CN102681901A - Segmental reconfigurable hardware task arranging method - Google Patents
Segmental reconfigurable hardware task arranging method Download PDFInfo
- Publication number
- CN102681901A CN102681901A CN2012101394675A CN201210139467A CN102681901A CN 102681901 A CN102681901 A CN 102681901A CN 2012101394675 A CN2012101394675 A CN 2012101394675A CN 201210139467 A CN201210139467 A CN 201210139467A CN 102681901 A CN102681901 A CN 102681901A
- Authority
- CN
- China
- Prior art keywords
- task
- reconfigurable hardware
- ready
- hardware task
- reconfigurable
- 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.)
- Granted
Links
Images
Landscapes
- Design And Manufacture Of Integrated Circuits (AREA)
Abstract
The invention provides a segmental reconfigurable hardware task arranging method. The method comprises the following steps of: constructing a two-dimensional coordinate system consisting of an axis X and an axis Y on a two-dimensional reconfigurable resource model; partitioning a reconfigurable hardware task set into a plurality of groups according to the closeness degree of communication among tasks; partitioning the reconfigurable resource model into a plurality of resource regions which correspond to hardware task groups one by one; selecting a layout region which corresponds to a task according to a group where the task is to be arranged; and distributing the task to be arranged in the region according to a corresponding layout strategy. The reconfigurable hardware task set is partitioned into a plurality of groups according to the closeness degree of communication among tasks specific to the condition of communication closeness among the reconfigurable hardware tasks, device resources are managed in a segmental way on the basis of the two-dimensional reconfigurable resource model, and tasks of different groups are distributed into different regions, so that difficulty in wiring in a reconfigurable device is lowered, and effective management of hardware tasks in the group and the resource utilization ratio of reconfigurable hardware are effectively increased.
Description
Technical field
The invention belongs to field of computer technology, be specifically related to a kind of laying method of subregional two-dimentional reconfigurable hardware task.
Background technology
In restructurable computing system; The restructural resource is abstract usually to be two kinds of models of peacekeeping two dimension; In two-dimentional restructural resource model, the restructural resource is divided with the mode of two-dimensional array and is organized, and the least unit of reconstruct is a rectangular element in the two-dimensional array; A reconfigurable hardware task takies a rectangular area in the array; A plurality of tasks are carried out layout through the mode of scheduling in the restructural resource, because the placement of each task in resource space has two-dimensional freedom, therefore normal employing clear area scanning and these two kinds of methods of state matrix record are that task to be scheduled is searched suitable placement position.The clear area scan method mainly is to come resource is described through the record clear area; Wherein the most important thing is maximum idle rectangle method MER (Maximal Empty Rectangle); The idling-resource rectangle tabulation that one of MER method record does not overlap mutually; Obtain to wait to dispatch the placement location of hardware task through the idling-resource rectangle of searching coupling, this method need safeguard the tabulation of resource rectangle, and it is higher relatively to search the complexity of coupling idling-resource during task scheduling; The state matrix recording method comes management resource through the matrix of safeguarding record restructural computing unit (Reconfigurable Computing Unit) state; But and search the placement location of waiting to dispatch hardware task through the scanning mode matrix; The method need scan and search state matrix, and spended time is longer relatively.What the placement strategy that when the task layout, adopts was commonly used has two kinds; A kind of is first fit laying method (First Fit); Another kind is best laying method (Best Fit); These two kinds of laying methods are not all considered the interior wiring of two-dimensional device difficulty and are waited to dispatch the level of intimate of communicating by letter between hardware task, under the situation that layout is comparatively disperseed, caused huge difficulty for the wiring of two-dimentional reconfigurable device relatively closely in communication between the part hardware task; Even can't lead to by cloth, increased the difficulty of hardware task coordinated management.
Summary of the invention
The object of the present invention is to provide a kind of subregional reconfigurable hardware task laying method, thereby reduce the difficulty of wiring in the reconfigurable device, effectively the high-efficiency management of hardware task in the raising group.
