CN102508704A - Method for implementing task decomposition and parallel processing in computer software system - Google Patents
Method for implementing task decomposition and parallel processing in computer software system Download PDFInfo
- Publication number
- CN102508704A CN102508704A CN2011103545730A CN201110354573A CN102508704A CN 102508704 A CN102508704 A CN 102508704A CN 2011103545730 A CN2011103545730 A CN 2011103545730A CN 201110354573 A CN201110354573 A CN 201110354573A CN 102508704 A CN102508704 A CN 102508704A
- Authority
- CN
- China
- Prior art keywords
- subtask
- task
- computer software
- scheduling unit
- parallel
- 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
- 238000012545 processing Methods 0.000 title claims abstract description 63
- 238000000034 method Methods 0.000 title claims abstract description 60
- 238000000354 decomposition reaction Methods 0.000 title claims abstract description 20
- 239000011159 matrix material Substances 0.000 claims abstract description 41
- 238000000547 structure data Methods 0.000 claims abstract description 20
- 230000001419 dependent effect Effects 0.000 abstract 1
- 230000006870 function Effects 0.000 description 41
- 230000008569 process Effects 0.000 description 17
- 238000010586 diagram Methods 0.000 description 12
- 230000008859 change Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
Images
Landscapes
- Multi Processors (AREA)
Abstract
The invention relates to a method for implementing task decomposition and parallel processing in a computer software system. The computer software system is provided with a task scheduling unit connected between a task receiving end and a processing unit. The method includes that the task scheduling unit decomposes a task into subtasks, stores dependent relations among the subtasks as matrix structure data, and confirms sequence of the subtasks which can be processed parallelly according to the matrix structure data, and a processor is used for parallel processing. Using the method can respond to various changes in a configuration manner, corresponding adjustments can be performed dynamically to automatically implement task decomposition, manual interference is not needed, parallel task decomposition efficiency is improved, and parallel task processing capacity of a computer is improved greatly. In addition, the method for implementing task decomposition and parallel processing in the computer software system is simple and convenient in application way, low in implementation cost and wider in application range.
Description
Technical field
The present invention relates to the computer software technical field, particularly task processing method technical field in the computer software, the realization task in a kind of computer software of specifically being meant are decomposed and the method for parallel processing control.
Background technology
In the computer system that only has a monokaryon CPU, all tasks are all handled with serial mode, and its processing speed is slower, and efficient is lower.Along with, having double-core or multi-core CPU or have being widely used of computer system of a plurality of CPU, the task processing mode of parallel processing also is applied in the computer software more and more.
Parallel processing is meant that computing machine is in synchronization or two or more character of completion work identical or inequality in the same time interval.The most significant advantage of parallel processing is to have improved arithmetic speed.N bit-serial arithmetic and the computing of n bit parallel are come comparison, and under the identical situation of element processing speed, latter's arithmetic speed almost rises to the former n times.This is the method for parallel processing that a kind of resource repeats, and it is that principle according to " winning victory with quantity " increases substantially arithmetic speed.But parallel processing does not also terminate in the simple repetition of equipment, and it also has more implication, like time-interleaving and resource sharing etc.So-called time-interleaving is according to the pipeline processes technology, and a plurality of processing procedures are staggered in time each other, uses the several sections of same set of equipment in turn.Resource sharing then is according to the principle of " timesharing is shared ", makes a plurality of users use same set of equipment in chronological order.
In parallel task was handled, very strict to preceding making demands, task resolution was the problem that at first need solve.Generally speaking, use parallel computation not only can not promote performance in a large number, can run counter to desire on the contrary.Because inappropriate task resolution model and inappropriate scheduling of resource all can cause performance of entire system on the contrary not as unit one process single-threaded system.In task resolution, problem (Problem), scene (context), condition (Forces), scheme (Solutions) all will be considered carefully.
One of parallel task resolution model is task resolution model (also having data decomposition, data stream to decompose).At present; When the task of employing resolution model; Under the situation that the priority and the processing speed of subtask changes, need carry out manual intervention and come the adjustment program to adapt to this variation, and computer software can't be configured the generation of dealing with this variation automatically; Thereby the application to the parallel task resolution model has caused limitation, and the parallel task processing power of computing machine is reduced.
Summary of the invention
The objective of the invention is to have overcome above-mentioned shortcoming of the prior art, provide a kind of when adopting the task resolution model, various variations that can configuration reply subtask; Dynamically adjust accordingly, thereby need not the manual work intervention, improved the parallel task decomposing efficiency; Significantly promoted the parallel task processing power of computing machine; And application mode is easy, realize with low cost, range of application comparatively widely in the computer software realization task decompose and the method for parallel processing control.
