CN112395059A - CMP task scheduling method for improving firefly algorithm - Google Patents
CMP task scheduling method for improving firefly algorithm Download PDFInfo
- Publication number
- CN112395059A CN112395059A CN202011279152.1A CN202011279152A CN112395059A CN 112395059 A CN112395059 A CN 112395059A CN 202011279152 A CN202011279152 A CN 202011279152A CN 112395059 A CN112395059 A CN 112395059A
- Authority
- CN
- China
- Prior art keywords
- firefly
- fitness value
- population
- fitness
- beta
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 241000254158 Lampyridae Species 0.000 title claims abstract description 130
- 238000000034 method Methods 0.000 title claims abstract description 30
- 239000002245 particle Substances 0.000 claims abstract description 15
- 238000010521 absorption reaction Methods 0.000 claims abstract description 5
- 238000005457 optimization Methods 0.000 abstract description 28
- 230000007547 defect Effects 0.000 description 3
- 238000011423 initialization method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F9/00—Arrangements for program control, e.g. control units
- G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
- G06F9/46—Multiprogramming arrangements
- G06F9/48—Program initiating; Program switching, e.g. by interrupt
- G06F9/4806—Task transfer initiation or dispatching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Software Systems (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a CMP task scheduling method for improving a firefly algorithm, which comprises the following steps of 1: the expected input number N of the firefly population, the absorption coefficient of the medium to light is gamma, the initial step factor alpha and the maximum attraction degree beta0Threshold value of attraction betaM(ii) a Step 2: initializing the quantity and the position of the firefly population according to an initialization strategy; and step 3: calculating the fitness value of the firefly according to the position of the firefly; and 4, step 4: each firefly flies to a firefly with higher brightness than the firefly, the fitness value of the firefly after the firefly reaches the new position is calculated, if the fitness value is better than the fitness value of the firefly, the firefly reaches the new position, and if the fitness value is not better than the fitness value of the firefly, the firefly stays at the original position; and 5: and (4) judging the optimizing result, if the optimizing result meets the termination condition, ending the iteration process, otherwise, repeating the operation in the step (3) and iterating the firefly particles again. The invention accelerates the convergence rate of the firefly particles, greatly reduces the possibility of trapping in the local optimal solution in the optimization process, and reduces unnecessary overlappingThe generation times shorten the completion time of task scheduling.
Description
Technical Field
The invention relates to a CMP task scheduling method, in particular to a CMP task scheduling method for improving a firefly algorithm, and belongs to the technical field of task scheduling.
Background
With the increasing development of the technology level of computer software, the requirements of modern applications on the performance of computer hardware are continuously increased. Under the limit of relatively slow development of semiconductor technology, a Multi-core Processor (Chip Multi-core Processor) is produced by simply increasing the dominant frequency of a single core to be insufficient for maintaining moore's law. Under the support of a multi-core processor, one chip collects a plurality of processor cores, and the requirements of improving the system performance and balancing the load are met at the minimum cost. Of course, a good task scheduling algorithm is necessary to make the advantages of the multi-core processor be realized and embodied.
In order to obtain an optimal task scheduling strategy, a related scholars use a group intelligent algorithm in the solution of a task scheduling sequence and prove the practical possibility of the theoretical system in practical application. As a relatively novel group intelligence algorithm, the idea of the firefly algorithm is derived from simulating the information exchange behavior between fireflies. Each individual (firefly particle) in the population is a candidate solution to the corresponding problem. The search of the firefly algorithm relies on the attraction between individuals to produce movement, and the firefly with poor fitness (darker) moves towards the firefly with better fitness (brighter).
Although the firefly algorithm can show better performance in the aspect of optimization problem, the firefly algorithm still has some defects, such as slow convergence rate and easy falling into local optimization in the aspect of complex problem.
Disclosure of Invention
The invention aims to provide a CMP task scheduling method for improving a firefly algorithm in order to solve the problems of low convergence rate and easy falling into local optimization on complex problems.
The purpose of the invention is realized as follows:
a CMP task scheduling method for improving a firefly algorithm comprises the following steps:
step 1: defining the meaning of the parameters, and initializing the basic parameters as follows:
the expected input number N (N is a positive integer) of the firefly population, the absorption coefficient of the medium to light is gamma, the initial step factor alpha and the maximum attraction beta0Threshold value of attraction betaM;
Step 2: initializing the quantity and the position of the firefly population according to an initialization strategy;
and step 3: calculating the fitness value of the firefly according to the position of the firefly, wherein the higher the fitness value is, the higher the brightness is;
and 4, step 4: each firefly flies to a firefly with higher brightness than the firefly, the fitness value of the firefly after the firefly reaches the new position is calculated, if the fitness value is better than the fitness value of the firefly, the firefly reaches the new position, and if the fitness value is not better than the fitness value of the firefly, the firefly stays at the original position;
and 5: and (4) judging the optimizing result, if the optimizing result meets the termination condition, ending the iteration process, otherwise, repeating the operation in the step (3) and iterating the firefly particles again.