In order to achieve the above object, the technical scheme that the present invention adopted is:
The first step is that the two-dimentional restructural resource model of W * H is the basis with the scale, and W, H are respectively the width and the height of restructural resource model, and the reconfigurable hardware task in the reconfigurable hardware task-set is divided into groups according to the tightness degree of association;
Second step was divided into several resource area with two-dimentional restructural resource model, and resource area is corresponding one by one with the grouping of reconfigurable hardware task;
The 3rd step; For each the reconfigurable hardware task to be placed in the reconfigurable hardware task-set; The corresponding resource area of branch group selection based on reconfigurable hardware task to be placed place; According to the placement strategy of resource area with reconfigurable hardware task layout to be placed to resource area, and record layout position;
In the 4th step, after the reconfigurable hardware task was complete, the reconfigurable hardware task withdrawed from from the resource area at place according to the concluding time successively, deleted the record of placement position according to the placement strategy of reconfigurable hardware task place resource area.
Said two-dimentional restructural resource model is: make up the two-dimensional coordinate system of X, Y axle composition, X, Y axle least unit are a restructural computing unit, based on the reconfigurable hardware task T of two-dimensional coordinate system
iRequired reconstruct resource identifies with rectangle, and rectangle size is described as (w
i, h
i), wherein i is a positive integer, w
iBe reconfigurable hardware task T
iTake the width of reconstruct resource, h
iBe reconfigurable hardware task T
iTake the height of reconstruct resource; Reconfigurable hardware task T
iTo occupy high h of being in the two-dimensional coordinate system
i, wide is w
iThe rectangular area, be called the task put area.
The concrete grammar that divides into groups in the said first step is: with the reconfigurable hardware task T in the reconfigurable hardware task-set
iAccording to the level of intimate of intertask communication, according to based on bunch the responsive partitioning algorithm of level be divided into the m group, G divides into groups to be designated as
j, j is natural number and j≤m.
Said based on bunch the execution in step of the responsive partitioning algorithm of level be:
Step 1); Adopt directed acyclic graph to represent the relation of the reconfigurable hardware task in the reconfigurable hardware task-set; Then in the directed acyclic graph between task the quantity of directed edge just represent the level of intimate of reconfigurable hardware intertask communication; The reconfigurable hardware task that root node is corresponding in the directed acyclic graph is called the root node task, newly-built grouping G
1As current group, first root node task in the directed acyclic graph is added grouping G
1In, generate ready task formation Rdy simultaneously, if when having a plurality of root node task, then all the other root node tasks are added among ready task formation Rdy; When the direct precursor task of a certain task had all been finished by division, then this task was ready task; Desirable number of tasks l=n/m in the calculating group, wherein n is the reconfigurable hardware task sum in the reconfigurable hardware task-set, m is the group number that task is divided into groups, 2≤m≤9;
Step 2); If be the root node task among the ready task formation Rdy, then will have the root node task adding current group of identical follow-up task, and upgrade ready task formation Rdy with root node task in the current group; From ready task formation Rdy, select ready task to add current group then successively; Add the front number if make limit number between other tasks in current group and the reconfigurable hardware task-set be less than or equal to after adding, then this ready task is added, and renewal ready task formation Rdy; Otherwise continue to search, up to having searched all ready tasks;
Step 3); After ready task is searched and is finished, if number of tasks c≤l-m in the current group then adds the task of belonging to task immediate successor in the current group among the ready task queue Rdy in the current group; Upgrade ready task formation Rdy then, return step 2) continue to carry out; If number of tasks c satisfies l-m in the current group<c, the task queue of then writing down current group, newly-built then another grouping G
jAs current group, j≤m, and, upgrade the ready task formation with the affair of the first to be appointed to an office among ready task queue Rdy adding current group, and return step 2) continue to carry out; After last completion of dividing into groups was searched, in the grouping that number of tasks in the ungrouped ready task adding group among the ready task queue Rdy is minimum, the reconfigurable hardware task in the reconfigurable hardware task-set divided into groups to finish;
Step 4), the back output grouping task queue of dividing into groups to finish.