In order to realize above-mentioned purpose; In computer software of the present invention the realization task decompose and the method for parallel processing control in; Have the task scheduling unit that is connected between task receiving end and the processing unit in the described computer software, described method may further comprise the steps:
(1) described task scheduling unit receives task from described task receiving end;
(2) described task scheduling unit is split as a plurality of subtasks that can independently carry out with a task;
(3) dependence between described each subtask is confirmed in described task scheduling unit;
(4) described task scheduling unit is stored as the matrix structure data with the dependence between described each subtask;
(5) but the subtask sequence of executed in parallel is confirmed in described task scheduling unit according to described matrix structure data;
(6) described task scheduling unit is sent to described processing unit for parallel execution according to described subtask sequence with described task.
In this computer software the realization task decompose and the method for parallel processing control in; Described task scheduling unit is split as a plurality of subtasks that can independently carry out with a task, is specially: described task scheduling unit is split as a plurality of subtasks with independent function with a task according to the function of its required operation.
In this computer software the realization task decompose and the method for parallel processing control in; Dependence between described each subtask; Be specially: is condition if having first subtask of an independent function with another result with second subtask of an independent function, then exists first subtask to depend on the relation of second subtask.
In this computer software the realization task decompose and the method for parallel processing control in; The transverse axis and the longitudinal axis are tactic subtask in described matrix structure data, and the transverse axis subtask that a certain transverse axis coordinate is corresponding with this position of data representation of ordinate of orthogonal axes crossover location in the described matrix structure data is for the dependence of corresponding longitudinal axis subtask, this position.
In this computer software the realization task decompose and the method for parallel processing control in; The data of described transverse axis coordinate and ordinate of orthogonal axes crossover location are represented with binary data; " 1 " representes that there is dependence corresponding transverse axis subtask, this position for corresponding longitudinal axis subtask, this position, and " 0 " representes that corresponding transverse axis subtask, this position does not have dependence for corresponding longitudinal axis subtask, this position.
In this computer software the realization task decompose and the method for parallel processing control in; The subtask for being relied on that data " 1 " are arranged in the binary data on the crossover location of each matrix structure that described longitudinal axis subtask is corresponding, the binary data on the crossover location of each matrix structure that described longitudinal axis subtask is corresponding are the subtask for not relied on of " 0 ".
In this computer software the realization task decompose and the method for parallel processing control in; Described step is specially: described task scheduling unit is with the number of the described subtask number that is not relied on as the subtask sequence, but and the subtask sequence of definite executed in parallel.
Adopted realization task in the computer software of this invention to decompose and the method for parallel processing control; It is in the task scheduling unit; Dependence between each subtask is stored as the matrix structure data; But confirm the subtask sequence of executed in parallel again according to described matrix structure data, and then by the processor executed in parallel.Thereby various variations that can configuration reply subtask are dynamically adjusted accordingly, realize the decomposition of task automatically; Need not the manual work intervention; Improved the parallel task decomposing efficiency, significantly promoted the parallel task processing power of computing machine, and in the computer software of the present invention the realization task decompose and the method application mode of parallel processing control easy; Realize with low costly, range of application is comparatively extensive.
Description of drawings
Fig. 1 is that the realization task is decomposed and the process flow diagram of the method for parallel processing control in the computer software of the present invention.
Fig. 2 is that the webmaster of handling with an one process in the prior art is handled the schematic flow sheet that trap alarms program.
Fig. 3 is that the webmaster of handling with an one process in the prior art is handled 4 the required CPU time unit of function synoptic diagram that trap alarms program.
Fig. 4 handles the dependence synoptic diagram of 4 functions of trap alarm program for webmaster.
Fig. 5 utilizes realization task in the computer software of the present invention to decompose with the method for parallel processing control webmaster is handled the dependence of 4 functions that trap alarms program with the synoptic diagram of matrix structure data representation.
The webmaster that Fig. 6 handles for the method for decomposition of realization task and parallel processing control in the employing computer software of the present invention is handled the required CPU time unit of trap alarm program synoptic diagram.
Fig. 7 decomposes for realization task in the employing computer software of the present invention and the method for parallel processing control is opened two required CPU time units of trap alarm program of four thread process synoptic diagram on four CPU.
Fig. 8 is for utilizing two required CPU time units of trap alarm program of single-threaded processing synoptic diagram on two CPU in the prior art.
Fig. 9 is the basic composition synoptic diagram of existing computing machine.
Figure 10 is for utilizing realization task in the computer software of the present invention in the practical application and decompose and the method for parallel processing control being carried out the internal data Message Processing process flow diagram of EMS program.
Embodiment
In order more to be expressly understood technology contents of the present invention, the special following examples of lifting specify.
See also shown in Figure 1ly, be that realization task in the computer software of the present invention is decomposed and the process flow diagram of the method for parallel processing control.
