CN108595366A - A kind of multicomputer system structural model fault-tolerant in real time - Google Patents
A kind of multicomputer system structural model fault-tolerant in real time Download PDFInfo
- Publication number
- CN108595366A CN108595366A CN201810338577.1A CN201810338577A CN108595366A CN 108595366 A CN108595366 A CN 108595366A CN 201810338577 A CN201810338577 A CN 201810338577A CN 108595366 A CN108595366 A CN 108595366A
- Authority
- CN
- China
- Prior art keywords
- task
- subtask
- time
- processor
- real
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F15/00—Digital computers in general; Data processing equipment in general
- G06F15/16—Combinations of two or more digital computers each having at least an arithmetic unit, a program unit and a register, e.g. for a simultaneous processing of several programs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/07—Responding to the occurrence of a fault, e.g. fault tolerance
- G06F11/14—Error detection or correction of the data by redundancy in operation
- G06F11/1479—Generic software techniques for error detection or fault masking
- G06F11/1489—Generic software techniques for error detection or fault masking through recovery blocks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Computer Hardware Design (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Software Systems (AREA)
- Quality & Reliability (AREA)
- Hardware Redundancy (AREA)
Abstract
Herein in conjunction with the demand of the multiprocessor system architecture of real-time fault tolerance, to ensure the needs that all tasks are correctly completed before deadline according to real-time system, and it is completed using software by heuritic approach for traditional Real-Time Scheduling, its performance with it is less efficient, for the disadvantage of the fault-tolerant ability difference of multiprocessor real-time system.Real-time task is decomposed into linear incoherent subtask, it is proposed a kind of multicomputer system structural model fault-tolerant in real time, real-time system is optimized, the scheduling hardware model of centre type is used, and uses task redundancy scheme to realize fault-tolerant ability.The present invention distributes three Optimal Parameters of deviation ratio and earliest finish time in the task reject rate of processor, multiprocessor run time and is substantially better than conventional method, and fitness is high, meets various needs.
Description
Technical field
The present invention is applied to the structure design of real-time fault-tolerant multicomputer system.
Background technology
With the rapid decline of the increasingly mature and hardware cost of computer technology, various functions and it is complicated in real time
It computer system and is widely used, real-time computer system refers to that can make a response to incoming event within the time of restriction
System.Therefore, if a computer system needs to export number from external environment input data within the exact time and to it
According to, or executing some other processing, then this system is exactly real-time computer system.In order to ensure the reliable of computer system
Property, there has been proposed a series of fault-tolerance approach, it is fault-tolerant refer to computer during operation, when software or hardware break down,
System can automatically switch resource, adopt remedial measures, to ensure that system continues correct execution task.In some special applications fields
It closes, such as aerospace, national defense and military, nuclear energy electric power, industrial chemical, hygiene medical treatment first aid key sector, a department of computer science
System mistake this may result in irremediable massive losses, must be used in the design of these critical systems a large amount of
Fault-toleranr technique ensures that the computation error that happens suddenly will not lead to the failure of whole system in operation.
In recent years, the urgent growth with people for calculating demand, multicomputer system are very universal.Due to
Multiprocessor has very strong hardware computation ability, therefore is commonly utilized in various types of computers, embedded system and each
In kind mobile device.However, traditional Real-Time Scheduling by heuritic approach using software complete, performance with it is less efficient,
It is poor for the fault-tolerant ability of multiprocessor real-time system.It is therefore proposed that a kind of multicomputer system structure mould fault-tolerant in real time
Type optimizes real-time system, and realizes fault-tolerant ability using task redundancy scheme, reasonably can using these technologies
It is preferably promoted so that real-time fault tolerance system obtains.
Invention content
Technical problems based on background technology, the present invention propose a kind of fault-tolerant multicomputer system structure in real time
Model method.
A kind of fault-tolerant multicomputer system structural model in real time proposed by the present invention, at dedicated task
Manage device, central scheduler, parallel processing cores.
Preferably, dedicated task processor is used in real-time multicomputer system structure, obtains real-time task module
The address set of the task memory address of middle generation and the partial parameters of task；According to the operation description of task and task structure
Coupling and interaction, carry out the decomposition and processing of task.Resulting bottle set of tasks continue to be transferred to task processing module start it is excellent
Change series processing.