The concrete grammar in said second step is: two-dimentional restructural resource model is divided into m resource area manages, resource area is designated as A
j, j is natural number and j≤m, resource area A
jWith reconfigurable hardware task T
iGrouping G
jCorresponding one by one, can adopt placement strategies such as First Fit, Best Fit or DATS in the resource area, separate between resource area, unified management.
The concrete grammar in said the 3rd step is: during reconfigurable hardware task layout, according to reconfigurable hardware task T
iThe grouping G at place
jSelect corresponding resource area A
j, according to grouping G
jIn the characteristic of reconfigurable hardware task select placement strategy, (if this group task is most for rule and take the less task of resource, then adopt First Fit placement strategy, improve layout speed), according to the placement strategy of selecting with reconfigurable hardware task T
iLayout is to resource area A
j, and record reconfigurable hardware task T
iPlacement position.
Beneficial effect of the present invention is embodied in: subregional reconfigurable hardware task laying method is to the tight situation of intertask communication in the reconfigurable hardware task-set; Utilize the scale information of two-dimentional restructural resource model; Through subregional laying method; With the level of intimate packet partition territory layout management of reconfigurable hardware task according to intertask communication; Reduce the inner wiring difficulty of reconfigurable device, improved the efficiency of management of regional interior reconfigurable hardware task, in the reconfigurable system applicable to real time execution.
Description of drawings
Fig. 1 is a two-dimensional coordinate system synoptic diagram according to the invention, and the unit of horizontal ordinate X and ordinate Y is the reconfigurable cell number;
Fig. 2 is that the grouping of reconfigurable hardware task concerns synoptic diagram in the reconfigurable hardware task-set according to the invention;
Fig. 3 is the schematic layout pattern of reconfigurable hardware task according to the invention.
Embodiment
Below in conjunction with accompanying drawing and embodiment the present invention is done detailed description.
A kind of subregional reconfigurable hardware task laying method may further comprise the steps:
The first step; In scale (W, H are respectively the width and the height of restructural resource model) on the two-dimentional restructural resource model basis of W * H; Make up the two-dimensional coordinate system of X, Y axle composition, X, Y axle least unit are a restructural computing unit, based on the reconfigurable hardware task T of two-dimensional coordinate system
iRequired reconstruct resource identifies with rectangle, and rectangle size is described as: (w
i, h
i), wherein i is a positive integer, w
iBe reconfigurable hardware task T
iTake the width of resource, h
iBe reconfigurable hardware task T
iTake the height of resource; Reconfigurable hardware task T
iTo occupy high h of being in the two-dimensional coordinate system
i, wide is w
iThe rectangular area, be called the task put area, and with the reconfigurable hardware task T in the reconfigurable hardware task-set
iAccording to the level of intimate of intertask communication, according to based on bunch the responsive partitioning algorithm of level be divided into the m group, be designated as G
j(j is natural number and j≤m).
With reference to Fig. 1, A is that scale is the two-dimentional restructural resource model of W * H, and wherein wide W=10, high H=10, B are that restructural calculates minimum unit.Fig. 2 is the grouping graph of a relation of the reconfigurable hardware task in the reconfigurable hardware task-set, adopts directed acyclic graph (DAG) to represent the relation between the reconfigurable hardware task.According to identifying among Fig. 2, the reconfigurable hardware task in the reconfigurable hardware task-set is divided into two groups according to the level of intimate of intertask communication: G
1Group (T
1, T
2, T
3, T
5) and G
2Group (T
4, T
6, T
7, T
8), the directed edge among Fig. 2 between task is represented the communication between task, then the quantity of directed edge is promptly represented the level of intimate of intertask communication between task.Adopt based on bunch concrete the execution as follows of the responsive partitioning algorithm of level:
Step 1); Adopt directed acyclic graph to represent the relation of the reconfigurable hardware task in the reconfigurable hardware task-set; Then in the directed acyclic graph between task the quantity of directed edge just represent the level of intimate of reconfigurable hardware intertask communication; The reconfigurable hardware task that root node is corresponding in the directed acyclic graph is called the root node task, newly-built grouping G