The invention also includes such features:
the step 2 initialization strategy is as follows:
(2) adding new firefly particles in a random generation mode;
(3) if the relative attraction between the inputted new firefly and the firefly closest to the inputted new firefly is larger than betaMIf the fitness is less than beta, one bit with lower fitness is eliminatedMAdding the new strain into the initial population;
(4) repeating the step (3) until the initial number of fireflies of each group is reachedThe initial number of the population reaches N.
The step 5 termination conditions are as follows:
and judging the optimizing result, setting a threshold value max _ step, recording the number f _ step of the optimizing result which is kept unchanged, and terminating the optimizing process if f _ step is greater than max _ step.
Compared with the prior art, the invention has the beneficial effects that:
the invention improves the initialization method of the firefly algorithm, so that the initial position distribution of each firefly is more uniform, thereby accelerating the convergence speed of firefly particles and greatly reducing the possibility of falling into the local optimal solution in the optimization process; by improving the optimization termination strategy, unnecessary iteration times are reduced, and the completion time of task scheduling is shortened.
Drawings
Fig. 1 is a flow chart of CMP task scheduling based on the modified firefly algorithm.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.
The invention aims to provide an improved CMP task scheduling method of a firefly algorithm.
The firefly algorithm can generally show better performance in the aspect of optimization problem, but still has some non-negligible defects. For example, the standard firefly algorithm employs a randomly generated initialization method, which may cause the initial position of the firefly population to be unevenly distributed, so that the convergence rate of the firefly population becomes slow, and the firefly population is likely to fall into a local optimal solution on the complex optimization problem. In addition, the standard firefly algorithm employs an optimized termination strategy that terminates when the maximum number of iterations is reached. In practical application, if the maximum iteration number is set too large, unnecessary iteration time and calculation cost are increased, otherwise, a large deviation between the output result and the actual optimal value may be caused.
Aiming at the problems of the firefly algorithm, the invention improves the initialization mode and the optimization termination strategy of the standard firefly algorithm.
Firstly, in order to reduce the possibility of falling into a local optimal solution in the optimization process, the solution space is grouped during initialization, and the firefly particles are screened through the attraction degree among the firefly particles, so that the effect of uniform initialization distribution of the firefly population is achieved.
Secondly, the invention introduces a threshold value max _ step and the number f _ step of the optimization result which is kept unchanged in the optimization process, and if f _ step is more than max _ step, the optimization process is terminated.
The improved firefly algorithm comprises the following steps:
step 1: defining the meaning of the parameters, and initializing the basic parameters as follows:
the expected input number N (N is a positive integer) of the firefly population, the absorption coefficient of the medium to light is gamma, the initial step factor alpha and the maximum attraction beta0Threshold value of attraction betaM;
Step 2: initializing the quantity and the position of the firefly population according to an improved initialization strategy;
and step 3: calculating the fitness value of the firefly according to the position of the firefly, wherein the higher the fitness value is, the higher the brightness is;
and 4, step 4: each firefly flies to a firefly with higher brightness than the firefly, the fitness value of the firefly after the firefly reaches the new position is calculated, if the fitness value is better than the fitness value of the firefly at the original position, the firefly reaches the new position, and if the fitness value is not better than the fitness value of the firefly at the original position, the firefly stays at the;
and 5: and (4) judging the optimization result according to the improved termination condition, if the optimization result meets the condition, ending the iteration process, otherwise, repeating the operation in the step (3) and iterating the firefly particles again.
The invention improves the initialization mode of the firefly algorithm on the original basis, and screens N high-quality firefly samples by solving the attraction degree among the firefly samples, so that the initial position distribution of each firefly is more uniform, thereby accelerating the convergence speed of firefly particles and greatly reducing the possibility of trapping in a local optimal solution in the optimization process. And in the later stage of the optimization process, after N fireflies are iterated for multiple times, the optimization condition is judged, and when the conditions are met, the optimization process is ended, so that the optimal solution target is obtained, and the time and space cost caused by unnecessary iteration times is reduced.
