time interval between the starting time of the first historical time interval and the starting time of the second historical time interval as the first historical time interval.

10. An apparatus for determining a state of occupancy of a resource by a simulation accelerator, the apparatus comprising:

11. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor implements the steps of the method of any one of claims 1 to 9 when the computer program is executed.

12. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any of claims 1 to 9.

13. A computer program product comprising a computer program, characterized in that the computer program, when being executed by a processor, implements the steps of the method of any one of claims 1 to 9.