In one embodiment, have the task scheduling unit that is connected between task receiving end and the processing unit in the described computer software, described method may further comprise the steps:
(1) described task scheduling unit receives task from described task receiving end;
(2) described task scheduling unit is split as a plurality of subtasks that can independently carry out with a task;
(3) dependence between described each subtask is confirmed in described task scheduling unit;
(4) described task scheduling unit is stored as the matrix structure data with the dependence between described each subtask;
(5) but the subtask sequence of executed in parallel is confirmed in described task scheduling unit according to described matrix structure data;
(6) described task scheduling unit is sent to described processing unit for parallel execution according to described subtask sequence with described task.
In a kind of more preferably embodiment; Described task scheduling unit is split as a plurality of subtasks that can independently carry out with a task, is specially: described task scheduling unit is split as a plurality of subtasks with independent function with a task according to the function of its required operation.Dependence between then described each subtask is specially: is condition if having first subtask of an independent function with another result with second subtask of an independent function, then exists first subtask to depend on the relation of second subtask.
In further preferred embodiment; The transverse axis and the longitudinal axis are tactic subtask in the described matrix structure data, and the transverse axis subtask that a certain transverse axis coordinate is corresponding with this position of data representation of ordinate of orthogonal axes crossover location in the described matrix structure data is for the dependence of corresponding longitudinal axis subtask, this position.Wherein, The data of described transverse axis coordinate and ordinate of orthogonal axes crossover location are represented with binary data; " 1 " representes that there is dependence corresponding transverse axis subtask, this position for corresponding longitudinal axis subtask, this position, and " 0 " representes that corresponding transverse axis subtask, this position does not have dependence for corresponding longitudinal axis subtask, this position.The subtask for being relied on that data " 1 " are arranged in the binary data on the crossover location of each matrix structure that described longitudinal axis subtask is corresponding, the binary data on the crossover location of each matrix structure that described longitudinal axis subtask is corresponding are the subtask for not relied on of " 0 ".
In preferred embodiment; Described step (5) but in confirm the subtask sequence of executed in parallel according to described matrix structure data; Be specially: described task scheduling unit is with the number of the described subtask number that is not relied on as the subtask sequence, but and the subtask sequence of definite executed in parallel.
In practical application; Build can be automatically, highly configurable subtask carries out sequence and subtask when taking the distributed system of CPU sequential; The pattern that data decomposition and data stream are decomposed all must rely on fixing data layout; The demand and the function that is to say this type systematic will be limited; And, can also accomplish abstract to hardware (CPU core number, network rate, hard disk I/O speed etc.) scheduling of resource to the functional definition very high level conceptual of system based on the parallel task disposal system of task resolution model.
Adopt matrix disassembling method to define parallel subtasks; Can accomplish configurableization of total system sequence of operations adjustment; Automatic identification to system's initial conditions change comes automatic adjustment system usefulness; The change that system is relied on hardware, network environment can be adjusted flexibly, and the whole efficiency that the parallel task that makes is handled is optimum all the time.
Therefore, the basic technical scheme of the present invention is in existing parallel task disposal system, (to comprise distributed and the unit deployment), retrieves each analysable subtask, discerns its dependence and the computing degree of depth, and with the document form record.Simultaneously, newly-increased task scheduling unit, inner with matrix-style loading subtask information, and press matrix internal information generation subtask loading sequence.
Below to handle trap alarm program with webmaster be example, specify.
As shown in Figure 2, handle in the trap alarm program at the webmaster of handling with an one process, 162 ports are monitored in a plurality of trap alarm serials in the program socket receives.Have 1 to n thread on it and handle, can move these 4 functions of analy (), insertDB (), notifyA () and notifyB () in each thread successively and handle the trap alarm.
Four functions as shown in Figure 3, above-mentioned, 50,200,200,300 cpu clock cycles consuming time respectively.Then do not have under the prerequisite of obvious bottleneck in clock period scheduling on the CPU, a trap alarm program is finished by complete process needs 750 x of chronomere.
Adopt the method for task decomposition of the present invention and parallel processing control, it at first need confirm the dependence between each function.The dependence of above-mentioned 4 functions, as shown in Figure 4, be specially:
Function A, analy () do not have dependence;
Function B, insertDB () rely on analy ();
Function C, notifyA () rely on analy ();
Function D, notifyB () rely on insertDB () and analy ().
Then, above-mentioned dependence is explained with matrix format data as shown in Figure 5.Wherein, seeing from transverse axis, is 0 entirely under function C and the function D, representes that it is the independent subtask that can be split; See on the longitudinal axis that function B, function C and function D have dependence to function A, function D has dependence to function B.
In this example, can set up the matrix format data of Fig. 5 easily by the actual dependence in program subtask of Fig. 4.And in the commercial production of complicacy; Can be in the message packets that is processed the treatment scheme information of recording messages packet; Create the dependence matrix of each subtasks in the affairs: X by Automatic Program; If have subtask nested (for example perhaps function A inside exist function A1 and A2) certainly, also can correspondingly create corresponding submatrix X-A.
As can beappreciated from fig. 5; Function C, D be not by any other functional dependence (C, two of D are 0 entirely on the longitudinal axis); Can be splitted into two independently Message Processing subtasks, and seen that from transverse axis function C has relied on function A; Function D has relied on function A and B, two subtasks such as Fig. 6 after then can obtaining splitting.
After according to above-mentioned matrix format data this task being split as two subtasks sequences as shown in Figure 6; In this parallel task; A trap alarming assignment utilizes 2 threads to handle, 550 x of chronomere consuming time altogether, than before 750 required x of chronomere of described single-threaded processing saved 200 x of chronomere; Thereby significantly shortened the task processing time, promoted treatment effeciency.
Shown in Figure 7 on four CPU, carrying out the release line journey, handle 2 trap alarm programs with 550 x of chronomere.It is compared with the diagram of on two CPU, handling 2 trap alarm programs with 750 x of chronomere shown in Figure 8, can obviously find out both gaps.
In addition; Because the basic composition of current computer is as shown in Figure 9, the bus travelling speed that connects north bridge control chip is obviously faster than the bus travelling speed that connects south bridge control chip; Caused the time of 4 above-mentioned function internal operations not wait, the computing power utilization ratio of CPU has not been waited.Wherein, Analy () function operation is the fastest, and it is pure CPU and the computing between internal memory, and fastest, the CPU utilization ratio is the highest.There are network scheduling and hard disk I/O operation in insertDB () function inside, and consuming time the longest, there is obstruction inside, and the CPU utilization ratio is minimum.NotifyA () and notifyB () have Network Transmission, need to wait for Drive Layer buffer zone message informing, and obstruction is also arranged, and be consuming time relatively longer.Therefore, except above-mentioned carrying out the simple matrix decomposition of putting down in writing, similarly, can also adopt the record function to the matrix of CPU usage and matrix of record function I/O blocking rate or the like according to the dependence between function.These matrixes also can be applied in the method for the present invention, and carry out computing according to existing mathematical programming.
Figure 10 is for utilizing realization task in the computer software of the present invention in the practical application and decompose and the method for parallel processing control being carried out the internal data Message Processing process flow diagram of EMS program.Wherein, the left side square frame is the ACCESS component processes, and middle square frame is the CONFIG component processes; The right side square frame is the ALARM component processes; Each independently passes through the communication of SOCKET mode between the process, inside has one or more thread units to come processing messages data (handler and task), because independence between process and thread process unit is relatively independent; Can flow change and configuration management, improve the overall treatment efficiency of affairs.
In the present patent application file; Only with " EMS network supervisor " Fig. 1 to Figure 10 give an example simple declaration with matrix-style storage, calculate, management is the principle of the parallel task of basic model with the Message Processing; In the practical project exploitation; Not only write down the dependence of subtask, also can write down running status and load or the like the information of subtask, through pre-configured matrix template and matrix operation formula with matrix with matrix; Come to calculate in real time, analyze the computing bottleneck of subtask, by the programming automation or the artificial sequence of calculation of adjusting the subtask.Thereby the transaction capabilities that elevator system is whole.
Through the matrix decomposition subtask, the overall performance of system improves.Adjustment System needed update routine, develops again, disposes and could improve in the past.If come analysis task and task resolution through the method for matrix decomposition; Sequence is carried out in subtask that just can the configuration system; Save cost of development greatly and dispose efficient; Make that under the large-scale cluster solution of distributed, multiprocessor, multi-process multithreading commercial applications is promoted cost and reduced.
Adopted realization task in the computer software of this invention to decompose and the method for parallel processing control; It is in the task scheduling unit; Dependence between each subtask is stored as the matrix structure data; But confirm the subtask sequence of executed in parallel again according to described matrix structure data, and then by the processor executed in parallel.Thereby various variations that can configuration reply subtask are dynamically adjusted accordingly, realize the decomposition of task automatically; Need not the manual work intervention; Improved the parallel task decomposing efficiency, significantly promoted the parallel task processing power of computing machine, and in the computer software of the present invention the realization task decompose and the method application mode of parallel processing control easy; Realize with low costly, range of application is comparatively extensive.
In this instructions, the present invention is described with reference to its certain embodiments.But, still can make various modifications and conversion obviously and not deviate from the spirit and scope of the present invention.Therefore, instructions and accompanying drawing are regarded in an illustrative, rather than a restrictive.
Claims (7)
1. the realization task is decomposed and the method for parallel processing control in the computer software; It is characterized in that; Have the task scheduling unit that is connected between task receiving end and the processing unit in the described computer software, described method may further comprise the steps:
(1) described task scheduling unit receives task from described task receiving end;
(2) described task scheduling unit is split as a plurality of subtasks that can independently carry out with a task;
(3) dependence between described each subtask is confirmed in described task scheduling unit;
(4) described task scheduling unit is stored as the matrix structure data with the dependence between described each subtask;
(5) but the subtask sequence of executed in parallel is confirmed in described task scheduling unit according to described matrix structure data;
(6) described task scheduling unit is sent to described processing unit for parallel execution according to described subtask sequence with described task.
2. the method for decomposition of realization task and parallel processing control in the computer software according to claim 1 is characterized in that described task scheduling unit is split as a plurality of subtasks that can independently carry out with a task, is specially:
Described task scheduling unit is split as a plurality of subtasks with independent function with a task according to the function of its required operation.
3. the method for decomposition of realization task and parallel processing control in the computer software according to claim 2 is characterized in that the dependence between described each subtask is specially:
If having first subtask of an independent function is condition with another result with second subtask of an independent function, then exist first subtask to depend on the relation of second subtask.
4. decompose according to realization task in each described computer software in the claim 1 to 3 and the method for parallel processing control; It is characterized in that; The transverse axis and the longitudinal axis are tactic subtask in described matrix structure data, and the transverse axis subtask that a certain transverse axis coordinate is corresponding with this position of data representation of ordinate of orthogonal axes crossover location in the described matrix structure data is for the dependence of corresponding longitudinal axis subtask, this position.
5. the method for decomposition of realization task and parallel processing control in the computer software according to claim 4; It is characterized in that; The data of described transverse axis coordinate and ordinate of orthogonal axes crossover location are represented with binary data; " 1 " representes that there is dependence corresponding transverse axis subtask, this position for corresponding longitudinal axis subtask, this position, and " 0 " representes that corresponding transverse axis subtask, this position does not have dependence for corresponding longitudinal axis subtask, this position.
6. the method for decomposition of realization task and parallel processing control in the computer software according to claim 5; It is characterized in that; The subtask for being relied on that data " 1 " are arranged in the binary data on the crossover location of each matrix structure that described longitudinal axis subtask is corresponding, the binary data on the crossover location of each matrix structure that described longitudinal axis subtask is corresponding are the subtask for not relied on of " 0 ".
7. the method for decomposition of realization task and parallel processing control in the computer software according to claim 6 is characterized in that described step (5) is specially:
Described task scheduling unit is with the number of the described subtask number that is not relied on as the subtask sequence, but and the subtask sequence of definite executed in parallel.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011103545730A CN102508704A (en) | 2011-11-10 | 2011-11-10 | Method for implementing task decomposition and parallel processing in computer software system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011103545730A CN102508704A (en) | 2011-11-10 | 2011-11-10 | Method for implementing task decomposition and parallel processing in computer software system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102508704A true CN102508704A (en) | 2012-06-20 |
Family
ID=46220796
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2011103545730A Pending CN102508704A (en) | 2011-11-10 | 2011-11-10 | Method for implementing task decomposition and parallel processing in computer software system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102508704A (en) |
Cited By (31)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102929718A (en) * | 2012-09-17 | 2013-02-13 | 江苏九章计算机科技有限公司 | Distributed GPU (graphics processing unit) computer system based on task scheduling |
| CN103279390A (en) * | 2012-08-21 | 2013-09-04 | 中国科学院信息工程研究所 | Parallel processing system for small operation optimizing |
| CN103593179A (en) * | 2012-08-17 | 2014-02-19 | 通用电气航空系统有限责任公司 | Method for developing software in a parallel computing environment |
| CN104166538A (en) * | 2013-05-16 | 2014-11-26 | 北大方正集团有限公司 | Data task processing method and system |
| CN104319875A (en) * | 2014-09-18 | 2015-01-28 | 许继电气股份有限公司 | Dynamic task loop scheduling method suitable for multifunctional integrative LED |
| CN105354087A (en) * | 2015-12-14 | 2016-02-24 | 北京交控科技有限公司 | Real-time virtual task processing method and device |
| CN106648850A (en) * | 2015-11-02 | 2017-05-10 | 佳能株式会社 | Information processing apparatus and method of controlling the same |
| CN106919451A (en) * | 2015-12-24 | 2017-07-04 | 北京趣拿科技软件有限公司 | Data processing method and device |
| CN107087010A (en) * | 2016-02-14 | 2017-08-22 | 阿里巴巴集团控股有限公司 | Intermediate data transmission method and system, distributed system |
| CN107092778A (en) * | 2017-03-30 | 2017-08-25 | 江苏骏龙光电科技股份有限公司 | A kind of optical fiber measurement parallel calculating method based on multi-core CPU |
| CN107479962A (en) * | 2016-06-08 | 2017-12-15 | 阿里巴巴集团控股有限公司 | A kind of method and apparatus for mission dispatching |
| CN107679726A (en) * | 2017-09-26 | 2018-02-09 | 成都优易数据有限公司 | A kind of evaluation method of Optimum distribution formula calculating process efficiency |
| CN108595157A (en) * | 2018-04-28 | 2018-09-28 | 百度在线网络技术（北京）有限公司 | Processing method, device, equipment and the storage medium of block chain data |
| CN108958789A (en) * | 2018-05-20 | 2018-12-07 | 湖北九州云仓科技发展有限公司 | A kind of parallel streaming calculation method, electronic equipment, storage medium and system |
| CN109327490A (en) * | 2017-07-31 | 2019-02-12 | 杭州华为数字技术有限公司 | A kind of method and apparatus for disposing cloud service component |
| CN109766515A (en) * | 2018-12-26 | 2019-05-17 | 北京思朗科技有限责任公司 | Matrix decomposition processing unit and method |
| CN110088730A (en) * | 2017-06-30 | 2019-08-02 | 华为技术有限公司 | Task processing method, device, medium and its equipment |
| WO2019218708A1 (en) * | 2018-05-15 | 2019-11-21 | 华为技术有限公司 | Task processing method and device, and computer system |
| CN110597613A (en) * | 2018-06-12 | 2019-12-20 | 成都鼎桥通信技术有限公司 | Task processing method, device, equipment and computer readable storage medium |
| CN110622128A (en) * | 2017-05-16 | 2019-12-27 | 甲骨文国际公司 | Dynamic parallelization of computational processing |
| CN111381946A (en) * | 2018-12-29 | 2020-07-07 | 上海寒武纪信息科技有限公司 | Task processing method and device and related product |
| CN111611479A (en) * | 2020-05-07 | 2020-09-01 | 北京达佳互联信息技术有限公司 | Data processing method and related device for network resource recommendation |
| CN112527471A (en) * | 2019-09-19 | 2021-03-19 | 中移(苏州)软件技术有限公司 | Task processing method and device and storage medium |
| CN112631559A (en) * | 2020-12-24 | 2021-04-09 | 中国航发控制系统研究所 | Software architecture and scheduling design method based on model development |
| CN112988364A (en) * | 2021-05-20 | 2021-06-18 | 西安芯瞳半导体技术有限公司 | Dynamic task scheduling method, device and storage medium |
| CN113254078A (en) * | 2021-06-23 | 2021-08-13 | 北京睿芯高通量科技有限公司 | Data stream processing method for efficiently executing matrix addition on GPDPU simulator |
| CN113641956A (en) * | 2021-08-05 | 2021-11-12 | 中国科学院软件研究所 | A High-performance Realization Method of Level 1 and Level 2 BLAS Function Library for SW26010-Pro Processor |
| CN113806051A (en) * | 2021-09-22 | 2021-12-17 | 海通证券股份有限公司 | Task management method and device of computing equipment, storage medium and computing equipment |
| CN114582313A (en) * | 2022-03-18 | 2022-06-03 | 广东电网有限责任公司 | Active noise control method and device based on FxLMS algorithm |
| CN118941213A (en) * | 2024-10-12 | 2024-11-12 | 君和智通(山东)大数据科技有限公司 | An intelligent warehouse scheduling system |
| CN119376698A (en) * | 2024-12-27 | 2025-01-28 | 奥特酷智能科技(南京)有限公司 | Development and operation optimization system and method for large-scale multi-component systems based on reinforcement learning |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001318798A (en) * | 2000-03-16 | 2001-11-16 | Square Co Ltd | Parallel object task engine, and parallel processing method |
| US6832378B1 (en) * | 2000-06-20 | 2004-12-14 | International Business Machines Corporation | Parallel software processing system |
| CN102077172A (en) * | 2008-07-02 | 2011-05-25 | Nxp股份有限公司 | A multiprocessor circuit using run-time task scheduling |
-
2011
- 2011-11-10 CN CN2011103545730A patent/CN102508704A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2001318798A (en) * | 2000-03-16 | 2001-11-16 | Square Co Ltd | Parallel object task engine, and parallel processing method |
| US6832378B1 (en) * | 2000-06-20 | 2004-12-14 | International Business Machines Corporation | Parallel software processing system |
| CN102077172A (en) * | 2008-07-02 | 2011-05-25 | Nxp股份有限公司 | A multiprocessor circuit using run-time task scheduling |
Cited By (48)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103593179A (en) * | 2012-08-17 | 2014-02-19 | 通用电气航空系统有限责任公司 | Method for developing software in a parallel computing environment |
