CN101799838B - Method for seeking mapping scheme between tasks and nodes of network on chip - Google Patents
Method for seeking mapping scheme between tasks and nodes of network on chip Download PDFInfo
- Publication number
- CN101799838B CN101799838B CN2010100281577A CN201010028157A CN101799838B CN 101799838 B CN101799838 B CN 101799838B CN 2010100281577 A CN2010100281577 A CN 2010100281577A CN 201010028157 A CN201010028157 A CN 201010028157A CN 101799838 B CN101799838 B CN 101799838B
- Authority
- CN
- China
- Prior art keywords
- scheme
- group
- energy consumption
- response time
- network
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a method for seeking a mapping scheme between tasks and nodes of a network on chip, which comprises the following steps: initializing a scheme group Jt comprising K allocation schemes ( K indicates number); determining network on chip energy consumption, response time and a fitness value in each scheme in the scheme group Jt; according to the size of the fitness value and the like, prioritizing, grouping and exchanging and carrying out other processes to obtain the preferred scheme finally. The energy consumption and the response time are separated to be used as preferred indexes of an evaluation scheme, and the energy consumption and the response time are simultaneously considered in the process of seeking the optimal scheme, so that the sought scheme can minimize the energy consumption and system delay, and thereby, the performance of the network designed by the preferred mapping scheme between the tasks and the nodes is the best, and the method has the characteristics of complexity and simple implementation compared with other methods for seeking preferred scheme of mapping between the tasks and the nodes in a plurality object functions. The simulation confirms that the method of the invention has quick convergence rate.
Description
Affiliated field
This patent belongs to the technical field of integrated circuit (IC) design, especially at the search procedure of the mapping scheme of the multitask of network-on-chip and multinode.
Technical background
Increase along with the computation complexity of future communications terminal and equipment, the integrated scale demand of complication system chip will increase rapidly in real time, may integrated dozens of or hundreds of processing unit on the chip piece, become the development trend of software radio based on the signal processing platform of multinuclear.In such synthesization system, reliable design, at a high speed, communication system becomes SOC (system on a chip) (System on Chip, SoC) Fa Zhan challenge and opportunity in the high performance chips of low-power consumption.The bus type communication structure of tradition multinuclear SoC can't satisfy the demand of exchanges data in the extensive complicated digital signal processing procedure in real time, has therefore proposed interconnected and communication mechanism---network-on-chip on the sheet of new generation.
The system that network-on-chip (going up each node processor just) is realized mainly is complex communication system or the digital information processing system that real-time is had relatively high expectations.Communication has real-time and processing delay requirement with signal Processing, and therefore, the mapping scheme of NoC don't fail to make that the time-delay of system is as far as possible little.In addition, network-on-chip is to realize on the chip of integrated circuit, and for the work that guarantees that chip can be stable, the temperature of chip can not be high.The temperature of chip is that energy consumption is big more by the decision of the energy consumption of chip, and the temperature of chip is high more.Therefore, when selecting mapping scheme, must select to make energy consumption to hang down and time delay few scheme of trying one's best as far as possible.
By Cha Xin with extensively collect documents and materials, we find disclosedly to be used to seek on the network-on-chip that the method for mapping scheme has following a few class between task and node:
Document " Zhou Ganmin; Yin Yongsheng; Hu Yonghua etc. based on the NoC mapping of ant colony optimization algorithm. computer engineering and application .2005; 41 (18): 7-10. " adopt a kind of ant colony optimization algorithm effectively to realize the network-on-chip mapping, when duty mapping is arrived the network-on-chip node, reduced the communication power consumption of system as much as possible.The objective function of this method is a power consumption, and this method is roughly as follows:
(1) number according to ant is that each unappropriated node is according to probability assignments task (this probability is the probability that a node is assigned to certain task that calculates by a kind of mode) circularly, every circulation primary all will make all nodes all be assigned to corresponding task, adopt two exchange processes to carry out Local Search then, find out the optimum solution of power consumption minimum, and upgrade the pheromones of the ant of optimum solution correspondence with the value of the power consumption number pheromones dough softening of optimum solution and current information element, by said process repeatedly after the iteration, is selected the allocative decision that makes the power consumption minimum at last.
The mapping scheme that this method obtains has less communication power consumption, but does not have the delay performance of taking into account system.
Document " Yang Shengguang; Li Li; brilliant human relations etc. towards the NoC mapping method of energy consumption and time-delay. electronic letters, vol .2008; 36 (5): 937-942 " based on two-dimensional grid structure NoC platform, set up the unified goal function that is intended to optimization system communication energy consumption and link load variance based on ant group algorithm.The calculating formula of the objective function cost of this method is:
cost＝λ×E+(1-λ)×VAR
Wherein, λ is a scale-up factor, and E is an energy consumption, and VAR is the link load variance.This method is roughly as follows:
(1) number according to ant is that each unappropriated node is according to probability assignments task (probability is the probability that a node is assigned to certain task that calculates by a kind of mode) circularly, every circulation primary all will make all nodes all be assigned to corresponding task, adopt two exchange processes to carry out Local Search then, find out the power consumption minimal solution, and upgrade the pheromones of the ant of optimum solution correspondence with the value of the power consumption number pheromones dough softening of optimum solution and current information element, by said process repeatedly after the iteration, is selected the allocative decision that makes objective function cost minimum at last.
This method do not consider to communicate by letter energy consumption and execution time is diverse physical quantity, and the mapping scheme of acquisition not necessarily has excellent performance.
