CN105024926B - A kind of fault message transmission method applied to Mesh topology network-on-chips - Google Patents
A kind of fault message transmission method applied to Mesh topology network-on-chips Download PDFInfo
- Publication number
- CN105024926B CN105024926B CN201510436479.8A CN201510436479A CN105024926B CN 105024926 B CN105024926 B CN 105024926B CN 201510436479 A CN201510436479 A CN 201510436479A CN 105024926 B CN105024926 B CN 105024926B
- Authority
- CN
- China
- Prior art keywords
- router
- row
- link
- failure
- link failure
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/28—Routing or path finding of packets in data switching networks using route fault recovery
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/06—Management of faults, events, alarms or notifications
- H04L41/0654—Management of faults, events, alarms or notifications using network fault recovery
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses a kind of link failure information transmission method applied to Mesh topology network-on-chips, comprise the following steps:S1. the width that the orthogonal direction of link failure information is propagated is set, the width means that the orthogonal direction of link failure information is propagated are N;S2. newly-increased link failure information is obtained, this is increased newly link failure information and travels in the propagation regions centered on faulty link;S3. fault zone is the region that can produce the faulty link composition interfered between each other, and each router updates current failure region according to the newly-increased link failure information received;S4. the also Unrecorded link failure information of each router is propagated in current failure region.The present invention passes through the setting for the width propagated orthogonal direction, and the self-adapting detecting to fault zone, determine fault propagation region, each router that link failure information is traveled in this region, requirement of the Fault-tolerant Routing Algorithm for link failure information of Different Optimization target can be met.
Description
Technical field
The present invention relates to interconnection technique field on integrated circuit chip, is applied to Mesh topologys piece more particularly to one kind and surfs the Net
The link failure information transmission method of network.
Background technology
With the progress of integrated technique, on-chip system(System-on-Chip, SoC)In integrate functional unit species
Increased sharply with quantity.Traditional shared communication structure and direct communication structure can not all be tackled between chip multi-core increasingly well
The communication pressure of growth.Network-on-chip(Network-on-Chip, NoC)It is a kind of new communicating between multi-kernel structure, compared to
Traditional communication structure, network-on-chip can provide higher bandwidth and flexibility, while be obtained more between performance and cost
Good balance.
Relatively conventional a kind of network-on-chip is Mesh topology network-on-chips under study for action, Mesh topology network-on-chips Road
By device according to ranks regular array, the rectangle that m rows × n is arranged is formed.Router can use its coordinate(x,y)(0≤x≤m-1,0≤
y≤n-1)Represent, wherein x is the row coordinate of router(Or X-axis coordinate), y is the row coordinate of router(Or Y-axis is sat
Mark).The port of router is in X-direction or Y direction respectively.X-direction where port(Or Y direction)Referred to as
Port direction, Y direction(Or X-direction)Referred to as orthogonal direction.
But as the rapid reduction of process, the device stability of integrated circuit gradually reduce.Therefore, electricity is integrated
The significant challenge that road surface is faced is to realize reliable system using insecure device.In order to reach this target, fault-tolerant skill
Art becomes extremely important.
Common failure has link failure and router failure etc. in network-on-chip, and link failure refers to for some reason
Cause the communication from some router to its some neighbour unsuccessful.Router failure can be regarded as each port of router
All there occurs link failure.For the router of a failure, the failure of router line direction port is referred to as line direction event
Barrier, includes two link failures;Similarly, the failure of router column direction port is referred to as column direction failure, includes two links
Failure.
In network-on-chip, a kind of very important fault-toleranr technique is exactly Fault-tolerant Routing Algorithm(fault-tolerant
routing algorithm).These algorithms are by using known network topological information, packet information and device fault
Information, packet is bypassed the device of failure, ensure the accessibility of packet.For this method, the propagation side of fault message
Method and routing algorithm are of equal importance.The quantity of fault message known to routing algorithm directly governs routing algorithm and prolonged in propagation
Late, the number of faults and the performance of pattern etc. supported.In brief, if fault message is more known to routing algorithm,
The performance and Potential performance of routing algorithm are higher.
Certainly, nor all fault messages are necessary to inform each router.When the road for carrying out router-level topology
During by device and distant defective device, fault message has no influence to the router-level topology of this router.In this case, should
Fault propagation region should be reduced as far as possible, to reduce the cost of information propagation.When multiple defective devices are closer to the distance, can satisfy the need again
Footpath produces overlaying influence, at this moment, needs suitably to expand propagation regions again so that routing algorithm can make full use of failure to believe
Breath.
It is most of not use fault propagation method in existing design, only with the failure of local router
The fault message that detection circuit obtains is route, and its spread scope is only limitted to the neighbor node of failed router;Small part is set
Fault propagation method has been used in meter, its spread scope, which is also confined to distance fault router, to jump the interior place reached 2,
As shown in Figure 1.
The content of the invention
It is an object of the invention to overcome the deficiencies of the prior art and provide a kind of chain applied to Mesh topology network-on-chips
Road fault message transmission method, by the setting for the width propagated in orthogonal direction failure, and to the adaptive of fault zone
It should detect, determine fault propagation region, each router that link failure information is traveled in this region, can meet not
With requirement of the Fault-tolerant Routing Algorithm for link failure information of optimization aim.
The purpose of the present invention is achieved through the following technical solutions:A kind of chain applied to Mesh topology network-on-chips
Road fault message transmission method, comprises the following steps:
S1. the propagation regions of link failure information are set:Set the width that the orthogonal direction of link failure information is propagated, chain
The width means that the orthogonal direction of road fault message is propagated are N;Failed router is that the router of link failure, failure end occurs
Mouthful for occur link failure port, then the propagation regions of link failure information for failed router non-working port column or
All-router on the adjacent N column or row of all-router and failed router on row;
S2. newly-increased link failure information is spread:Newly-increased link failure information is obtained, increasing this newly link failure information propagates
Into the propagation regions centered on faulty link;
S3. current failure region is updated:Fault zone is that can produce the faulty link composition interfered between each other
Region, each router update current failure region according to the newly-increased link failure information received;
S4. supplement diffusion link failure information:The also Unrecorded link event of each router is propagated in current failure region
Hinder information.
Each router is according to the mode in the newly-increased link failure information renewal current failure region received in step S3:
When router previously without record fault message when, newly-increased fault zone is obtained, using newly-increased fault zone as ought
Prior fault region;
When router precedence record during fault message, obtain newly-increased fault zone, by known fault region and it is newly-increased therefore
Hinder the union in region as current failure region.
The mode for propagating the also Unrecorded link failure information of each router in step S4 in current failure region is:Will
Newly-increased link failure information is broadcast to the router in known fault region, and known link fault message is broadcast into newly-increased failure
Router in region.
The faulty link interfered described in step S3 is faulty link adjacent in line direction or in a column direction.
Fault zone described in step S3 is:
If only a row or column has faulty link, the row or column is defined as fault zone.
It is continuous that if more row or column, which have faulty link and the row or column of faulty link be present,:When consecutive rows of per a line
When faulty link be present, by the behavior fault zone between consecutive rows of initial row and end line, initial row and end line are determined
Justice is the border of fault zone;When each row of continuation column have link failure, by the starting row and end column of continuation column
Between definition be fault zone, starting row and end column are defined as the border of fault zone.
If there is faulty link and the row or column of faulty link be present to be discontinuous in more row or column:Then event be present by all
The union for hindering the row and column of link is defined as fault zone.
Described known fault region and the acquisition pattern of the union of newly-increased fault zone are:To being in together on line direction
Fault zone takes union, and union is taken to the fault zone being in together on column direction.
The beneficial effects of the invention are as follows:Present invention determine that the pattern that link failure information is propagated, and can specify
The size of propagation regions, by the self-adapting detecting to fault zone, fault propagation region is determined, link failure information is propagated
To each router in this region, the Fault-tolerant Routing Algorithm of Different Optimization target can be met for link failure information
It is required that.
Brief description of the drawings
Fig. 1 is the propagation regions schematic diagram in existing link failure information transmission method;
Fig. 2 is the FB(flow block) for the link failure information transmission method that the present invention is applied to Mesh topology network-on-chips;
Fig. 3 is the link failure propagation regions schematic diagram of the present invention;
Fig. 4 is network-on-chip coordinate-system and the definition schematic diagram of fault zone;
Fig. 5 is the particular flow sheet of the present invention;
The link failure information that Fig. 6 is failure I when the width for having two link failures and orthogonal direction to propagate is 1 is propagated
Schematic diagram;
Fig. 7 is that the link failure information of failure II when the width for having two link failures and orthogonal direction to propagate is 1 is propagated
Schematic diagram;
Fig. 8 is that supplement spreads showing for link failure information when the width for having two link failures and orthogonal direction to propagate is 1
It is intended to;
The link failure information propagation that Fig. 9 is failure I when the width for having two link failures and orthogonal direction to propagate is 2 is shown
It is intended to;
Figure 10 is that the link failure information of failure II when the width for having two link failures and orthogonal direction to propagate is 2 is propagated
Schematic diagram;
Figure 11 is that supplement spreads showing for link failure information when the width for having two link failures and orthogonal direction to propagate is 2
It is intended to;
The link failure information that Figure 12 is failure I when the width for having three link failures and orthogonal direction to propagate is 1 is propagated
Schematic diagram;
Figure 13 is that the link failure information of failure II when the width for having three link failures and orthogonal direction to propagate is 1 is propagated
Schematic diagram;
Figure 14 is that the link failure information of failure III when the width for having three link failures and orthogonal direction to propagate is 1 is propagated
Schematic diagram;
Figure 15 is the of supplement diffusion link failure information when the width for having three link failures and orthogonal direction to propagate is 1
One schematic diagram;
Figure 16 is the of supplement diffusion link failure information when the width for having three link failures and orthogonal direction to propagate is 1
Two schematic diagrames.
Embodiment
Technical scheme is described in further detail below in conjunction with the accompanying drawings, but protection scope of the present invention is not limited to
It is as described below.
In the present invention, the base unit of failure is link failure, i.e., from some router to its some neighboring router
Communication it is unsuccessful;Router failure be then represented by failed router to its all neighboring router communication not into
Work(.
As shown in Fig. 2 a kind of link failure information transmission method applied to Mesh topology network-on-chips, including following step
Suddenly:
S1. the propagation regions of link failure information are set:Set the width that the orthogonal direction of link failure information is propagated, chain
The width means that the orthogonal direction of road fault message is propagated are N;Failed router is that the router of link failure, failure end occurs
Mouthful for occur link failure port, then the propagation regions of link failure information for failed router non-working port column or
All-router on the adjacent N column or row of all-router and failed router on row, as shown in Figure 3.
The propagation regions of one isolated link failure, it is network size in the length of port direction, in that orthogonal direction
Depending on according to demand, the border of propagation regions is referred to as to propagate border.
For each link failure, the propagation of its link failure information is divided into port direction and propagated and orthogonal direction propagation.
Port direction propagates the all-router passed to link failure information on the column or row of faulty link place;Orthogonal direction is propagated
It is then the router being broadcast to link failure information on the adjacent column or row of faulty link.
S2. newly-increased link failure information is spread:Newly-increased link failure information is obtained, increasing this newly link failure information propagates
Into the propagation regions centered on faulty link, the propagation regions are a rectangles, are whole column or row in port direction
On all-router, be in that orthogonal direction the all-router on 2N+1 column or row.
For some router(Referred to as local router)For, its approach for obtaining link failure information has two kinds:The
A kind of approach is directly to be obtained by failure detector circuit, and second of approach is transmitted to by neighboring router.Two kinds of ways
Footpath all can directly inform this local router, and the non-working port of failed router breaks down, or other equivalent information lattice
Formula.
, three kinds of situations be present in the newly-increased link failure information of diffusion:The first situation, if failed router and local routing
Device is gone together in the port direction of non-working port or same column, and link failure information passes to the port in port direction first
Pass;, will also be to the orthogonal direction transmission of non-working port if the width that orthogonal direction is propagated is more than 0;Second of situation, it is discontented with
During the first situation of foot, local router is in the width that the orthogonal direction of the non-working port of failed router is propagated and is not to pass
Border is broadcast, then is continued to external diffusion link failure information;The third situation, non-working port of the local router in failed router
Orthogonal direction propagate width outside or propagate border on, then this link failure information so far no longer spreads.
S3. current failure region is updated:Fault zone is that can produce the faulty link composition interfered between each other
Region, each router update current failure region according to the newly-increased link failure information received.
Each router is according to the mode in the newly-increased link failure information renewal current failure region received in step S3:
When router previously no record fault message, newly-increased fault zone is obtained, using newly-increased fault zone as current failure area
Domain;When router precedence record during fault message, newly-increased fault zone is obtained, by known fault region and newly-increased fault zone
Union as current failure region.
When updating current failure region, union only is taken to the fault zone in same direction, with orthogonal propagation width
It is unrelated.In addition, the link failure information for retaining known fault region is still to, to carry out supplement diffusion.
The faulty link interfered described in step S3 is faulty link adjacent in line direction or in a column direction.
Fault zone described in step S3 is:If only a row or column has faulty link, the row or column is determined
Justice is fault zone.It is continuous that if more row or column, which have faulty link and the row or column of faulty link be present,:When consecutive rows of
When faulty link be present per a line, by the behavior fault zone between consecutive rows of initial row and end line, initial row and knot
Beam row is defined as the border of fault zone;When link failure be present in each row of continuation column, by the starting of continuation column row and
Definition between end column is fault zone, and starting row are defined as the border of fault zone with end column.
If there is faulty link and the row or column of faulty link be present to be discontinuous in more row or column:Then event be present by all
The union for hindering the row and column of link is defined as fault zone.
Described known fault region and the acquisition pattern of the union of newly-increased fault zone are:To being in together on line direction
Fault zone takes union, and union is taken to the fault zone being in together on column direction.