1As current group, first root node task in the directed acyclic graph is added grouping G
1In, generate ready task formation Rdy simultaneously, if when having a plurality of root node task, then all the other root node tasks are added among ready task formation Rdy; When the direct precursor task of a certain task had all been finished by division, then this task was ready task; Desirable number of tasks l=n/m in the calculating group, wherein n is the reconfigurable hardware task sum in the reconfigurable hardware task-set, m is the group number that task is divided into groups, 2≤m≤9, group number m counts the n statistical computation according to hardware task and can know.If the group number then can cause the reconfigurable hardware resource subregion too much too much; Be unfavorable for management, and increased the difficulty of wiring between group, lost the meaning of subregion algorithm; If the group number then can not reach the purpose that algorithm is placed in the subregion very little, therefore need the balanced grouping situation of selecting.Less than 50 hardware task formation, the group number generally should be 2 for number of tasks, under the many situation of number of tasks, should keep hardware task group number can not surpass 9 at most, and value is 3,4,6,8 and 9, mainly is the district management of reconfigurable hardware resource for ease;
Step 2); If be the root node task among the ready task formation Rdy, then will have the root node task adding current group of identical follow-up task, and upgrade ready task formation Rdy with root node task in the current group; From ready task formation Rdy, select ready task to add current group then successively; Add the front number if make limit number between other tasks in current group and the reconfigurable hardware task-set be less than or equal to after adding, then this ready task is added, and renewal ready task formation Rdy; Otherwise continue to search, up to having searched all ready tasks;
Step 3); After ready task is searched and is finished, if number of tasks c≤l-m in the current group then adds the task of belonging to task immediate successor in the current group among the ready task queue Rdy in the current group; Upgrade ready task formation Rdy then, return step 2) continue to carry out; If number of tasks c satisfies l-m in the current group<c, the task queue of then writing down current group, newly-built then another grouping G
jAs current group, j≤m, and, upgrade the ready task formation with the affair of the first to be appointed to an office among ready task queue Rdy adding current group, and return step 2) continue to carry out; After last completion of dividing into groups was searched, in the grouping that number of tasks in the ungrouped ready task adding group among the ready task queue Rdy is minimum, the reconfigurable hardware task in the reconfigurable hardware task-set divided into groups to finish;
Step 4), the back output grouping task queue of dividing into groups to finish.
According to based on bunch the responsive partitioning algorithm of level task shown in Figure 2 is divided: a newly-built G
1Group is with ready task T
1In (root task) adding should be organized, ready task formation this moment Rdy={T
2, T
3And T
4, as ready task T
2Add G
1During group, this group is 3 limit numbers before equaling to add to the limit number of other tasks, therefore with task T
2Add G
1Group, and upgrade ready task formation Rdy={T
3And T
4; As ready task T
3Add G
1During group, this group is 3 limit numbers before equaling to add to the limit number of other tasks, therefore with task T
3Add G
1Group; This moment task T
5Direct precursor task T
2And T
3All divided, then task T
5Just become ready task, upgrade ready task formation Rdy={T
4And T
5; As ready task T
4Add G
1During group, this group is 4 several 3 greater than the limit before adding to the limit number of other tasks, so task T
4Do not add G
1Group continues to search ready task formation Rdy; As ready task T
5Add G
1During group, this group is 2 several 3 less than the limit before adding to the limit number of other tasks, therefore with task T
5Add G
1Group is upgraded ready task formation Rdy={T
4, then all ready tasks have been searched and have been finished, then G
1The task of group is: T
1, T
2, T
3And T
5, and G
1Number of tasks c=4 > in the group; L-m=2, then logger task group G
1, newly-built another G
2Group is as current group, with the first to be appointed to an office affair T of ready task queue
4Add and be somebody's turn to do in the group, because task T
6And T
7Direct precursor task T
4Divided, then upgrading ready task is formation Rdy={T
6And T
7, search ready task according to above method equally, remaining task is divided the G of generation
2The task of group is: T
4, T
6, T
7And T
8, and G
2Number of tasks c=4 > in the group; L-m=2, then logger task group G
2Reconfigurable hardware task division in this reconfigurable hardware task-set finishes.
Second step was divided into m resource area with two-dimentional restructural resource model and manages, and was designated as A
j(group of j≤reconfigurable hardware task is counted m), and corresponding one by one with the group number of hardware task, can adopt identical or different placement strategies such as First Fit, Best Fit, DATS in the zone, interregional separate, unified management.
With reference to figure 3, two-dimentional restructural resource model is divided into up and down two resource area along dotted line as shown in Figure 3 manages, be designated as A
1And A
2, G wherein
1Group hardware task layout is to A
1The zone, G
2Group hardware task layout is to A
2The scale information of hardware task is considered in the zone, and First Fit placement strategy is all used in two zones here, and the mode that adopts diagonal line to place is in opposite directions managed, i.e. A
1The zone is the inside layout of coordinate starting point with the lower left corner, A
2The zone is the inside layout of coordinate starting point with the upper right corner, and this way to manage can effectively reduce the wiring difficulty in the reconfigurable device, also can the clear area be put together simultaneously, conveniently for taking the more placement task of treating of resource enough free spaces is provided.
The 3rd step is during reconfigurable hardware task layout, according to waiting to place reconfigurable hardware task T
iThe group G at place
j, the layout areas A that the selection task is corresponding
j,, select corresponding placement strategy according to the characteristic of this group task; (if this group task majority is rule and takies the less task of resource; Then can adopt First Fit placement strategy, improve layout speed), will wait to place reconfigurable hardware task T according to this placement strategy
iLayout is to regional A
j, and the record layout position.
With reference to Fig. 3, task T
1After the arrival, according to the G at its place
1Group is then with task T
1Layout is to regional A
1, according to the placement strategy First Fit of current region, then task T
1Placement location is [(0,0) (3,2)], T
1The reconstruct resource that occupies is a rectangular area, and width is 3-0=3, highly is 2-0=0; Task T
2, T
3After arriving respectively, layout in the same way, its placement location is respectively [(0,2) (4,5)] and [(4,0) (6,4)]; Task T
4After the arrival, according to the G at its place
2Group is then with task T
4Layout is to regional A
2, according to the placement strategy First Fit of current region, then task T
4Placement location is [(4,8) (10,10)]; Task T
5After the arrival, according to the G at its place
1Group is then with task T
5Layout is to regional A
1, its placement location is [(6,0) (9,3)], T
5The reconstruct resource that occupies is a rectangular area, and width is 9-6=3, highly is 3-0=3; Task T
6, T
7, T
8After arriving respectively, according to the G at task place
2Group then arrives regional A with their layouts
2, then placement location is respectively [(7,5) (10,8)], [(1,6) (4,10)] and [(4,6) (7,8)].
In the 4th step, after the reconfigurable hardware task was complete, task withdrawed from from current region according to the concluding time successively, the placement strategy that adopts according to the current region of task simultaneously, the location records that deletion is corresponding.
With reference to Fig. 3, as task T
1After carrying out end, according to current region A
1The placement strategy First Fit that is adopted withdraws from current region, deletes current position information simultaneously, and the modification task status is complete; Other tasks withdraw from according to carrying out end successively in proper order, and the deletion current position information, the updating task state.
Claims (6)
1. subregional reconfigurable hardware task laying method is characterized in that: may further comprise the steps:
The first step is that the two-dimentional restructural resource model of W * H is the basis with the scale, and W, H are respectively the width and the height of restructural resource model, and the reconfigurable hardware task in the reconfigurable hardware task-set is divided into groups according to the tightness degree of association;
Second step was divided into several resource area with two-dimentional restructural resource model, and resource area is corresponding one by one with the grouping of reconfigurable hardware task;
The 3rd step; For each the reconfigurable hardware task to be placed in the reconfigurable hardware task-set; The corresponding resource area of branch group selection based on reconfigurable hardware task to be placed place; According to the placement strategy of resource area with reconfigurable hardware task layout to be placed to resource area, and record layout position;
In the 4th step, after the reconfigurable hardware task was complete, the reconfigurable hardware task withdrawed from from the resource area at place according to the concluding time successively, and deleted the record of placement position according to the placement strategy of reconfigurable hardware task place resource area.
2. according to the said a kind of subregional reconfigurable hardware task laying method of claim 1; It is characterized in that: said two-dimentional restructural resource model is: the two-dimensional coordinate system that makes up X, Y axle composition; X, Y axle least unit are a restructural computing unit, based on the reconfigurable hardware task T of two-dimensional coordinate system
iRequired reconstruct resource identifies with rectangle, and rectangle size is described as (w
i, h
i), wherein i is a positive integer, w
iBe reconfigurable hardware task T
iTake the width of reconstruct resource, h
iBe reconfigurable hardware task T
iTake the height of reconstruct resource; Reconfigurable hardware task T
iTo occupy high h of being in the two-dimensional coordinate system
i, wide is w
iThe rectangular area.
3. according to the said a kind of subregional reconfigurable hardware task laying method of claim 1, it is characterized in that: the concrete grammar that divides into groups in the said first step is: with the reconfigurable hardware task T in the reconfigurable hardware task-set
iAccording to the level of intimate of intertask communication, according to based on bunch the responsive partitioning algorithm of level be divided into the m group, G divides into groups to be designated as
j, j is natural number and j≤m.
4. according to the said a kind of subregional reconfigurable hardware task laying method of claim 3, it is characterized in that: said based on bunch the execution in step of the responsive partitioning algorithm of level be:
Step 1); Adopt directed acyclic graph to represent the relation of the reconfigurable hardware task in the reconfigurable hardware task-set; Then in the directed acyclic graph between task the quantity of directed edge just represent the level of intimate of reconfigurable hardware intertask communication; The reconfigurable hardware task that root node is corresponding in the directed acyclic graph is called the root node task, newly-built grouping G
1As current group, first root node task in the directed acyclic graph is added grouping G
1In, generate ready task formation Rdy simultaneously, if when having a plurality of root node task, then all the other root node tasks are added among ready task formation Rdy; When the direct precursor task of a certain task had all been finished by division, then this task was ready task; Desirable number of tasks l=n/m in the calculating group, wherein n is the reconfigurable hardware task sum in the reconfigurable hardware task-set, m is the group number that task is divided into groups, 2≤m≤9;
Step 2); If be the root node task among the ready task formation Rdy, then will have the root node task adding current group of identical follow-up task, and upgrade ready task formation Rdy with root node task in the current group; From ready task formation Rdy, select ready task to add current group then successively; Add the front number if make limit number between other tasks in current group and the reconfigurable hardware task-set be less than or equal to after adding, then this ready task is added, and renewal ready task formation Rdy; Otherwise continue to search, up to having searched all ready tasks;
Step 3); After ready task is searched and is finished, if number of tasks c≤l-m in the current group then adds the task of belonging to task immediate successor in the current group among the ready task queue Rdy in the current group; Upgrade ready task formation Rdy then, return step 2) continue to carry out; If number of tasks c satisfies l-m in the current group<c, the task queue of then writing down current group, newly-built then another grouping G
jAs current group, j≤m, and, upgrade the ready task formation with the affair of the first to be appointed to an office among ready task queue Rdy adding current group, and return step 2) continue to carry out; After last completion of dividing into groups was searched, in the grouping that number of tasks in the ungrouped ready task adding group among the ready task queue Rdy is minimum, the reconfigurable hardware task in the reconfigurable hardware task-set divided into groups to finish;
Step 4), the back output grouping task queue of dividing into groups to finish.
5. according to the said a kind of subregional reconfigurable hardware task laying method of claim 1, it is characterized in that: the concrete grammar in said second step is: two-dimentional restructural resource model is divided into m resource area, and resource area is designated as A
j, j is natural number and j≤m, resource area A
jWith reconfigurable hardware task T
iGrouping G
jCorresponding one by one.