The detailed steps of the invention are as follows:
step 1: suppose Xi=(xi1,xi2,...,xiD) Is the ith firefly in the population, wherein i is 1, 2, N and D respectively represent the population size and problem dimension (N and D are positive integers), the absorption coefficient of the medium to light is gamma, and the initial value isInitial step factor α, maximum attraction β0Threshold value of attraction betaM;
Step 2: initializing the quantity and the position of the firefly population according to an improved initialization strategy, wherein the initialization steps are as follows:
(2) adding new firefly particles in a random generation mode;
(3) if the relative attraction between the inputted new firefly and the firefly closest to the inputted new firefly is larger than betaMIf the fitness is less than beta, one bit with lower fitness is eliminatedMAdding the initial population, wherein the formula of the relative attraction degree is as follows:
wherein r isijIs firefly XiTo firefly XjThe euclidean distance between.
(4) Repeating the step (3) until the initial number of fireflies of each group is reachedThe initial number of the population reaches N.
And step 3: calculating the fitness value of the firefly according to the position of the firefly, wherein the higher the fitness value is, the higher the brightness is;
and 4, step 4: every firefly flies to the firefly that is higher than own luminance, calculates its fitness value after arriving new position, if be superior to the original position, then reaches new position, otherwise stops in the original position, and the update mode of position is according to following formula:
xid(t+1)=xid(t)+β(rij)·(xjd(t)-xid(t))+αε
and 5: and judging the optimization result, setting a threshold value max _ step, recording the number f _ step of the optimization result which is kept unchanged, and terminating the optimization process if f _ step is greater than max _ step, otherwise, performing the next iteration.
The application example of the present invention is to select the optimal scheduling policy for each core through a modified firefly algorithm for the application of CMP architecture, but the scope of the present invention is not limited thereto, and any person skilled in the art should be covered by the scope of the present invention without changing or replacing the technical solution of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
In summary, the following steps: the invention provides a CMP task scheduling method for improving a firefly algorithm. Although the basic firefly algorithm can show better performance in the aspect of optimization problem, the basic firefly algorithm still has some defects, such as slow convergence speed, easy falling into local optimization in the aspect of complex problem and the like. Aiming at the problems, the invention improves the initialization of the firefly algorithm, and screens out N high-quality firefly particles according to the attraction degree among the firefly particles, so that the initial position distribution of each firefly is more uniform, thereby accelerating the convergence speed of the firefly population and greatly reducing the possibility of trapping in a local optimal solution in the optimization process. And in the later stage of the optimization process, after the N fireflies are iterated for multiple times, the optimization condition is judged, and when the conditions are met, the optimization process is ended, so that the optimal solution target is obtained, and meanwhile, the unnecessary iteration times are reduced.
Claims (3)
1. A CMP task scheduling method for improving a firefly algorithm is characterized by comprising the following steps:
step 1: defining the meaning of the parameters, and initializing the basic parameters as follows:
the expected input number N (N is a positive integer) of the firefly population, the absorption coefficient of the medium to light is gamma, the initial step factor alpha and the maximum attraction beta0Threshold value of attraction betaM;
Step 2: initializing the quantity and the position of the firefly population according to an initialization strategy;
and step 3: calculating the fitness value of the firefly according to the position of the firefly, wherein the higher the fitness value is, the higher the brightness is;
and 4, step 4: each firefly flies to a firefly with higher brightness than the firefly, the fitness value of the firefly after the firefly reaches the new position is calculated, if the fitness value is better than the fitness value of the firefly, the firefly reaches the new position, and if the fitness value is not better than the fitness value of the firefly, the firefly stays at the original position;
and 5: and (4) judging the optimizing result, if the optimizing result meets the termination condition, ending the iteration process, otherwise, repeating the operation in the step (3) and iterating the firefly particles again.