Preferably, central scheduler is used for the scheduling and control of dedicated task processor resulting bottle task queue.
Preferably.Operation and processing of the parallel processing cores for subtask and its Redundant task, are additionally operable to parallel task
Task, data acquisition system and its yuan are realized in mutual exchange data, as a result merger and collision detection between processor and program
Element is at different levels comprehensive parallel.
In the present invention, system bus connects dedicated task processor, central scheduler and parallel core.Real-time task is passed through
Dedicated task processor handles the subtask sequence optimized, and the scheduling and control of task, son are carried out by central scheduler
Task carries out operation and processing by parallel processing cores, final to obtain correctly solution.
Description of the drawings
Fig. 1 is a kind of fault-tolerant multicomputer system structural model schematic diagram in real time proposed by the present invention.
Specific implementation mode
The present invention is made further to explain with reference to specific embodiment.
Embodiment
With reference to figure 1, the present embodiment proposes a kind of fault-tolerant multicomputer system structural model method in real time.Pass through setting
Dedicated task processor carries out task and handles and central scheduler monitors subtask state, executes backup tasks and ensures
The mode of task normal operation, to realize fault-tolerance.The task reject rate of traditional real-time object model, multiprocessor fortune simultaneously
The row time distributes deviation ratio and the undesirable situation of parameter on earliest finish time is optimized, using optimization subtask team
The method of row is allocated processor according to allocation algorithm according to processing task, processing time and priority is estimated.Optimization
Task is sent to central scheduler by task queue, by task scheduling in the way of meeting real-time and fault-tolerance, and is transmitted
To parallel processing cores, the task sequence during which caches according to current each task queue carries out timely task
Response, is transmitted to after being disposed and completes queue.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
Any one skilled in the art in the technical scope disclosed by the present invention, according to the technique and scheme of the present invention and its
Inventive concept is subject to equivalent substitution or change, should be covered by the protection scope of the present invention.
Claims (3)
1. a kind of multicomputer system structural model fault-tolerant in real time, it is characterised in that：It can ensure under multicomputer system
The fault-tolerance of task scheduling processing meets the real-time of task, and improves the performance of multi-task scheduling, to improve system effect
Rate.The system structure includes dedicated task processor, central scheduler, parallel processing cores.
The dedicated task processor is used in real-time multicomputer system structure, for obtaining real-time task module
The address set of the task memory address of middle generation and the partial parameters of task；Task-decomposing module is retouched according to the operation of task
State with the coupling of task structure and interaction, carry out the decomposition and processing of task.Resulting bottle set of tasks continues to be transferred at task
Reason module starts optimization processing.The optimization processing procedure is：The son of algorithm in operation task processing module firmware
Task triplet sets preserve subtask list when iteration reaches setting number or subtask triad sequence is stablized.
The Task-decomposing module analyzes a real-time task, is decomposed into linear incoherent subtask, ensures real
When task parallel processing result have correctness.
Task-decomposing module is that real-time task carries out parallelization, can be regarded as to multicomputer system structure task fault-tolerant in real time
The pretreatment of the task of scheduling model is the bridge that task and multiple processor cores are contacted in system structure, dedicated task processing
Dedicated module is arranged in device.In multiprocessor computer structure, task is difficult to make full use of the efficient place of multiple processors
Reason ability allows the efficiency of multicomputer system and benefit to have a greatly reduced quality, and task, which enters, carries out parallelization in Task-decomposing module
Processing, serial task are broken down into multiple linear incoherent subtasks, in order to the next scheduling to subtask and
Parallel processing.Task-decomposing module carries out parallel optimization to task allows its multiple operation to be carried out at the same time, to improve system
Performance and complicated real-time task is allowed to be possibly realized in the corresponding time.
The task processing module is handled Task-decomposing module resulting bottle set of tasks in dedicated task processor, son
Task is calculated based on relevant algorithm and obtains best or preferable feasible solution, and the standard deviation and task of multiprocessor are prevented
Reject rate it is excessively high, reduce processor utilization and system effectiveness.
The task handles firmware Firmware, refers to and pre-installs optimization processing routine in read-only memory, supports son
Task tuple-set sequence is selected excellent.
In handling Task-decomposing module resulting bottle task list, dedicated task processing system appoints the setting of each subtask
Be engaged in id, and each subtask processing task and processing time are estimated record, right according to the description that subtask set element is held
Task-set call relevant algorithm ask the processor scheduling strategy of wherein each subtask and subtask backup tasks optimal or compared with
Excellent solution, described method includes following steps：
Step 1：Unique id is distributed in the whole subtask list elements of scanning gained, each subtask, and all subtask elements are safeguarded
One triple<Di,Zi,Ri>, indicate that subtask id, subtask run processor number, the operation of subtask backup tasks respectively
Processor is numbered.
Step 2：Generate the sequence of the subtask of N number of whole subtasks triple composition, subtask triple Z in each sequenceiWith
RiIt is provided at random by system.
Step 3：System handles task according to each subtask and processing time assesses every sequence, and every sequence obtains
Adaptation parameter.
Step 4：It is screened according to every sequence adaptation parameter, the sequence that adaptation parameter reaches threshold value is selected, other sequences are abandoned
Row.
Step 5：Obtained sequence combines two-by-two, and each two sequence generates random number R andomNum, and sequence is along RandomNum points
At two sections of left and right, intersection is combined into new sequence, calculates new formation sequence adaptation parameter.
Step 6：Return to step 4 no longer changes until recycling given number or sequence, selects the maximum sequence of adaptation parameter
Row output.
2. the central scheduler is used for the scheduling and control of dedicated task processor resulting bottle task queue, inside there is main task
Scheduler module, Redundant task scheduler module, task controller, dispatching algorithm firmware and error detection hardware.Central scheduler pair
Subtask queue scheduling and it is fault-tolerant controlled, subtask error, then backup tasks are executed, to ensure task normal operation.
The task scheduling modules complete subtask scheduling and the scheduling of subtask backup tasks.
The task controller, the task distribution, task erasing and task that memory queue is handled for each processor tasks are refused
Absolutely.
The error detection hardware generates mistake to processor processing task and is detected and transmits.
The dispatching algorithm firmware, refers to and pre-installs dispatching algorithm program in read-only memory, supports subtask tuple sequence
Scheduling of multiprocessor.
The task tuple that central scheduler is is scheduled, and its step are as follows：
Step 1：A task is taken from subtask triad sequence obtained by dedicated task processor, it, will according to its alignment processing device
Subtask is put into processor Z as early as possibleiQueue caching in, Redundant task then only under the premise of meeting real-time, is being located
Manage device RiQueue caching is just scheduled when empty.
Step 2：If the subtask and its Redundant task can not normal consistency execute, which is placed into mission failure team
Row.
Step 3：If detection show that the subtask or its Redundant task exceed the stipulated time, which is put into mission failure team
Row.
Step 4：If subtask has executed, Redundant task release is corresponded to.
Step 5：If system detectio subtask is corresponded to Redundant task priority and is shifted to an earlier date, execute the redundancy as a result, it has been found that mistake
Task.
Step 6：If subtask successful execution or Redundant task are successfully scheduled and execute, existing time-scale is updated.
3. the parallel processing cores are additionally operable to parallel task each other for the operation and processing of subtask and its Redundant task
Exchange data, as a result merger and collision detection realize that task, data acquisition system and its element are each between processor and program
Grade it is comprehensive parallel.Multiprocessor respectively handles task sequence element by being obtained in processor respectively queue caching.
The shared instruction data high-speed caching improves processor speed for storing the data and instruction that processor is commonly used
Degree.
The processor queue caching, stores the subtask waited in line in processor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810338577.1A CN108595366A (en) | 2018-04-16 | 2018-04-16 | A kind of multicomputer system structural model fault-tolerant in real time |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810338577.1A CN108595366A (en) | 2018-04-16 | 2018-04-16 | A kind of multicomputer system structural model fault-tolerant in real time |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108595366A true CN108595366A (en) | 2018-09-28 |
Family
ID=63622675
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810338577.1A Pending CN108595366A (en) | 2018-04-16 | 2018-04-16 | A kind of multicomputer system structural model fault-tolerant in real time |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108595366A (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101069161A (en) * | 2004-12-01 | 2007-11-07 | 索尼计算机娱乐公司 | Scheduling method, scheduling device, and multiprocessor system |
| US20090077561A1 (en) * | 2007-07-05 | 2009-03-19 | International Business Machines Corporation | Pipeline Processing Method and Apparatus in a Multi-processor Environment |
| CN102999385A (en) * | 2012-11-06 | 2013-03-27 | 苏州懿源宏达知识产权代理有限公司 | Multiprocessor co-processing method in computation equipment |
-
2018
- 2018-04-16 CN CN201810338577.1A patent/CN108595366A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101069161A (en) * | 2004-12-01 | 2007-11-07 | 索尼计算机娱乐公司 | Scheduling method, scheduling device, and multiprocessor system |
| US20090077561A1 (en) * | 2007-07-05 | 2009-03-19 | International Business Machines Corporation | Pipeline Processing Method and Apparatus in a Multi-processor Environment |
| CN102999385A (en) * | 2012-11-06 | 2013-03-27 | 苏州懿源宏达知识产权代理有限公司 | Multiprocessor co-processing method in computation equipment |
Non-Patent Citations (3)
| Title |
|---|
| 化二为一: "容错关键技术", 《HTTPS://WENKU.BAIDU.COM/VIEW/1FB65654BE1E650E52EA99CD.HTML》 * |
| 李莉丽: "实时分布系统容错调度算法的设计与实现", 《中国优秀博硕士学位论文全文数据库 (硕士)信息科技辑》 * |
| 袁云: "基于多核处理器并行系统的任务调度算法研究", 《中国优秀硕士论文全文数据库-信息科技辑》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Gu et al. | Liquid: Intelligent resource estimation and network-efficient scheduling for deep learning jobs on distributed GPU clusters | |
| US8893145B2 (en) | Method to reduce queue synchronization of multiple work items in a system with high memory latency between processing nodes | |
| CN101958808B (en) | Cluster task dispatching manager used for multi-grid access | |
| CN108710536B (en) | Multilevel fine-grained virtualized GPU (graphics processing Unit) scheduling optimization method | |
| CN101986272A (en) | Task scheduling method under cloud computing environment | |
| Gotoda et al. | Task scheduling algorithm for multicore processor system for minimizing recovery time in case of single node fault | |
| CN110795254A (en) | Method for processing high-concurrency IO based on PHP | |
| JP2010079622A (en) | Multi-core processor system and task control method thereof | |
| CN103500119B (en) | A kind of method for allocating tasks based on pre-scheduling | |
| US20060015876A1 (en) | Light weight context switching technique | |
| Faragardi et al. | Optimal task allocation for maximizing reliability in distributed real-time systems | |
| Socci et al. | Time-triggered mixed-critical scheduler on single and multi-processor platforms | |
| CN108595366A (en) | A kind of multicomputer system structural model fault-tolerant in real time | |
| US8291419B2 (en) | Fault tolerant system for execution of parallel jobs | |
| Li et al. | Redundant and fault-tolerant algorithms for real-time measurement and control systems for weapon equipment | |
| Sun et al. | Hybrid overloading and stochastic analysis for redundant real-time multiprocessor systems | |
| Dong et al. | A general analysis framework for soft real-time tasks | |
| CN115599507A (en) | Data processing method, execution workstation, electronic device and storage medium | |
| Chuprat et al. | Evaluation of a linear programming approach towards scheduling divisible real-time loads | |
| Han et al. | A resilient hierarchical distributed loop self-scheduling scheme for cloud systems | |
| Sabu et al. | SMILEY: a mixed-criticality real-time task scheduler for multicore systems | |
| CN117389625B (en) | Process synchronization method, system, equipment and medium based on active interrupt instruction | |
| CN105045659B (en) | Task based access control is overlapping with the fault-tolerant method for scheduling task of virtual machine (vm) migration in a kind of cloud | |
| Guo et al. | QoS-aware fault-tolerant rate-monotonic first-fit scheduling in real-time systems | |
| CN114866612B (en) | Electric power micro-service unloading method and device |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180928 |
|
| WD01 | Invention patent application deemed withdrawn after publication |