6. according to the said a kind of subregional reconfigurable hardware task laying method of claim 1, it is characterized in that: the concrete grammar in said the 3rd step is: during reconfigurable hardware task layout, according to reconfigurable hardware task T
iThe grouping G at place
jSelect corresponding resource area A
j, according to grouping G
jIn the characteristic of reconfigurable hardware task select placement strategy, according to the placement strategy of selecting with reconfigurable hardware task T
iLayout is to resource area A
j, and record reconfigurable hardware task T
iPlacement position.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210139467.5A CN102681901B (en) | 2012-05-08 | 2012-05-08 | Segmental reconfigurable hardware task arranging method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210139467.5A CN102681901B (en) | 2012-05-08 | 2012-05-08 | Segmental reconfigurable hardware task arranging method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102681901A true CN102681901A (en) | 2012-09-19 |
CN102681901B CN102681901B (en) | 2014-04-23 |
Family
ID=46813869
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201210139467.5A Expired - Fee Related CN102681901B (en) | 2012-05-08 | 2012-05-08 | Segmental reconfigurable hardware task arranging method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102681901B (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103631659A (en) * | 2013-12-16 | 2014-03-12 | 武汉科技大学 | Schedule optimization method for communication energy consumption in on-chip network |
CN104834571A (en) * | 2015-05-25 | 2015-08-12 | 南京大学 | Data prefetching method applied to cloud workflow scheduling |
CN105528243A (en) * | 2015-07-02 | 2016-04-27 | 中国科学院计算技术研究所 | A priority packet scheduling method and system utilizing data topological information |
CN105740518A (en) * | 2016-01-25 | 2016-07-06 | 深圳市同创国芯电子有限公司 | FPGA resource placement method and apparatus |
CN106991009A (en) * | 2017-02-28 | 2017-07-28 | 武汉科技大学 | A kind of reconfigurable device chip area management optimization method |
CN107256281A (en) * | 2017-05-05 | 2017-10-17 | 西安交通大学 | A kind of non-rectangle layout method of FPGA restructural resources based on cutting-out method |
US10452605B2 (en) | 2015-03-27 | 2019-10-22 | Huawei Technologies Co., Ltd. | Method and apparatus for task scheduling on heterogeneous multi-core reconfigurable computing platform |
US10552222B2 (en) | 2015-01-16 | 2020-02-04 | Huawei Technologies Co., Ltd. | Task scheduling method and apparatus on heterogeneous multi-core reconfigurable computing platform |
CN111880933A (en) * | 2020-07-27 | 2020-11-03 | 北京神舟航天软件技术有限公司 | Reconfigurable hardware task dynamic allocation method based on heterogeneous computing platform |
US11061882B2 (en) | 2018-06-25 | 2021-07-13 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for generating a wide table |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101833368A (en) * | 2010-04-13 | 2010-09-15 | 杭州电子科技大学 | Method for managing energy of local dynamic reconfigurable system by coordinatively processing software and hardware |
CN102262612A (en) * | 2011-07-28 | 2011-11-30 | 西安交通大学 | Method for lowering restructurable hardware task layout conflict |
-
2012
- 2012-05-08 CN CN201210139467.5A patent/CN102681901B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101833368A (en) * | 2010-04-13 | 2010-09-15 | 杭州电子科技大学 | Method for managing energy of local dynamic reconfigurable system by coordinatively processing software and hardware |
CN102262612A (en) * | 2011-07-28 | 2011-11-30 | 西安交通大学 | Method for lowering restructurable hardware task layout conflict |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103631659B (en) * | 2013-12-16 | 2017-02-15 | 武汉科技大学 | Schedule optimization method for communication energy consumption in on-chip network |
CN103631659A (en) * | 2013-12-16 | 2014-03-12 | 武汉科技大学 | Schedule optimization method for communication energy consumption in on-chip network |
US10552222B2 (en) | 2015-01-16 | 2020-02-04 | Huawei Technologies Co., Ltd. | Task scheduling method and apparatus on heterogeneous multi-core reconfigurable computing platform |
US10452605B2 (en) | 2015-03-27 | 2019-10-22 | Huawei Technologies Co., Ltd. | Method and apparatus for task scheduling on heterogeneous multi-core reconfigurable computing platform |