2. The method for scheduling CMP task of firefly algorithm improvement according to claim 1, wherein the step 2 initialization strategy is as follows:
(2) adding new firefly particles in a random generation mode;
(3) if the relative attraction between the inputted new firefly and the firefly closest to the inputted new firefly is larger than betaMIf the fitness is less than beta, one bit with lower fitness is eliminatedMAdding the new strain into the initial population;
3. The method for scheduling CMP tasks of firefly algorithm as claimed in claim 1, wherein the step 5 termination condition is as follows:
and judging the optimizing result, setting a threshold value max _ step, recording the number f _ step of the optimizing result which is kept unchanged, and terminating the optimizing process if f _ step is greater than max _ step.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011279152.1A CN112395059A (en) | 2020-11-16 | 2020-11-16 | CMP task scheduling method for improving firefly algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011279152.1A CN112395059A (en) | 2020-11-16 | 2020-11-16 | CMP task scheduling method for improving firefly algorithm |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112395059A true CN112395059A (en) | 2021-02-23 |
Family
ID=74600414
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011279152.1A Pending CN112395059A (en) | 2020-11-16 | 2020-11-16 | CMP task scheduling method for improving firefly algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112395059A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112947450A (en) * | 2021-02-20 | 2021-06-11 | 哈尔滨工程大学 | Multi-module ship cooperative thrust distribution method |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2018036282A1 (en) * | 2016-08-24 | 2018-03-01 | 深圳市中兴微电子技术有限公司 | Task scheduling method, device and computer storage medium |
CN110533151A (en) * | 2019-07-29 | 2019-12-03 | 湘潭大学 | A kind of firefly optimization algorithm based on the law of universal gravitation |
AU2020101065A4 (en) * | 2020-06-19 | 2020-07-23 | Hubei University Of Technology | Method for scheduling UAVs based on chaotic adaptive firefly algorithm |
-
2020
- 2020-11-16 CN CN202011279152.1A patent/CN112395059A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2018036282A1 (en) * | 2016-08-24 | 2018-03-01 | 深圳市中兴微电子技术有限公司 | Task scheduling method, device and computer storage medium |
CN110533151A (en) * | 2019-07-29 | 2019-12-03 | 湘潭大学 | A kind of firefly optimization algorithm based on the law of universal gravitation |
AU2020101065A4 (en) * | 2020-06-19 | 2020-07-23 | Hubei University Of Technology | Method for scheduling UAVs based on chaotic adaptive firefly algorithm |
Non-Patent Citations (2)
Title |
---|
李成辉;李仁旺;杨强光;贾江鸣;: "基于改进萤火虫算法的云计算任务调度算法", 浙江理工大学学报(自然科学版), no. 03 * |
陈亚峰;张晓明;曹国清;周泽?戴波;: "双种群协同下带混沌闪烁机制的萤火虫算法研究", 西安交通大学学报, no. 03 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112947450A (en) * | 2021-02-20 | 2021-06-11 | 哈尔滨工程大学 | Multi-module ship cooperative thrust distribution method |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110503192B (en) | Resource efficient neural architecture | |
CN109242878B (en) | Image multi-threshold segmentation method based on self-adaptive cuckoo optimization method | |
CN113242568A (en) | Task unloading and resource allocation method in uncertain network environment | |
CN110780938B (en) | Computing task unloading method based on differential evolution in mobile cloud environment | |
CN110233755B (en) | Computing resource and frequency spectrum resource allocation method for fog computing in Internet of things | |
CN112988345A (en) | Dependency task unloading method and device based on mobile edge calculation | |
CN112463189B (en) | Distributed deep learning multi-step delay updating method based on communication operation sparsification | |
CN110689345B (en) | Unlicensed blockchain consensus method and system for adjusting block weights, and P2P network | |
US20220374722A1 (en) | Intelligent ammunition co-evolution task assignment method | |
CN113255138B (en) | Load distribution optimization method for power system | |
CN113485826A (en) | Load balancing method and system for edge server | |
CN112990420A (en) | Pruning method for convolutional neural network model | |
CN110689113A (en) | Deep neural network compression method based on brain consensus initiative | |
CN114625506A (en) | Edge cloud collaborative task unloading method based on adaptive covariance matrix evolution strategy | |
CN112395059A (en) | CMP task scheduling method for improving firefly algorithm | |
CN116107754A (en) | Memory management method and system for deep neural network | |
CN114186671A (en) | Large-batch decentralized distributed image classifier training method and system | |
CN117081895B (en) | Automatic modulation identification method based on self-adaptive noise reduction | |
CN111930484B (en) | Power grid information communication server thread pool performance optimization method and system | |
CN116824232A (en) | Data filling type deep neural network image classification model countermeasure training method | |
CN116582502A (en) | TD3 algorithm-based Coflow scheduling system | |
Li et al. | Adafl: Adaptive client selection and dynamic contribution evaluation for efficient federated learning | |
CN115034634A (en) | Phased array radar resource scheduling management method based on greedy algorithm | |
CN112838905B (en) | Interference suppression method, device and equipment | |
CN114138416A (en) | DQN cloud software resource self-adaptive distribution method facing load-time window |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |