CN103631751B - A kind of multitask set partitioning method based on connection features - Google Patents

A kind of multitask set partitioning method based on connection features Download PDF

Info

Publication number
CN103631751B
CN103631751B CN201310692186.7A CN201310692186A CN103631751B CN 103631751 B CN103631751 B CN 103631751B CN 201310692186 A CN201310692186 A CN 201310692186A CN 103631751 B CN103631751 B CN 103631751B
Authority
CN
China
Prior art keywords
task
multitask
tasks
correspondence
connection features
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.)
Active
Application number
CN201310692186.7A
Other languages
Chinese (zh)
Other versions
CN103631751A (en
Inventor
胡威
邹代坤
黎文飞
张凯
郭宏
李伟强
张若凡
徐景
谭练
江若成
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Juliang Sci Tech Innovation Technology Co.,Ltd.
Original Assignee
Wuhan University of Science and Engineering WUSE
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Wuhan University of Science and Engineering WUSE filed Critical Wuhan University of Science and Engineering WUSE
Priority to CN201310692186.7A priority Critical patent/CN103631751B/en
Publication of CN103631751A publication Critical patent/CN103631751A/en
Application granted granted Critical
Publication of CN103631751B publication Critical patent/CN103631751B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Landscapes

  • Multi Processors (AREA)

Abstract

The present invention relates to a kind of multitask set partitioning method based on connection features.The step of its technical scheme is: step 1 is for multi task model is set up in multitask; Step 2 is the connection features factor θ calculating each task i; Step 3 is the associated task set of setting up each task; Step 4 divides multitask set T by task relation integration; Step 5 is according to connection features factor θ icarry out multitask set division.The present invention is applicable to the division multiple task being carried out to set of tasks, takes full advantage of the connection between multiple task and correspondence, fast and effeciently multiple task division can be become different set.The present invention can provide basic multitask to divide set for multiple task management, multi-task scheduling and multitask mapping etc., the efficiency improving management, scheduling and map.

Description

A kind of multitask set partitioning method based on connection features
Technical field
The invention belongs to network-on-chip technical field, particularly relate to a kind of multitask set partitioning method based on connection features.
Background technology
According to Moore's Law, the speed of microprocessor and degree of monolithic integration will double for every 18 months.Semi-conductor industry is follow the rhythm of Moore's Law in the development of nearly decades always, and the frequency of microprocessor also constantly rises.Along with the dominant frequency of general processor breaks through 4GHz, it is found that the way of single lifting dominant frequency can not improve performance more effectively, but bring rising sharply of power consumption on the contrary, high-frequency road has gone to the end gradually.
So the research for computer processor starts the direction turning to multiprocessing core.Mostly early stage symmetric multiprocessor (SMP, SymmetricMulti-Processor) is to adopt the mode collecting one group of CPU on the same computer, shared drive subsystem and bus structure between them.Afterwards due to the introducing of nanometer fabrication technology, SMP starts to change chip multiprocessors (CMP, ChipMultiprocessor) into, i.e. integrated multiple process core on the same chip, defines our said polycaryon processor now.Between multi-core, direct shared buffer memory and bus structure, greatly reduce wire delay, significantly improve communication efficiency.
By process core equity whether, polycaryon processor can be divided into isomorphism multinuclear and heterogeneous polynuclear.Process nuclear phase with, status reciprocity be called isomorphism multinuclear.The polycaryon processor of now popular on the market Intel and AMD is exactly the polycaryon processor of isomorphism.Process core is different, status is not reciprocity is called heterogeneous polynuclear.The design that heterogeneous polynuclear generally adopts " primary processor+coprocessor ".The model of the Cell processor that IBM, Sony and Toshiba etc. put out jointly this isomery framework just.The structural relation of processor itself is to the area of whole chip, power consumption and performance.How the achievement of inherit and development conventional processors directly has influence on performance and the performance period of polycaryon processor.
Sometimes need to carry out data sharing with synchronous between the program of each process core execution of polycaryon processor, therefore its hardware configuration must support intercore communication.Efficient communication mechanism is the high performance important leverage of polycaryon processor.On current sheet, efficient communication mechanism has two kinds usually: based on the cache structure of shared bus, based on the interconnection structure of network-on-chip.Cache structure based on shared bus refers to that each process core has shared secondary or three grades of cache, for preserving relatively more conventional data, and is communicated by bus.The advantage of this system is that structure is simple, and communication speed is fast; Shortcoming is poor expandability.
Shared bus obviously cannot meet the needs of large scale system.Interconnection network are used for system-on-chip designs, solve the Communication between assembly on sheet, Here it is network-on-chip.Network-on-chip (-NetworkOnChip, NoC) technology is accessed with its support simultaneously, reliability is high, reusability high is considered to more desirable extensive CMP interconnection technique.Network-on-chip overcomes the shortcoming of bus structure poor expandability, is to provide a kind of feasible SOC (system on a chip) communication mechanism 1,000,000,000 transistor epoch.Network-on-chip, except connecting more IP assembly, compared with bus structure, also has the features such as high reusability.
In system-on-chip designs, reusability is an important principle of design.Reusability design can save design cost, improves the reliability of design, shortens the market periods of product.Based in the system-on-chip designs of bus, each IP assembly is reused, but communication structure cannot be reused.Each design needs to redesign communication structure.In network-on-chip, except each assembly is reusable, the Communications service on Communication structure and sheet is also reusable.When designing new system, original system is added router and new functional part just passable, former design obtains reuses, and greatly accelerates the progress of design.Simultaneously, network-on-chip also has the feature of low-power consumption, it adopts the design of Global Asynchronous, local synchronization, communication modes end to end, the assembly participating in communication is only had to be activate, avoid in bus structure the power wastage adopting broadcast mode to carry out the system that communication causes, therefore greatly reduce the power consumption of system.
In network-on-chip, provide abundant computational resource, realize the parallel of multitask, this has just needed scheduling, the management of multitask, especially maps.How will duty mapping to the processor core of network-on-chip be very important problem, and first this depend on how to divide multitask set.Carry out in the mode dividing task image traditionally, very consuming time, and owing to being np complete problem, optimum solution can only be sought with the algorithm of complexity.
Summary of the invention
The present invention is intended to overcome prior art defect, and object is to provide a kind of multitask set partitioning method based on connection features, and the method can improve the division efficiency of multitask set.
For achieving the above object, the step of the technical solution used in the present invention is:
Step 1, set up multi task model
For multitask, set up multi task model G (T, P, Q), wherein:
T is the set of task, T={t 0, t 1..., t m.
P is p ijset, p ij=1 represents task t iwith task t jbetween there is correspondence, p ij=0 represents task t iwith task t jbetween there is not correspondence.
Q is q ijset, q ij=1 represents task t iwith task t jbetween there is not correspondence, but by task t iand the correspondence between other tasks and by task t jand the correspondence between other tasks, task t iwith task t jcan be connected.
The attribute that multi task model G (T, P, Q) has is:
D (q ij) be connected relation q ijattribute, represent task t iwith task t jbetween connection required for the task quantity of process.
W ijtask t iattribute, w ijexpression task t iwith task t jbetween the traffic, W is w ijset.
L itask t iattribute, represent with task t ithere is the quantity of the task of correspondence.
H itask t iattribute, represent task t iall traffic sums.
Step 2, calculate the connection features factor θ of each task i
Task t iconnection features factor θ ifor:
θ i=L i×lg(H i)(1)
Then to all tasks according to connection features factor θ isize carry out descending sort, form multitask set T '; In sequencer procedure, if multiple task has the connection features factor of formed objects, then carry out descending sort according to the sequence number size of multiple task.
Step 3, set up task t iassociated task set
For the task t in multitask set T i, task t iassociated task S set ifor with task t ithere is the set of all tasks of correspondence or connected relation.For associated task S set iin with task t ithere is the task t of correspondence j, S i(t j)=0; For associated task S set iin with task t ithere is the task t of connected relation j, S i(t j)=D (q ij).
Wherein, S i(t j) be task t iwith task t jbetween association required for the task quantity of process; If S i(t j)=0, represents task t iwith task t jbetween association required for the task quantity of process be 0; If S i(t j)=D (q ij), represent task t iwith task t jbetween association required for the task quantity of process be D (q ij).
Step 4, by task relation integration Si, multitask set T to be divided
According to task t irelevance, to multitask set T according to task t ibetween association divide, be divided into g each other without any the set V of association 1, V 2..., V g.Concrete steps are:
Step 4.1, for task t 0, by task t 0and S set 0in all tasks join set V 1in the middle of.
Step 4.2, for not set V 1in task t i, by task t iand S set iin all tasks join set V 2in the middle of.
Step 4.3, for not set V 1with set V 2in task t j, by task t jand S set jin all tasks join set V 3in the middle of.
Step 4.4, when to proceed to kth step according to step 4.1, step 4.2 and step 4.3, for not at set V 1, V 2... V k-1in task t c, by task t cand S set cin all tasks join set V kin the middle of; Until complete g step, multitask set T is divided into set V 1, V 2..., V g.
Step 5, according to connection features factor θ icarry out multitask set division
For the set V generated in step 4 1, V 2..., V gfurther divide, concrete steps are:
Step 5.1, arrange converging factor I, I is 0 or natural number.
Step 5.2, for set V 1, V 2..., V gin one set V i, for gathering V iin and multitask set T ' sort first task t x, according to the sequence in multitask set T ', check task t xwith task t in multitask set T ' ybetween connected relation; If S x(t y) <I, then by task t yfrom set V imiddle removal, sets up set V g+1, and by task t yadd set V i'.
Step 5.3, to all set V i', all operate according to step 5.1 and step 5.2, generate until no longer include new set; Wherein each when operating according to step 5.1, need converging factor I be reset.
Step 5.4, for only having a task t iset, pass through p ijfind the task t that corresponding mission number is minimum jthe set at place, by task t ithe set at place and task t jthe set at place merges, and division completes.
Owing to adopting technique scheme, present invention utilizes the correspondence between multitask and the traffic, calculate the connection features factor, and with the connection features factor, multitask is divided, realize the quick division of multitask from new angle, improve the efficiency of division.The present invention compared with prior art, has following good effect:
(1) high efficiency.Often can support a large amount of tasks in network-on-chip, due to task One's name is legion, how carry out the division of set of tasks, thus support that mapping efficiently, reaching effective utilization of network-on-chip is the problem needing to solve.In the present invention, carry out the division of multitask with the connection features factor, decrease the complexity of division methods, improve the speed of division, therefore there is higher efficiency.
(2) practicality.In existing task division, most method of complexity that all adopts realizes, although can reach certain effect of optimization, because method is complicated, is often faced with the problem of various reality when realizing.In the present invention, decrease the dependence to condition, thus the complexity of the method that reduces when realizing, thus there is very strong practicality.
Therefore, the present invention is applicable to the division multiple task being carried out to set of tasks, takes full advantage of the connection between multiple task and correspondence, fast and effectively multiple task division can be become different set.For multiple task management, multi-task scheduling and multitask mapping etc. provide basic multitask to divide set, the efficiency improving management, scheduling and map.
Accompanying drawing explanation
Fig. 1 is a kind of method schematic diagram of the present invention;
Fig. 2 is a kind of multitask relation schematic diagram of the present invention.
Embodiment
Below in conjunction with the drawings and specific embodiments, the present invention will be further described, the restriction not to its protection domain.
A kind of multitask set partitioning method based on connection features.The step of the method is as shown in Figure 1:
Step 1, set up multi task model
For multitask, set up multi task model G (T, P, Q), wherein:
T is the set of task, T={t 0, t 1..., t m.
P is p ijset, p ij=1 represents task t iwith task t jbetween there is correspondence, p ij=0 represents task t iwith task t jbetween there is not correspondence.
Q is q ijset, q ij=1 represents task t iwith task t jbetween there is not correspondence, but by task t iand the correspondence between other tasks and by task t jand the correspondence between other tasks, task t iwith task t jcan be connected.
The attribute that multi task model G (T, P, Q) has is:
D (q ij) be connected relation q ijattribute, represent task t iwith task t jbetween connection required for the task quantity of process.
W ijtask t iattribute, w ijexpression task t iwith task t jbetween the traffic, W is w ijset.
L itask t iattribute, represent with task t ithere is the quantity of the task of correspondence.
H itask t iattribute, represent task t iall traffic sums.
For the set of tasks with 11 tasks, its multi task model G (T, P, Q) is as follows:
T={t 0,t 1,t 2,t 3,t 4,t 5,t 6,t 7,t 8,t 9,t 10}。
P, Q, W as shown in Figure 2, p ij=1 represents two task t iand t jbetween have line to exist, p ij=0 represents two task t iand t jbetween there is no line; Two task t iand t jbetween line on numeral w ij; If there is line between any two tasks these two tasks to be linked together, then these two tasks are communicated with.
L ibe worth as shown in table 1:
The L of table 1 task ivalue
t 0 t 1 t 2 t 3 t 4 t 5 t 6 t 7 t 8 t 9 t 10
L i 4 3 2 2 2 3 2 2 3 4 3
Hi value is as shown in table 2:
The H of table 2 task ivalue
t 0 t 1 t 2 t 3 t 4 t 5 t 6 t 7 t 8 t 9 t 10
H i 900 630 500 430 430 780 560 420 650 670 410
Step 2, calculate the connection features factor θ of each task i
Task t iconnection features factor θ i:
θ i=L i×lg(H i)(1)
Then to all tasks according to connection features factor θ isize carry out descending sort, form multitask set T '; In sequencer procedure, if two tasks have the connection features factor of formed objects, then carry out descending sort according to the sequence number size of multiple task.
For the aforementioned multitask set T={t with 11 tasks 0, t 1, t 2, t 3, t 4, t 5, t 6, t 7, t 8, t 9, t 10,
The connection features factor calculating 11 tasks according to table 1 and table 2 is as shown in table 3:
The connection features factor θ of table 3 task i
t 0 t 1 t 2 t 3 t 4 t 5 t 6 t 7 t 8 t 9 t 10
θ i 11.817 8.398 5.398 5.267 5.267 8.676 5.496 5.246 8.439 11.304 7.838
According to connection features factor θ isize sort after formed new set T ' be:
T’={t 0,t 9,t 5,t 8,t 1,t 10,t 6,t 2,t 3,t 4,t 7}。
Step 3, set up task t iassociated task set
For the task t in multitask set T i, task t iassociated task S set ifor with task t ithere is the set of all tasks of correspondence or connected relation.For associated task S set iin with task t ithere is the task t of correspondence j, S i(t j)=0; For associated task S set iin with task t ithere is the task t of connected relation j, S i(t j)=D (q ij).
Wherein, S i(t j) be task t iwith task t jbetween association required for the task quantity of process; If S i(t j)=0, represents task t iwith task t jbetween association required for the task quantity of process be 0; If S i(t j)=D (q ij), represent task t iwith task t jbetween association required for the task quantity of process be D (q ij).
For the set of tasks with 11 tasks, its multi task model G (T, P) is as follows:
T={t 0,t 1,t 2,t 3,t 4,t 5,t 6,t 7,t 8,t 9,t 10};
Then according to shown in Fig. 2, its associated task set is as shown in table 4:
The associated task set of table 4 task
Step 4, by task relation integration, multitask set to be divided
According to task t irelevance, to multitask set T according to task t ibetween association divide, be divided into g each other without any the set V of association 1, V 2..., V g.Concrete steps are:
Step 4.1, for task t 0, by task t 0and S set 0in all tasks join set V 1in the middle of.
Step 4.2, for not set V 1in task t i, by task t iand S set iin all tasks join set V 2in the middle of.
Step 4.3, for not set V 1with set V 2in task t j, by task t jand S set jin all tasks join set V 3in the middle of.
Step 4.4, when to proceed to kth step according to step 4.1, step 4.2 and step 4.3, for not at set V 1, V 2... V k-1in task t c, by task t cand S set cin all tasks join set V kin the middle of; Until complete g step, multitask set T is divided into set V 1, V 2..., V g.
For the set of tasks with 11 tasks, its multi task model G (T, P) is as follows:
T={t 0,t 1,t 2,t 3,t 4,t 5,t 6,t 7,t 8,t 9,t 10}。
The set then marked off is as follows:
Step 4.1, V 1={ t 0, t 1, t 5, t 8, t 9, t 10, t 3, t 4.
Step 4.2, V 2={ t 2, t 6, t 7.
Step 5, carry out multitask set division according to the connection features factor
For the V generated in step 4 1, V 2..., V gfurther divide, concrete steps are:
Step 5.1, arrange converging factor I, I is 0 or natural number.
Step 5.2, for set V 1, V 2..., V gin one set V i, for gathering V iin and multitask set T ' sort first task t x, according to the sequence in multitask set T ', check task t xwith task t in multitask set T ' ybetween connected relation; If S x(t y) <I, then by task t yfrom set V imiddle removal, sets up set V g+1, and by task t yadd set V i'.
Step 5.3, to all set V i', all operate according to step 5.1 and step 5.2, generate until no longer include new set; Wherein each when operating according to step 5.1, need converging factor I be reset.
Step 5.4, for only having a task t iset, pass through p ijfind the task t that corresponding mission number is minimum jthe set at place, by task t ithe set at place and task t jthe set at place merges, and division completes.
For the set of tasks with 11 tasks, its multi task model G (T, P) is as follows:
T={t 0,t 1,t 2,t 3,t 4,t 5,t 6,t 7,t 8,t 9,t 10}。
By step 4, the set V marked off is:
V 1={t 0,t 1,t 5,t 8,t 9,t 10,t 3,t 4}。
V 2={t 2,t 6,t 7}。
Arranging converging factor is 8, then for V 1, due to t 10, t 3, t 4in these three tasks, S 0(t 10), S 0(t 4) and S 0(t 3) be all greater than converging factor 8, then t 10, t 3, t 4be added into new set V 3.
New converging factor 5 is set again, then t 10, t 3, t 4the set V formed 3no longer adjust;
For V 2, arranging new converging factor is 5, then V 2no longer adjust;
Set after division is:
V 1={t 0,t 9,t 5,t 8,t 1}
V 2={t 2,t 6,t 7}
V 3={t 10,t 3,t 4}
This embodiment make use of correspondence between multitask and the traffic, calculates connection features factor θ i, and with connection features factor θ imultitask is divided, realizes the quick division of multitask from new angle, improve the efficiency of division.The present invention compared with prior art, has following good effect:
(1) high efficiency.Often can support a large amount of tasks in network-on-chip, due to task One's name is legion, how carry out the division of set of tasks, thus support that mapping efficiently, reaching effective utilization of network-on-chip is the problem needing to solve.In the present invention, with connection features factor θ icarry out the division of multitask, decrease the complexity of division methods, improve the speed of division, therefore there is higher efficiency;
(2) practicality.In existing task division, most method of complexity that all adopts realizes, although can reach certain effect of optimization, because method is complicated, is often faced with the problem of various reality when realizing.In the present invention, decrease the dependence to condition, thus the complexity of the method that reduces when realizing, thus there is very strong practicality.
Therefore, the present invention is applicable to the division multiple task being carried out to set of tasks, takes full advantage of the connection between multiple task and correspondence, fast and effectively multiple task division can be become different set.For multiple task management, multi-task scheduling and multitask mapping etc. provide basic multitask to divide set, the efficiency improving management, scheduling and map.

Claims (1)

1., based on a multitask set partitioning method for connection features, it is characterized in that the step realizing the method is as follows:
Step 1, set up multi task model
For multitask, set up multi task model G (T, P, Q), wherein:
T is the set of task, T={t 0, t 1..., t m;
P is p ijset, p ij=1 represents task t iwith task t jbetween there is correspondence, p ij=0 represents task t iwith task t jbetween there is not correspondence;
Q is q ijset, q ij=1 represents task t iwith task t jbetween there is not correspondence, but by task t iand the correspondence between other tasks and by task t jand the correspondence between other tasks, task t iwith task t jcan be connected;
The attribute that multi task model G (T, P, Q) has is:
D (q ij) be connected relation q ijattribute, represent task t iwith task t jbetween connection required for the task quantity of process;
W ijtask t iattribute, w ijexpression task t iwith task t jbetween the traffic, W is w ijset;
L itask t iattribute, represent with task t ithere is the quantity of the task of correspondence;
H itask t iattribute, represent task t iall traffic sums;
Step 2, calculate the connection features factor θ of each task i
Task t iconnection features factor θ ifor:
θ i=L i×lg(H i)(1)
Then to all tasks according to connection features factor θ isize carry out descending sort, form multitask set T '; In sequencer procedure, if multiple task has the connection features factor of formed objects, then carry out descending sort according to the sequence number size of multiple task;
Step 3, set up task t iassociated task set
For the task t in multitask set T i, task t iassociated task S set ifor with task t ithere is the set of all tasks of correspondence or connected relation; For associated task S set iin with task t ithere is the task t of correspondence j, S i(t j)=0; For associated task S set iin with task t ithere is the task t of connected relation j, S i(t j)=D (q ij);
Wherein, S i(t j) be task t iwith task t jbetween association required for the task quantity of process; If S i(t j)=0, represents task t iwith task t jbetween association required for the task quantity of process be 0; If S i(t j)=D (q ij), represent task t iwith task t jbetween association required for the task quantity of process be D (q ij);
Step 4, by associated task S set imultitask set T is divided
According to task t irelevance, to multitask set T according to task t ibetween association divide, be divided into g each other without any the set V of association 1, V 2..., V g, concrete steps are:
Step 4.1, for task t 0, by task t 0and S set 0in all tasks join set V 1in the middle of;
Step 4.2, for not set V 1in task t i, by task t iand S set iin all tasks join set V 2in the middle of;
Step 4.3, for not set V 1with set V 2in task t j, by task t jand S set jin all tasks join set V 3in the middle of;
Step 4.4, when to proceed to kth step according to step 4.1, step 4.2 and step 4.3, for not at set V 1, V 2... V k-1in task t c, by task t cand S set cin all tasks join set V kin the middle of; Until complete g step, multitask set T is divided into set V 1, V 2..., V g;
Step 5, according to connection features factor θ icarry out multitask set division
For the set V generated in step 4 1, V 2..., V gfurther divide, concrete steps are:
Step 5.1, arrange converging factor I, I is 0 or natural number;
Step 5.2, for set V 1, V 2..., V gin one set V i, for gathering V iin and multitask set T ' sort first task t x, according to the sequence in multitask set T ', check task t xwith task t in multitask set T ' ybetween connected relation; If S x(t y) <I, then by task t yfrom set V imiddle removal, sets up set V g+1, and by task t yadd set V i';
Step 5.3, to all set V i', all operate according to step 5.1 and step 5.2, generate until no longer include new set; Wherein each when operating according to step 5.1, need converging factor I be reset;
Step 5.4, for only having a task t iset, pass through p ijfind the task t that corresponding mission number is minimum jthe set at place, by task t ithe set at place and task t jthe set at place merges, and division completes.
CN201310692186.7A 2013-12-17 2013-12-17 A kind of multitask set partitioning method based on connection features Active CN103631751B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310692186.7A CN103631751B (en) 2013-12-17 2013-12-17 A kind of multitask set partitioning method based on connection features