CN104834571B (en) * | 2015-05-25 | 2018-05-25 | 南京大学 | A kind of data prefetching method applied to cloud workflow schedule |
CN104834571A (en) * | 2015-05-25 | 2015-08-12 | 南京大学 | Data prefetching method applied to cloud workflow scheduling |
CN105528243B (en) * | 2015-07-02 | 2019-01-11 | 中国科学院计算技术研究所 | A kind of priority packet dispatching method and system using data topology information |
CN105528243A (en) * | 2015-07-02 | 2016-04-27 | 中国科学院计算技术研究所 | A priority packet scheduling method and system utilizing data topological information |
CN105740518A (en) * | 2016-01-25 | 2016-07-06 | 深圳市同创国芯电子有限公司 | FPGA resource placement method and apparatus |
CN106991009A (en) * | 2017-02-28 | 2017-07-28 | 武汉科技大学 | A kind of reconfigurable device chip area management optimization method |
CN106991009B (en) * | 2017-02-28 | 2020-08-11 | 武汉科技大学 | Reconfigurable device on-chip area management optimization method |
CN107256281A (en) * | 2017-05-05 | 2017-10-17 | 西安交通大学 | A kind of non-rectangle layout method of FPGA restructural resources based on cutting-out method |
CN107256281B (en) * | 2017-05-05 | 2020-07-28 | 西安交通大学 | FPGA (field programmable Gate array) reconfigurable resource non-rectangular layout method based on cutting method |
US11061882B2 (en) | 2018-06-25 | 2021-07-13 | Beijing Didi Infinity Technology And Development Co., Ltd. | Systems and methods for generating a wide table |
CN111880933A (en) * | 2020-07-27 | 2020-11-03 | 北京神舟航天软件技术有限公司 | Reconfigurable hardware task dynamic allocation method based on heterogeneous computing platform |
CN111880933B (en) * | 2020-07-27 | 2023-09-22 | 北京神舟航天软件技术有限公司 | Reconfigurable hardware task dynamic allocation method based on heterogeneous computing platform |
Also Published As
Publication number | Publication date |
---|---|
CN102681901B (en) | 2014-04-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102681901B (en) | Segmental reconfigurable hardware task arranging method | |
CN103370691B (en) | Managing buffer overflow conditions | |
Verma et al. | Orchestrating an ensemble of MapReduce jobs for minimizing their makespan | |
CN101571810B (en) | Method for implementing program, method for verifying program result, devices and system | |
CN105069134A (en) | Method for automatically collecting Oracle statistical information | |
CN102937918B (en) | A kind of HDFS runtime data block balance method | |
CN104331749A (en) | AGV optimization scheduling method based on simulated annealing particle swarm | |
CN107291539B (en) | Cluster program scheduler method based on resource significance level | |
CN103246549B (en) | A kind of method and system of data conversion storage | |
CN103701920A (en) | Method for configuring virtual application server under cloud environment | |
CN103765381A (en) | Parallel operation on B+ trees | |
CN103412889A (en) | Data storage and query method of intelligent electric meter and system thereof | |
CN102262612B (en) | Method for lowering restructurable hardware task layout conflict | |
CN105912601A (en) | Partition storage method for distributed real-time memory database of energy management system | |
CN105991478A (en) | Server resource distribution method and system | |
CN103019859A (en) | Method and system for calling service request | |
CN104778077A (en) | High-speed extranuclear graph processing method and system based on random and continuous disk access | |
CN110058882A (en) | It is a kind of for CNN accelerate OPU instruction set define method | |
CN106648546A (en) | Collaborative optimization compilation method used for GPU register allocation and parallelism management | |
KR20100004605A (en) | Method for selecting node in network system and system thereof | |
CN104459781A (en) | Three-dimensional pre-stack seismic data random noise degeneration method | |
KR20170016168A (en) | Tile-based map data updating system and method thereof | |
CN101853178B (en) | Description method of reconfigurable hardware resource in scheduling | |
CN104239520B (en) | A kind of HDFS data block Placement Strategies based on historical information | |
CN101515284A (en) | Parallel space topology analyzing method based on discrete grid |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | 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 | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20140423 Termination date: 20180508 |