Document Tang lei, Shashi Kumar.A two-step genetic algorithm for mapping task graphs to a network onchip architecture.Proceedings of the Euromicro Symposium on Digital System Design 2003.Antalya, Turkey:IEEE, 2003:180-187. propose a kind of method that adopts two step genetic algorithms to realize the duty mapping problem.This method is roughly as follows:
(1) time-delay of at first supposing all limits in the task flow graph equates, is a constant, i.e. average; Be each Task Distribution functional unit (as processor, storer etc.), make that total time-delay of system is minimum
(2) use the limit of real data flow diagram to delay time, further optimize, functional unit is assigned on the node of NoC.
This method has been divided into two steps with the whole optimizing phase, thereby has reduced the complexity of calculating, and has shortened computing time, but the energy consumption of the mapping scheme that obtains and time-delay can be not minimum simultaneously.
Document Wenbiao Zhou, Yan Zhang, Zhigang Mao.Pareto based Multi-objective Mapping IP Coresonto NoC Architectures.IEEE Asia Pacific Conference on Circuits and Systems.2006:331-334. and Giuseppe Ascia, Vincenzo Catania, Maurizio Palesi.Multi-objective Mapping for Mesh-based NoCArchitectures.International Conference on Hardware/Software Dodesign and System Synthesis, 2004:182-187. propose to adopt the multi-target evolution algorithm that task on the network-on-chip is mapped on the node based on the network-on-chip structure of Mesh topology.The objective function of this method has two, and an objective function is an average data bag jumping figure (hop), and calculating formula is as follows:
Another objective function is hot(test)-spot temperature (T), calculating formula is: and T=max (R * P), wherein, R is the transferred-impedance matrix, utilize K.Skadron, people such as M.R.Stan and W.Huang are published in the heat modeling tool HotSpot that proposes in the concentrated article " Temperature-AwareMicro-architecture " of ISCA ' 03 meeting paper and obtain.This method is roughly as follows:
(1) selects some forecast scheme configuration initialization scheme groups at first at random, being published in method that " the Multi-objective optimization function optimization using non-dominatedsorting genetic algorithms " of EvolutionaryComputation1995 second volume propose according to N.Srinivas and K.Deb carries out the duplicating of genetic algorithm of non-domination, intersection and mutation operation and generates new scheme group, repeat then above-mentionedly to duplicate, intersection and mutation operation for several times, select the scheme of objective function minimum of making.
The non-domination sort operation complexity that adopts in this method performance higher, that separate is not outstanding especially.
Summary of the invention
This patent be the network-on-chip of multitask multinode propose a kind of searching make work energy consumption and response time simultaneously hour task and node between the method for mapping scheme, the characteristics of this method are: the mapping scheme performance height that obtains, fast convergence rate, complexity is low.
For convenience of description, we are with vectorial X=(x
1, x
2... x
n..., x
N) N task of expression network-on-chip a kind of scheme of shining upon to M node, the individual locational x of the n of vectorial X
n(0≤x
n≤ M-1) represented the node serial number of n duty mapping to certain node, i.e. x
nRepresented that the Task Distribution that will be numbered n given x
nNode.
The step of seeking minimum scheme of energy consumption and response time while is as follows:
1, generates a scheme group J who comprises K allocative decision at random
t, and be each allocative decision X=(x
1, x
2..., x
N) set the Probability p of an intersection
iProbability q with variation
i, and total iterations T, and the variable t=0 of order record number of iterations;
2, obtain scheme group J
tIn network-on-chip energy consumption and response time under every kind of scheme;
3, one of methods such as non-domination ranking method, exclusive method, banker method, challenge tournament method, recurrence method, quick sort or improved quick sort of employing Deb (Science Press published in 2007, and the author is for introducing in Zheng Jinhua " multi-target evolution algorithm and application thereof ") calculate scheme group J
tIn the fitness value of each scheme, the priority of this fitness value size reflection scheme, it is high more be worth more little priority, so-called priority level be little by energy consumption and response time be excellent.
4, with scheme group J
tIn K scheme be divided at random
Individual scheme group, wherein L is the scheme number in the every scheme group, selects in the every scheme group prepreerence scheme according to the precedence relationship of determining in the 3rd step and forms a priority scheme group Q
t
5, with priority scheme group Q
tIn scheme press random pair, described pairing is meant according to two schemes divides into groups for one group, and with the Probability p of the intersection set in the step 1
iThe numbering of the node on any one same position on two schemes in every pair of scheme is exchanged, and then all are merged by every pair of scheme after exchanging obtain scheme group Q
t';
6, the variation probability q to set in the step 1
iChange scheme group Q
t' in each scheme in the numbering of any one locational node, obtain scheme group G
t
7, the fitness value order of calculating by the 3rd step from small to large is from scheme group J
tThe middle selection
Individual scheme is with scheme group G
tScheme combine, constitute the new scheme group G comprise K scheme
t'
8, with scheme group J
tAnd G
tBe merged into scheme group R
t
9, numerical procedure group R
tIn network-on-chip energy consumption and response time under every kind of scheme;
10, the boundary set that adopts Deb to introduce in the article A Fast Elitist Non-dominated Sorting Genetic Algorithm for Multi-objective Optimization:NSGA-II that was published on magazine IEEE Transactions on Evolutionary Computation the 6th volume in 2002 produces method with scheme group R
tIn 2K scheme be divided into a plurality of groups, each group represents a boundary set, to little group #, the scheme in the less group of numbering is than the priority height of the scheme of the bigger group of numbering; So-called priority level is little of excellent by energy consumption and response time.
11, new scheme group F who does not have scheme of initialization
t, to divide the good group from the 10th step then, the numbering order from big to small by group joins F with the scheme in the group successively
tIn, up to scheme group F
tIn the scheme number surpass K, the scheme of the group that will add is at last all taken out then, and the group that will be added at last is designated as V;
12, determine in the 10th step the priority ranking of each scheme in each group, concrete grammar is as follows:
12-1 makes that the distance value of each scheme in the group is 0, and described distance value represents whether getting in touch of this scheme and other scheme be tight.
12-2 carries out inverted order according to the 9th size that goes on foot the energy consumption of each scheme that calculates to scheme and arranges, and the less scheme of energy consumption comes the front of the bigger scheme of energy consumption,
12-3 calculates the energy consumption distance of each scheme, and the energy consumption of each scheme distance deducts the resulting value of energy consumption of that scheme that comes its front for the energy consumption of that scheme of coming its back,
12-4 carries out the inverted order arrangement according to the response time size of each scheme that the 9th step calculated to scheme, and the response time, less scheme came the front of bigger scheme of response time,
12-5 calculates the response time distance of each scheme, and response time of each scheme distance deducts the resulting value of response time of that scheme that comes its front for response time of that scheme of coming its back,
12-6 obtains the distance value of scheme with the response time distance of each scheme and energy consumption apart from addition,
12-7 comes the foremost to the scheme of the scheme of energy consumption minimum and response time minimum, according to the distance size remaining scheme is sorted then, and come the back of the scheme of the scheme of energy consumption minimum and response time minimum, the scheme that distance value is bigger comes the front of the less scheme of distance value, the priority ratio that comes the scheme of front like this comes the priority height of the scheme of back
13, join scheme group F from the select progressively scheme of having passed through in the V of group after the ordering of the 12nd step according to priority from high to low
t, up to scheme group F
tIn the scheme number be K.
14, with scheme group F
tIn K scheme be divided at random
Individual scheme group, wherein L is the scheme number in the every scheme group, unites definite precedence relationship according to the 10th step and the 12nd step and selects priority scheme group F of prepreerence scheme composition in the every scheme group
t';
15, with priority scheme group F
t' in scheme random pair again, with the Probability p of the intersection set in the step 1
iThe numbering of the node on any one same position on two schemes in every pair of scheme is exchanged, and then all are merged by every pair of scheme after exchanging obtain scheme group H
t
16, with the probability q of the variation set in the step 1
iChange scheme group H
tIn in each scheme in the numbering of any one locational node, obtain scheme group H
t'
17, unite definite scheme group F according to the 10th step and the 12nd step
tThe priority ranking of middle scheme is according to priority order from high to low, from scheme group F
tThe middle selection
Individual scheme is with scheme group H
t' scheme combine, constitute the new scheme group J comprise K scheme
T+1
If 18 t＜T, then t=t+1 turns back to step 2 then, otherwise enters step 19
19, calculate scheme group J according to the described method of step 3
TIn the fitness value of each scheme, with the scheme of fitness value minimum, promptly minimum simultaneously scheme of energy consumption and response time as the task of network-on-chip to internodal mapping scheme.
Energy consumption of the present invention can for:
E＝E
PE+E
Net
Wherein, E is an energy consumption, E
PEBe the energy consumption sum of all processing nodes on the network-on-chip, E
NetFor connecting the energy consumption of the communication network of each processing node on the network-on-chip.The energy consumption E of communication network
NetAvailable Terry Tao Ye; Luca Benini, Giovanni DeMicheli are published in the proceeding in 02. year of DAC ' " Analysis of Power Consumption on Switch Fabrics inNetwork Routers " Calculation Method of Energy Consumption that proposes in the literary composition or J.Hu and R.Marculescu be published in the method that proposes in the article " Energy-andperformance-aware mapping for regular NoC architectures " of the 24th volume in 2005 of IEEE Transaction onComputer-Aided Design of Integrated Circuits and Systems and calculate.
The calculating formula of response time of the present invention (data are from being input to the longest time-delay of output experience) can for:
D＝P+TR+TW
Wherein, D is the response time, and P is the processing time sum that is input to all nodes on the outgoing route, and TR is the time-delay sum that is input to all links on the outgoing route, and TW is queuing and the stand-by period sum that is input to packet on the outgoing route.
Essence of an invention and effect
The present invention by with energy consumption and response time separately as the preferential index of the scheme of examination, but in seeking the optimal case process, consider energy consumption and response time again simultaneously, having accomplished to seek the scheme of coming out makes energy consumption and system delay minimum simultaneously, adopt the performance of the network that this conceptual design that preferably comes out comes out to reach best thereby make, and and other to the method for the priority scheme that shines upon between searching task and node under a plurality of objective functions relatively, has complexity, realize characteristic of simple, can also draw method fast convergence rate of the present invention by emulation.
Description of drawings
The task image that Fig. 1 is made up of 6 tasks.
The network-on-chip that Fig. 2 is made up of 16 nodes.
The 1st, task, the 2nd, the portfolio between the task, the 3rd, the node of network-on-chip, the 4th, the router of network-on-chip.
Embodiment
6 tasks are arranged in the task image shown in Fig. 1, and the network-on-chip shown in Fig. 2 has 16 nodes.Present embodiment be exactly for seek a kind of with 6 Task Distribution shown in Figure 1 to 16 nodes of network-on-chip shown in Figure 2, make the mapping scheme that the energy consumption of network-on-chip after the mapping and response time reach minimum value simultaneously.For convenience of description, we are with vectorial X=(x
1, x
2, x
3, x
4, x
5, x
6) 6 tasks of expression network-on-chip are to a kind of scheme of 16 nodes mapping, the n of vectorial X locational x
n(0≤x
n≤ 15) node serial number of n duty mapping to certain node, i.e. x have been represented
nRepresented that the Task Distribution that will be numbered n given x
nNode.
The step of seeking minimum scheme of energy consumption and response time while is as follows:
1, generates a scheme group J who comprises 20 allocative decisions at random
t, each allocative decision vector representation in the scheme group is: (11,1,15,9,0,15), (4,1,0,8,1,6), (7,5,15,8,7,8), (10,11,5,0,6,8), (0,6,7,6,15,9), (14,11,11,0,8,5), (15,5,14,0,15,1), (3,0,8,7,1,9), (2,2,1,8,2,8), (3,13,8,15,5,8), (8,8,9,12,10,2), (14,0,3,2,15,5), (5,2,4,9,0,8), (7,2,15,11,1,11), (7,1,15,2,1,11), (9,12,3,9,10,11), (11,3,12,14,14,6), (11,12,4,7,1,9), (4,8,12,12,9,4), (1,15,6,9,3,9).The probability of setting the intersection of each scheme is 0.4, and the probability of variation is 0.01, and total iterations is 100, and the variable t=0 of record number of iterations;
2, obtain scheme group J
tIn the energy consumption and response time of network-on-chip under every kind of scheme:
The computing formula of 2-1, energy consumption (E) adopts: E=E
PE+ E
Net, E
PEBe the energy consumption sum of all processing nodes on the network-on-chip, E
NetFor connecting the energy consumption of the communication network of each processing node on the network-on-chip.The energy consumption of all processing nodes equals all task handling times in the task image on the network, and each task handling time can obtain by the operand estimation to each task, when promptly setting the tasks figure, and the energy consumption that sets the tasks.In this example, for various scheme, E
PEThe value perseverance be 1.18.Adopt Terry TaoYe, Luca Benini, Giovanni De Micheli are published in the energy consumption E that the energy consumption calculation method that proposes in " Analysis of PowerConsumption on Switch Fabrics in Network Routers " literary composition in the proceeding in 02. year of DAC ' calculates communication network under the various schemes
Net, be respectively: 54.69,38.09,61.69,45.84,43.84,46.39,75.19,41.84,46.99,48.59,33.09,55.24,38.14,43.89,50.09,40.94,44.64,43.89,29.44,53.74.
The calculating formula of 2-2, response time (D) adopts: D=P+TR+TW, P is the processing time sum that is input to all nodes on the outgoing route, TR is the time-delay sum that is input to all links on the outgoing route, and TW is queuing and the stand-by period sum that is input to packet on the outgoing route.In this example, the response time of each scheme is respectively: 0.16,0.1601,0.1601,0.1601,0.1601,0.1601,0.1602,0.1601,0.1601,0.1601,0.1601,0.1601,0.16,0.16,0.1601,0.1601,0.1601,0.1601,0.16,0.1601.
3, adopt the non-domination ranking method (Science Press published in 2007, and the author is for introducing in Zheng Jinhua " multi-target evolution algorithm and application thereof ") of Deb to calculate scheme group J
tIn the fitness value of each scheme, for: 0.7895,1.6316,0.1579,1,1.1579,0.1579,0,1.3684,0.4737,0.4737,1.6316,0.4737,1.8947,1.6316,0.4737,1.6316,1,1.2632,2,0.7895 this is worth the priority of big or small reflection scheme, it is high more to be worth more little priority.
4, with scheme group J
tIn 20 schemes be divided into 10 scheme groups at random, in the every scheme group 2 schemes are arranged, select the prepreerence scheme of every scheme group according to the precedence relationship of determining in the 3rd step, will form a new scheme group Q from the prepreerence scheme that all scheme groups select then
t, new scheme group Q
tIn scheme be: (4,1,0,8,1,6), (10,11,5,0,6,8), (0,6,7,6,15,9), (3,0,8,7,1,9), (2,2,1,8,2,8), (8,8,9,12,10,2), (5,2,4,9,0,8), (9,12,3,9,10,11), (11,12,4,7,1,9), (4,8,12,12,9,3)
5, with scheme group Q
tIn scheme press random pair, described pairing is meant according to two schemes divides into groups for one group, and exchange, and then all are merged by every pair of scheme after exchanging obtain scheme group Q with the numbering of the node on random site between 0.4 pair of every pair of scheme of probability of the intersection set in the step 1
t', scheme group Q
t' in scheme be: (10,1,0,8,1,6), (4,11,5,0,6,8), (0,6,7,6,15,9), (3,0,8,7,1,9), (2,2,1,8,2,8), (8,8,9,12,10,2), (5,2,4,9,0,8), (9,12,3,9,10,11), (11,12,4,7,1,9), (4,8,12,12,9,3)
6, change scheme group Q with the variation probability of setting in the step 1 0.01
t' in the numbering of the node on random site in each scheme, obtain scheme group G
t, scheme group G
tIn scheme be: (10,1,0,8,1,6), (4,11,5,0,6,8), (0,6,7,6,15,9), (3,0,8,7,1,9), (2,2,1,8,2,8), (8,8,9,12,10,2), (5,2,4,9,0,8), (9,12,3,9,10,11), (11,12,4,7,1,9), (4,8,12,12,9,3)
7, press fitness order from small to large, from scheme group J
t10 schemes of middle selection are with scheme group G
tScheme combine, constitute the new scheme group G comprise 20 schemes
t'
8, with scheme group J
tAnd G
t' be merged into the scheme group R that comprises 40 schemes
t, that is: (11,1,15,9,0,15), (4,1,0,8,1,6), (7,5,15,8,7,8), (10,11,5,0,6,8), (0,6,7,6,15,9), (14,11,11,0,8,5), (15,5,14,0,15,1), (3,0,8,7,1,9), (2,2,1,8,2,8), (3,1,3,8,15,5,8), (8,8,9,12,10,2), (14,0,3,2,15,5), (5,2,4,9,0,8), (7,2,15,11,1,11), (7,1,15,2,1,11), (9,12,3,9,10,11), (11,3,12,14,14,6), (11,12,4,7,1,9), (4,8,12,12,9,4), (1,15,6,9,3,9), (4,1,0,8,1,6), (10,11,5,0,6,8), (0,6,7,6,15,9), (3,0,8,7,1,9), (2,2,1,8,2,8), (8,8,9,12,10,2), (5,2,4,9,0,8), (9,12,3,9,10,11), (11,12,4,7,1,9), (4,8,12,12,9,3), (11,3,12,14,14,6), (0,6,7,6,15,9), (11,12,4,7,1,9), (3,0,8,7,1,9), (4,1,0,8,1,6), (8,8,9,12,10,2), (7,2,15,11,1,11), (9,12,3,9,10,11), (5,2,4,9,0,8), (4,8,12,12,9,4)
9, numerical procedure group R
tIn network-on-chip energy consumption and response time under every kind of scheme, the energy consumption of each scheme is respectively: 54.69,38.09,61.69,45.84,43.84,46.39,75.19,41.84,46.99,48.59,33.09,55.24,38.14,43.89,50.09,40.94,44.64,43.89,29.44,53.74,38.09,45.84,43.84,41.84,46.99,33.09,38.14,40.94,43.89,40.84,44.64,43.84,43.89,41.84,38.09,33.09,43.89,40.94,38.14,29.44; The response time of each scheme is respectively:
10、0.16，0.1601，0.1601，0.1601，0.1601，0.1601，0.1602，0.1601，0.1601，0.1601，0.1601，0.1601，0.16，0.16，0.1601，0.1601，0.1601，0.1601，0.16，0.1601，0.1601，0.1601，0.1601，0.1601，0.1601，0.1601，0.16，0.1601，0.1601，0.1601，0.1601，0.1601，0.1601，0.1601，0.1601，0.1601，0.16，0.1601，0.16，0.16。
11, the boundary set that adopts Deb to introduce in the article A Fast Elitist Non-dominated Sorting Genetic Algorithm for Multi-objectiveOptimization:NSGA-II that was published on magazine IEEE Transactions on Evolutionary Computation the 6th volume in 2002 produces method with scheme group R
tIn 40 schemes be divided into 13 groups, each group represents a boundary set, to little group #, the scheme in the less group of numbering is than the priority height of the scheme of the bigger group of numbering; So-called priority level is little of excellent by energy consumption and response time.The scheme that the group that obtains as stated above is 1 li has scheme 19, scheme 40, and the scheme of group 2 the insides has scheme 13, scheme 27, scheme 39, the scheme that group is 3 li has scheme 2, scheme 11, scheme 21, scheme 26, scheme 35, scheme 36, the scheme of group 4 the insides has scheme 16, scheme 28, scheme 38, the scheme that group is 5 li has scheme 8, scheme 24, scheme 34, the scheme that group is 6 li has scheme 14, scheme 37, the scheme of group 7 the insides has scheme 18, scheme 29, scheme 33, the scheme that group is 8 li has scheme 5, scheme 23, scheme 32, the scheme of group 9 the insides has scheme 4, scheme 17, scheme 22, scheme 30, scheme 31, the scheme that group is 10 li has scheme 1, scheme 20, the scheme of group 11 the insides has scheme 9, scheme 10, scheme 12, scheme 15, scheme 25, the scheme that group is 12 li has scheme 3, scheme 6, and the scheme that group is 13 li has scheme 7.
12, new scheme group F who does not have scheme of initialization
t, divide the good group from the 10th step then, by the numbering order from small to large of group, from group 1 to group 13, successively the scheme in the group is joined F
tIn, up to scheme group F
tIn the scheme number surpass 20, the scheme of group's (being group 7) that will add at last this moment is all taken out.
13, the priority ranking of each scheme in each group that determines to obtain in the 10th step, concrete grammar is as follows:
12-1 makes that the distance value of each scheme in the group is 0, and described distance value represents whether getting in touch of this scheme and other scheme be tight.
12-2 carries out inverted order according to the 9th size that goes on foot the energy consumption of each scheme that calculates to scheme and arranges, and the less scheme of energy consumption comes the front of the bigger scheme of energy consumption,
12-3 calculates the energy consumption distance of each scheme, and the energy consumption of each scheme distance deducts the resulting value of energy consumption of that scheme that comes its front for the energy consumption of that scheme of coming its back,
12-4 carries out the inverted order arrangement according to the response time size of each scheme that the 9th step calculated to scheme, and the response time, less scheme came the front of bigger scheme of response time
12-5 calculates the response time distance of each scheme, and response time of each scheme distance deducts the resulting value of response time of that scheme that comes its front for response time of that scheme of coming its back,
12-6 obtains the distance value of scheme with the response time distance of each scheme and energy consumption apart from addition,
12-7 comes the foremost to the scheme of the scheme of energy consumption minimum and response time minimum, according to the distance size other schemes are sorted then, and come the back of the scheme of the scheme of energy consumption minimum and response time minimum, the scheme that distance value is bigger comes the front of the less scheme of distance value, and the priority ratio that comes the scheme of front like this comes the priority height of the scheme of back
14, in the group after the 12nd step ordering according to priority the height selection scheme add scheme group F
t, up to scheme group F
tIn the scheme number be 20, after finishing, scheme group F
tIn scheme be: (4,8,12,12,9,4), (4,8,12,12,9,4), (5,2,4,9,0,8), (5,2,4,9,0,8), (5,2,4,9,0,8), (4,1,0,8,1,6), (8,8,9,12,10,2), (4,1,0,8,1,6), (8,8,9,12,10,2), (4,1,0,8,1,6), (8,8,9,12,10,2), (9,12,3,9,10,11), (9,12,3,9,10,11), (9,12,3,9,10,11), (3,0,8,7,1,9), (3,0,8,7,1,9), (3,0,8,7,1,9), (7,2,15,11,1,11), (7,2,15,11,1,11), (11,12,4,7,1,9)
15, with scheme group F
tIn 20 schemes be divided into 10 scheme groups at random, in the every scheme group 2 schemes are arranged, unite definite precedence relationship according to the 10th and 12 steps and select the prepreerence scheme of every scheme group, will form a new scheme group F from the prepreerence scheme that all scheme groups select then
t', scheme group F
t' in scheme comprise (4,8,12,12,9,4), (5,2,4,9,0,8), (5,2,4,9,0,8), (8,8,9,12,10,2), (8,8,9,12,10,2), (8,8,9,12,10,2), (9,12,3,9,10,11), (3,0,8,7,1,9), (3,0,8,7,1,9), (7,2,15,11,1,11)
16, with scheme group F
t' in scheme random pair again, exchange with the numbering of the node on random site between 0.4 pair of every pair of scheme of probability of the intersection set in the step 1, and then all are combined obtain scheme group H
t, that is: (4,8,12,12,9,4), (5,2,4,9,0,8), (8,2,4,9,0,8), (5,8,9,12,10,2), (8,8,9,12,10,2), (8,8,9,12,10,2), (9,12,3,9,10,11), (3,0,8,7,1,9), (3,0,8,7,1,9), (7,2,15,11,1,11)
17, the probability 0.01 with the variation of setting in the step 1 changes scheme group H
tIn the numbering of the node on random site in each scheme, obtain scheme group H
t', that is: (4,8,12,12,9,4), (5,2,4,9,0,8), (8,2,4,9,0,8), (5,8,9,12,10,2), (8,8,9,12,10,2), (8,8,9,12,10,2), (9,12,3,9,10,11), (3,0,8,7,1,9), (3,0,8,7,1,9), (7,2,15,11,1,11)
18, press fitness order from small to large, from scheme group F
t10 schemes of middle selection are with scheme group H
t' scheme combine, constitute the new scheme group J comprise 20 schemes
T+1
If 19 iterations t are less than 100, then t=t+1 turns back to step 2 then, carries out new round interative computation, otherwise enters step 19;
20, calculate scheme group J according to the described method of step 3
100In the fitness value of each scheme, with the scheme of fitness value minimum, promptly scheme (5,2,9,9,5,4) as the task of network-on-chip to internodal mapping scheme.
Claims (5)
1. be used to seek the method for mapping scheme between the task of network-on-chip and node, it is characterized in that: seek energy consumption and the response time task and the node mapping scheme of minimum simultaneously of sening as an envoy to, the searching process is as follows:
One, generates a scheme group J who comprises K allocative decision at random
t, and be each allocative decision X=(x
1, x
2..., x
N) set the Probability p of one intersection
iProbability q with variation
i, and total iterations T, and the variable t=0 of order record number of iterations;
Two, obtain scheme group J
tIn network-on-chip energy consumption and response time under every kind of scheme;
Three, calculate scheme group J
tIn the fitness value of each scheme, the priority of this fitness value size reflection scheme, it is high more be worth more little priority, so-called priority level be little by energy consumption and response time be excellent;
Four, with scheme group J
tIn K scheme be divided at random
Individual scheme group, wherein L is the scheme number in the every scheme group, selects in the every scheme group prepreerence scheme according to the precedence relationship of determining in the 3rd step and forms a priority scheme group Q
t
Five, with priority scheme group Q
tIn scheme press random pair, described pairing is meant according to two schemes divides into groups for one group, and with the Probability p of the intersection set in the step 1
iThe numbering of the node on any one same position on two schemes in every pair of scheme is exchanged, and then all are merged by every pair of scheme after exchanging obtain scheme group Q '
t
Six, the variation probability q to set in the first step
iChange scheme group Q '
tIn in each scheme in the numbering of any one locational node, obtain scheme group G
t
Seven, the fitness value order of calculating by the 3rd step from small to large is from scheme group J
tThe middle selection
Individual scheme is with scheme group G
tScheme combine, constitute the new scheme group G ' comprise K scheme
t
Eight, with scheme group J
tAnd G '
tBe merged into scheme group R
t
Nine, numerical procedure group R
tIn network-on-chip energy consumption and response time under every kind of scheme;
Ten, with scheme group R
tIn 2K scheme be divided into a plurality of groups, each group represents a boundary set, to little group #, the scheme in the less group of numbering is than the priority height of the scheme of the bigger group of numbering; So-called priority level is little of excellent by energy consumption and response time;
11, new scheme group F who does not have scheme of initialization
t, from the good group of the tenth step branch, the numbering order from big to small by group joins F with the scheme in the group successively then
tIn, up to scheme group F
tIn the scheme number surpass K, the scheme of the group that will add is at last all taken out then, and the group that will be added at last is designated as V;
12, determine in the tenth step the priority ranking of each scheme in each group, concrete grammar is as follows:
12-1 makes that the distance value of each scheme in the group is 0, and described distance value represents whether getting in touch of this scheme and other scheme be tight;
12-2 carries out inverted order according to the 9th size that goes on foot the energy consumption of each scheme that calculates to scheme and arranges, and the less scheme of energy consumption comes the front of the bigger scheme of energy consumption,
12-3 calculates the energy consumption distance of each scheme, and the energy consumption of each scheme distance deducts the resulting value of energy consumption of that scheme that comes its front for the energy consumption of that scheme of coming its back,
12-4 carries out the inverted order arrangement according to the response time size of each scheme that the 9th step calculated to scheme, and the response time, less scheme came the front of bigger scheme of response time,
12-5 calculates the response time distance of each scheme, and response time of each scheme distance deducts the resulting value of response time of that scheme that comes its front for response time of that scheme of coming its back,
12-6 obtains the distance value of scheme with the response time distance of each scheme and energy consumption apart from addition,
12-7 comes the foremost to the scheme of the scheme of energy consumption minimum and response time minimum, according to the distance size remaining scheme is sorted then, and come the back of the scheme of the scheme of energy consumption minimum and response time minimum, the scheme that distance value is bigger comes the front of the less scheme of distance value, and the priority ratio that comes the scheme of front like this comes the priority height of the scheme of back;
13, join scheme group F from the select progressively scheme of having passed through in the V of group after the ordering of the 12 step according to priority from high to low
t, up to scheme group F
tIn the scheme number be K;
14, with scheme group F
tIn K scheme be divided at random
Individual scheme group, wherein L is the scheme number in the every scheme group, selects priority scheme group F ' of prepreerence scheme composition in the every scheme group according to the precedence relationship of determining in the tenth step and the 12 step
t
15, with priority scheme group F '
tIn scheme random pair again, with the Probability p of the intersection set in the first step
iThe numbering of the node on any one same position on two schemes in every pair of scheme is exchanged, and then all are merged by every pair of scheme after exchanging obtain scheme group H
t
16, with the probability q of the variation set in the first step
iChange scheme group H
tIn in each scheme in the numbering of any one locational node, obtain scheme group H '
t
17, unite definite scheme group F according to the tenth step and the 12 step
tThe priority ranking of middle scheme is according to priority order from high to low, from scheme group F
tThe middle selection
Individual scheme is with scheme group H
tScheme combine, constitute the new scheme group J comprise K scheme
T+1
If 18 t＜T, then t=t+1 turned back to for second step then, otherwise entered for the 19 step;
19, calculate scheme group J
tIn the fitness value of each scheme, with the scheme of fitness value minimum, promptly minimum simultaneously scheme of energy consumption and response time as the task of network-on-chip to internodal mapping scheme;
2. according to the method that is used to seek mapping scheme between the task of network-on-chip and node of claim 1, it is characterized in that: described energy consumption is the energy consumption that the energy consumption sum of all processing nodes on the network-on-chip is added the communication network that connects each processing node.
3. according to the method that is used to seek mapping scheme between the task of network-on-chip and node of claim 1 or claim 2, it is characterized in that: the described response time is that the processing time sum that is input to all nodes on the outgoing route adds that the time-delay sum that is input to all links on the outgoing route adds queuing and the stand-by period sum that is input to packet on the outgoing route.
4. according to the method that is used to seek mapping scheme between the task of network-on-chip and node of claim 1 or claim 2, it is characterized in that: the fitness value that calculates each scheme adopts one of exclusive method, banker method, challenge tournament method, recurrence method, quick sort or improved quick sort.
5. according to the method that is used to seek mapping scheme between the task of network-on-chip and node of claim 3, it is characterized in that: the fitness value that calculates each scheme adopts one of exclusive method, banker method, challenge tournament method, recurrence method, quick sort or improved quick sort.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010100281577A CN101799838B (en) | 2010-01-22 | 2010-01-22 | Method for seeking mapping scheme between tasks and nodes of network on chip |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2010100281577A CN101799838B (en) | 2010-01-22 | 2010-01-22 | Method for seeking mapping scheme between tasks and nodes of network on chip |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101799838A CN101799838A (en) | 2010-08-11 |
| CN101799838B true CN101799838B (en) | 2011-12-28 |
Family
ID=42595512
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2010100281577A Expired - Fee Related CN101799838B (en) | 2010-01-22 | 2010-01-22 | Method for seeking mapping scheme between tasks and nodes of network on chip |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101799838B (en) |
Families Citing this family (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102073765B (en) * | 2011-01-10 | 2012-06-27 | 四川大学 | Position probability distribution-based energy consumption model of network-on-chip (NoC) directory protocol |
| CN102075429B (en) * | 2011-01-21 | 2014-12-17 | 北京邮电大学 | Virtual network mapping method based on principle of proximity |
| CN102546232B (en) * | 2011-11-03 | 2014-12-17 | 北京邮电大学 | Multi-topology mapping method of virtual network |
| CN103678245B (en) * | 2013-12-20 | 2017-04-19 | 武汉科技大学 | Low-power-consumption on-chip network task mapping method |
| CN103761212B (en) * | 2014-01-21 | 2016-08-31 | 电子科技大学 | The method for designing of mapping scheme and topological structure between task and node in network-on-chip |
| CN103885842B (en) * | 2014-03-19 | 2017-08-25 | 浙江大学 | A kind of band accelerates the overall duty mapping method of the optimization of the network-on-chip of node |
| CN104780114B (en) * | 2015-05-08 | 2018-01-02 | 电子科技大学 | Wireless multi-core array focus removing method and framework based on butterfly network coding |
| CN107291553B (en) * | 2017-06-30 | 2020-05-12 | 电子科技大学 | Task scheduling method based on block splicing |
| CN107395507B (en) * | 2017-08-31 | 2019-09-03 | 电子科技大学 | A kind of test the map of perception method for network-on-chip NoC |
| CN108965337B (en) * | 2018-09-17 | 2021-07-30 | 新华三信息安全技术有限公司 | Rule matching method and device, firewall equipment and machine-readable storage medium |
| CN111324197B (en) * | 2020-02-07 | 2023-03-07 | 华侨大学 | Method for reducing system energy consumption based on three-speed periodic task |
| CN113822423A (en) * | 2021-09-30 | 2021-12-21 | 上海商汤智能科技有限公司 | Scheduling method, scheduling device, computer equipment, network on chip and storage medium |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1337807A (en) * | 2000-08-04 | 2002-02-27 | 因特隆公司 | Method and association for making the only connection in multi-node network to adapt the highest data rate |
-
2010
- 2010-01-22 CN CN2010100281577A patent/CN101799838B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1337807A (en) * | 2000-08-04 | 2002-02-27 | 因特隆公司 | Method and association for making the only connection in multi-node network to adapt the highest data rate |
Non-Patent Citations (2)
| Title |
|---|
| 凌翔等.基于FPGA阵列的超大规模SoC验证平台.《系统仿真学报》.2007,第19卷(第9期),第1967-1970页. * |
| 徐成等.延迟优化的片上网络低功耗映射.《计算机应用研究》.2009,第26卷(第10期),第3676-3679页. * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101799838A (en) | 2010-08-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101799838B (en) | Method for seeking mapping scheme between tasks and nodes of network on chip | |
| Wang et al. | Evolutionary Markov dynamics for network community detection | |
| CN112181867B (en) | On-chip network memory controller layout method based on multi-target genetic algorithm | |
| CN103761212B (en) | The method for designing of mapping scheme and topological structure between task and node in network-on-chip | |
| Trik et al. | Providing an Adaptive Routing along with a Hybrid Selection Strategy to Increase Efficiency in NoC‐Based Neuromorphic Systems | |
| CN109840154B (en) | Task dependency-based computing migration method in mobile cloud environment | |
| CN104145281A (en) | Neural network computing apparatus and system, and method therefor | |
| CN112468401B (en) | Network-on-chip routing communication method for brain-like processor and network-on-chip | |
| CN110163357A (en) | A kind of computing device and method | |
| Muppala et al. | Composite performance and availability analysis using a hierarchy of stochastic reward nets | |
| Hu et al. | Evaluating node importance with multi-criteria | |
| Zhang et al. | Random walks in modular scale-free networks with multiple traps | |
| CN112464611B (en) | Automatic PCB wiring system based on cloud-end collaborative intelligent processing | |
| CN103428804A (en) | Method for searching mapping scheme between tasks and nodes of network-on-chip (NoC) and network code position | |
| CN102325089A (en) | Fat tree type network-on-chip mapping method based on differential evolution and predatory search strategy | |
| CN104536831B (en) | A kind of multinuclear SoC software image methods based on multiple-objection optimization | |
| Zhang et al. | Trapping in scale-free networks with hierarchical organization of modularity | |
| CN105069517B (en) | Power distribution network multiple target fault recovery method based on hybrid algorithm | |
| Guo et al. | A novel cluster-head selection algorithm based on hybrid genetic optimization for wireless sensor networks | |
| Yang et al. | Trust-based scheduling strategy for cloud workflow applications | |
| Le et al. | Performance-driven assignment and mapping for reliable networks-on-chips | |
| CN112711906B (en) | Optimization method for wireless sensor network energy balance problem | |
| Tutsch et al. | Comparison of network-on-chip topologies for multicore systems considering multicast and local traffic | |
| Morgan et al. | Multi-objective optimization of NoC standard architectures using Genetic Algorithms | |
| CN112001141B (en) | Brain network inspired middle-large scale on-die interconnection system comprehensive method |
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 |
Granted publication date: 20111228 Termination date: 20150122 |
|
| EXPY | Termination of patent right or utility model |