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310692186.7A CN103631751B (en) 2013-12-17 2013-12-17 A kind of multitask set partitioning method based on connection features

Publications (2)

Publication Number Publication Date
CN103631751A CN103631751A (en) 2014-03-12
CN103631751B true CN103631751B (en) 2016-04-27

Family

ID=50212829

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310692186.7A Active CN103631751B (en) 2013-12-17 2013-12-17 A kind of multitask set partitioning method based on connection features

Country Status (1)

Country Link
CN (1) CN103631751B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106547701B (en) * 2015-09-17 2020-01-10 慧荣科技股份有限公司 Memory device and data reading method

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6779183B1 (en) * 1999-12-23 2004-08-17 Lucent Technologies Inc. Method and system for load scheduling multidimensional tasks among multiple processors
CN1818875A (en) * 2006-03-16 2006-08-16 浙江大学 Grouped hard realtime task dispatching method of built-in operation system
CN1818868A (en) * 2006-03-10 2006-08-16 浙江大学 Multi-task parallel starting optimization of built-in operation system
CN102681902A (en) * 2012-05-15 2012-09-19 浙江大学 Load balancing method based on task distribution of multicore system

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4781089B2 (en) * 2005-11-15 2011-09-28 株式会社ソニー・コンピュータエンタテインメント Task assignment method and task assignment device

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6779183B1 (en) * 1999-12-23 2004-08-17 Lucent Technologies Inc. Method and system for load scheduling multidimensional tasks among multiple processors
CN1818868A (en) * 2006-03-10 2006-08-16 浙江大学 Multi-task parallel starting optimization of built-in operation system
CN1818875A (en) * 2006-03-16 2006-08-16 浙江大学 Grouped hard realtime task dispatching method of built-in operation system
CN102681902A (en) * 2012-05-15 2012-09-19 浙江大学 Load balancing method based on task distribution of multicore system

Also Published As

Publication number Publication date
CN103631751A (en) 2014-03-12

Similar Documents

Publication Publication Date Title
US9479449B2 (en) Workload partitioning among heterogeneous processing nodes
Wang et al. SODA: Software defined FPGA based accelerators for big data
CN107463442B (en) Satellite-borne multi-core SoC task level load balancing parallel scheduling method
Zidenberg et al. Multiamdahl: How should i divide my heterogenous chip?
CN102193779A (en) MPSoC (multi-processor system-on-chip)-oriented multithread scheduling method
CN102253919A (en) Parallel numerical simulation method and system based on GPU and CPU cooperative operation
Huang et al. Triangle counting and truss decomposition using FPGA
Sato et al. Co-design and system for the supercomputer “Fugaku”
Fallin et al. The heterogeneous block architecture
Yu et al. A 16-core processor with shared-memory and message-passing communications
CN101625673A (en) Method for mapping task of network on two-dimensional grid chip
CN103631751B (en) A kind of multitask set partitioning method based on connection features
Majumder et al. High-throughput, energy-efficient network-on-chip-based hardware accelerators
CN103631659B (en) Schedule optimization method for communication energy consumption in on-chip network
Zidenberg et al. Optimal resource allocation with multiamdahl
CN103678245B (en) Low-power-consumption on-chip network task mapping method
CN102023846A (en) Shared front-end assembly line structure based on monolithic multiprocessor system
Hsiu et al. Multilayer bus optimization for real-time embedded systems
Wu et al. Runtime support for adaptive power capping on heterogeneous socs
CN104699520B (en) A kind of power-economizing method based on virtual machine (vm) migration scheduling
Bingham et al. Modeling energy-time trade-offs in VLSI computation
Pei et al. Reevaluating the overhead of data preparation for asymmetric multicore system on graphics processing
CN103634207B (en) The network-on-chip routing optimization method that the critical path of a kind of static state is preferential
CN103838631A (en) Multi-thread scheduling realization method oriented to network on chip
Iizuka et al. Power Analysis and Power Modeling of Directly-Connected FPGA Clusters

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
TR01 Transfer of patent right
TR01 Transfer of patent right

Effective date of registration: 20191230

Address after: 313000 1-C, building 1, No. 656, Qixing Road, high tech Zone, Wuxing District, Huzhou City, Zhejiang Province

Patentee after: Huzhou xinbeilian Network Technology Co.,Ltd.

Address before: 430081 construction of Qingshan District, Hubei, Wuhan

Patentee before: Wuhan University of Science and Technology

TR01 Transfer of patent right
TR01 Transfer of patent right

Effective date of registration: 20240705

Address after: No. 6-10, North of Building 2, Fengxiang East Street, Yangsong Town, Huairou District, Beijing 101400

Patentee after: Beijing Juliang Sci Tech Innovation Technology Co.,Ltd.

Country or region after: China

Address before: 313000 1-C, building 1, No. 656 Qixing Road, high tech Zone, Wuxing District, Huzhou City, Zhejiang Province

Patentee before: Huzhou xinbeilian Network Technology Co.,Ltd.

Country or region before: China