| CN103279390B (en) * | 2012-08-21 | 2016-09-28 | 中国科学院信息工程研究所 | A kind of parallel processing system (PPS) towards little optimization of job |
| CN103279390A (en) * | 2012-08-21 | 2013-09-04 | 中国科学院信息工程研究所 | Parallel processing system for small operation optimizing |
| CN102929718A (en) * | 2012-09-17 | 2013-02-13 | 江苏九章计算机科技有限公司 | Distributed GPU (graphics processing unit) computer system based on task scheduling |
| CN102929718B (en) * | 2012-09-17 | 2015-03-11 | 厦门坤诺物联科技有限公司 | Distributed GPU (graphics processing unit) computer system based on task scheduling |
| CN104166538A (en) * | 2013-05-16 | 2014-11-26 | 北大方正集团有限公司 | Data task processing method and system |
| CN104319875A (en) * | 2014-09-18 | 2015-01-28 | 许继电气股份有限公司 | Dynamic task loop scheduling method suitable for multifunctional integrative LED |
| CN104319875B (en) * | 2014-09-18 | 2016-06-08 | 许继电气股份有限公司 | A kind of dynamic task ring scheduling method being applicable to integrated multifunctional IED |
| CN106648850A (en) * | 2015-11-02 | 2017-05-10 | 佳能株式会社 | Information processing apparatus and method of controlling the same |
| CN106648850B (en) * | 2015-11-02 | 2020-11-10 | 佳能株式会社 | Information processing apparatus and control method thereof |
| CN105354087A (en) * | 2015-12-14 | 2016-02-24 | 北京交控科技有限公司 | Real-time virtual task processing method and device |
| CN105354087B (en) * | 2015-12-14 | 2018-10-23 | 交控科技股份有限公司 | The virtual processing method and processing device of real-time task |
| CN106919451A (en) * | 2015-12-24 | 2017-07-04 | 北京趣拿科技软件有限公司 | Data processing method and device |
| CN106919451B (en) * | 2015-12-24 | 2020-04-21 | 北京趣拿科技软件有限公司 | Data processing method and device |
| CN107087010A (en) * | 2016-02-14 | 2017-08-22 | 阿里巴巴集团控股有限公司 | Intermediate data transmission method and system, distributed system |
| CN107479962A (en) * | 2016-06-08 | 2017-12-15 | 阿里巴巴集团控股有限公司 | A kind of method and apparatus for mission dispatching |
| CN107092778A (en) * | 2017-03-30 | 2017-08-25 | 江苏骏龙光电科技股份有限公司 | A kind of optical fiber measurement parallel calculating method based on multi-core CPU |
| CN110622128A (en) * | 2017-05-16 | 2019-12-27 | 甲骨文国际公司 | Dynamic parallelization of computational processing |
| CN110088730B (en) * | 2017-06-30 | 2021-05-18 | 华为技术有限公司 | Task processing method, device, medium and equipment |
| CN110088730A (en) * | 2017-06-30 | 2019-08-02 | 华为技术有限公司 | Task processing method, device, medium and its equipment |
| CN109327490A (en) * | 2017-07-31 | 2019-02-12 | 杭州华为数字技术有限公司 | A kind of method and apparatus for disposing cloud service component |
| CN109327490B (en) * | 2017-07-31 | 2021-01-15 | 华为技术有限公司 | Method and server for deploying cloud service component |
| CN107679726A (en) * | 2017-09-26 | 2018-02-09 | 成都优易数据有限公司 | A kind of evaluation method of Optimum distribution formula calculating process efficiency |
| CN108595157A (en) * | 2018-04-28 | 2018-09-28 | 百度在线网络技术（北京）有限公司 | Processing method, device, equipment and the storage medium of block chain data |
| WO2019218708A1 (en) * | 2018-05-15 | 2019-11-21 | 华为技术有限公司 | Task processing method and device, and computer system |
| US11941434B2 (en) | 2018-05-15 | 2024-03-26 | Huawei Technologies Co., Ltd. | Task processing method, processing apparatus, and computer system |
| CN108958789A (en) * | 2018-05-20 | 2018-12-07 | 湖北九州云仓科技发展有限公司 | A kind of parallel streaming calculation method, electronic equipment, storage medium and system |
| CN108958789B (en) * | 2018-05-20 | 2021-07-09 | 湖北九州云仓科技发展有限公司 | Parallel stream type computing method, electronic equipment, storage medium and system |
| CN110597613A (en) * | 2018-06-12 | 2019-12-20 | 成都鼎桥通信技术有限公司 | Task processing method, device, equipment and computer readable storage medium |
| CN109766515B (en) * | 2018-12-26 | 2023-04-14 | 上海思朗科技有限公司 | Matrix decomposition processing device and method |
| CN109766515A (en) * | 2018-12-26 | 2019-05-17 | 北京思朗科技有限责任公司 | Matrix decomposition processing unit and method |
| CN111381946A (en) * | 2018-12-29 | 2020-07-07 | 上海寒武纪信息科技有限公司 | Task processing method and device and related product |
| CN111381946B (en) * | 2018-12-29 | 2022-12-09 | 上海寒武纪信息科技有限公司 | Task processing method and device and related products |
| CN112527471A (en) * | 2019-09-19 | 2021-03-19 | 中移(苏州)软件技术有限公司 | Task processing method and device and storage medium |
| CN111611479A (en) * | 2020-05-07 | 2020-09-01 | 北京达佳互联信息技术有限公司 | Data processing method and related device for network resource recommendation |
| CN111611479B (en) * | 2020-05-07 | 2024-02-13 | 北京达佳互联信息技术有限公司 | Data processing method and related device for network resource recommendation |
| CN112631559A (en) * | 2020-12-24 | 2021-04-09 | 中国航发控制系统研究所 | Software architecture and scheduling design method based on model development |
| CN112988364A (en) * | 2021-05-20 | 2021-06-18 | 西安芯瞳半导体技术有限公司 | Dynamic task scheduling method, device and storage medium |
| CN112988364B (en) * | 2021-05-20 | 2021-09-24 | 西安芯瞳半导体技术有限公司 | Dynamic task scheduling method, device and storage medium |
| CN113254078A (en) * | 2021-06-23 | 2021-08-13 | 北京睿芯高通量科技有限公司 | Data stream processing method for efficiently executing matrix addition on GPDPU simulator |
| CN113254078B (en) * | 2021-06-23 | 2024-04-12 | 北京中科通量科技有限公司 | Data stream processing method for efficiently executing matrix addition on GPDPU simulator |
| CN113641956A (en) * | 2021-08-05 | 2021-11-12 | 中国科学院软件研究所 | A High-performance Realization Method of Level 1 and Level 2 BLAS Function Library for SW26010-Pro Processor |
| CN113641956B (en) * | 2021-08-05 | 2023-05-30 | 中国科学院软件研究所 | High-performance implementation method of 1, 2-level BLAS function library facing SW26010-Pro processor |
| CN113806051B (en) * | 2021-09-22 | 2022-08-19 | 海通证券股份有限公司 | Task management method and device of computing equipment, storage medium and computing equipment |
| CN113806051A (en) * | 2021-09-22 | 2021-12-17 | 海通证券股份有限公司 | Task management method and device of computing equipment, storage medium and computing equipment |
| CN114582313A (en) * | 2022-03-18 | 2022-06-03 | 广东电网有限责任公司 | Active noise control method and device based on FxLMS algorithm |
| CN118941213A (en) * | 2024-10-12 | 2024-11-12 | 君和智通(山东)大数据科技有限公司 | An intelligent warehouse scheduling system |
| CN119376698A (en) * | 2024-12-27 | 2025-01-28 | 奥特酷智能科技(南京)有限公司 | Development and operation optimization system and method for large-scale multi-component systems based on reinforcement learning |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102508704A (en) | Method for implementing task decomposition and parallel processing in computer software system | |
| Wang et al. | Optimizing load balancing and data-locality with data-aware scheduling | |
| EP3256915B1 (en) | Extending a programmable logic controller with apps | |
| US20110296137A1 (en) | Performing A Deterministic Reduction Operation In A Parallel Computer | |
| Sugata et al. | Acceleration of publish/subscribe messaging in ROS-compliant FPGA component | |
| CN103488775A (en) | Computing system and computing method for big data processing | |
| Luo et al. | Adapt: An event-based adaptive collective communication framework | |
| WO2008121657A1 (en) | A recipe-and-component control module and methods thereof | |
| WO2023274278A1 (en) | Resource scheduling method and device and computing node | |
| JP2022508035A (en) | Large-scale model support for deep learning | |
| CN115204399A (en) | Quantum computing task computing method and device and quantum computer operating system | |
| EP2278454B1 (en) | Method for correct-by-construction development of real-time-systems | |
| CN116821041A (en) | Efficient queue access for user space packet processing | |
| Morozov et al. | ALCF MPI benchmarks: Understanding machine-specific communication behavior | |
| US20150067068A1 (en) | Data Communications In A Distributed Computing Environment | |
| Penmatsa et al. | Implementation of distributed loop scheduling schemes on the teragrid | |
| Tejedor et al. | An architecture for attacking the bottleneck communication in P systems | |
| US9170966B2 (en) | System method for deterministic message processing a direct memory access (DMA) adapter with head pointers and tail pointers for determining the in and out sequence of packets | |
| Pallickara et al. | Enabling large scale scientific computations for expressed sequence tag sequencing over grid and cloud computing clusters | |
| EP3502897A1 (en) | Efficient communication overlap by runtimes collaboration | |
| US9152590B2 (en) | Deterministic message processing in a direct memory access adapter | |
| US9027033B1 (en) | Administering message acknowledgements in a parallel computer | |
| US20230004322A1 (en) | Managing provenance information for data processing pipelines | |
| Wexler et al. | Solving problems with transputers: background and experience | |
| Li et al. | Improving performance by replicating simulations with alternative synchronization approaches |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20120620 |