As shown in figure 4, open circles indicate normal router, filled circles represent failed router, in the X-axis direction, from 0
To all faulty router of 5 rows, therefore the fault zone of X-axis is [0,5], and failure boundary is the 0th row and the 5th row.And Y direction
On, fault zone is more dispersed:Failure, and the 1st row, the 3rd row and the 6th row fault-free, therefore event are shown in 0th row, the 2nd row and 7
It is respectively [0,0], [2,2] and [7,7] to hinder region, and the border of this three fault zones is respectively:0th row, the 2nd row and the 7th
OK.4th and 5 row are all faulty, and the 3rd row and the 6th no failure of row, therefore fault zone be [4,5], and its failure boundary is the
4 rows and the 5th row.
S4. supplement diffusion link failure information:The also Unrecorded link event of each router is propagated in current failure region
Hinder information.
The mode for propagating the also Unrecorded link failure information of each router in step S4 in current failure region is:Will
Newly-increased link failure information is broadcast to the router in known fault region, and known link fault message is broadcast into newly-increased failure
Router in region.If router is on the propagation border in known fault region, to the orthogonal side backwards to known fault region
To all known link fault messages of transmission;If router is increasing the propagation border of fault zone newly, to backwards to newly-increased failure
The orthogonal direction in region transmits all newly-increased link failure informations.
The condition of newly-increased link failure information diffusion is that router is in the propagation border of newly-increased link failure information, and
In the propagation regions of known fault link failure information;Similarly, it is known that the condition of link failure information diffusion is router
The propagation border of known link fault message is in, and in the propagation regions of newly-increased link failure information.
As shown in figure 5, use router(X0, y0)Failed router is represented, port a represents non-working port, local router
Other routers are represented, then the medium process of link failure information is:
Local router obtains router by failure detector circuit or neighboring router(X0, y0)Port a failures
Link failure information, with fault message triple(X0, y0, a)Represent.
Judge local router and failure(X0, y0, a)Whether with port direction:If local router and failure
(X0, y0, a)With in port direction, then believe from local router to the router transmission link failure in port direction
Breath.And judge whether the width that orthogonal direction is propagated is more than 0.If the width that orthogonal direction is propagated is more than 0, by local router
To the router transmission link fault message on port a orthogonal direction.If local router and failure(X0, y0, a)Difference exists
In port direction, then judge local router whether in failure(X0, y0, a)Orthogonal direction propagate width in.If in orthogonal side
Into spread scope, then from local router to backwards to failure(X0, y0, a)Orthogonal direction on router transmission link
Fault message.
Judge whether current failure region needs to update:If current failure region needs to update, known fault region is taken
Whether the union with newly-increased fault zone is current failure region, and judge local router on the propagation side of known link failure
Boundary;If current failure region need not update, stop link failure information and propagate.
Judge local router whether on the propagation border of known link failure:If local router is in known link failure
Propagation border, then transmit all known link fault messages to backwards to the orthogonal direction port in known fault region, and judge
Local router is on the propagation border of newly-increased link failure;If local router not on the propagation border of known link failure,
Judge local router whether on the propagation border of newly-increased link failure.
Judge local router whether on the propagation border of newly-increased link failure:If local router is not in the event of newly-increased link
The propagation border of barrier, then stop link failure information and propagate;If ground router is increasing the propagation border of link failure newly, support or oppose
All newly-increased link failure informations are transmitted to the orthogonal direction of newly-increased fault zone, then stops link failure information and propagates.
Fig. 6 to Figure 16 respectively show three typical fault propagation examples, in figure, with × represent faulty link or failure
Router, it is indicated by means of an arrow the direction of link failure information propagation.The link failure for showing to propagate with different patterns in figure is believed
Breath, a kind of striped(Horizontal stripe, nicking or inclined stripe)Corresponding to a link failure information, the overlapping of striped represents router
Know a variety of link failure informations.
When two link failures:Link failure between router (3,3) and router (4,3)(Failure I)And router
(1,4) link failure between router (2,4)(Failure II).It is assumed that the width that orthogonal direction is propagated is 1, such as Fig. 6, Fig. 7 and
Shown in Fig. 8.
Failure I occurs, and router (3,3) and router (4,3) detect link failure, carry out new link failure information biography
Broadcast, as shown in Figure 6.
2nd row, the 3rd row and the 4th row router know failure I, and the fault zone of Y direction is [3,3], and X-direction does not have
Faulty region.
Failure II occurs, and router (Isosorbide-5-Nitrae) and router (2,4) detect link failure, carry out new link failure information
Propagate, as shown in Figure 7.
2nd row know failure I, and the 3rd row and the 4th row router know failure I and failure II, and the 5th row router knows event
Hinder II, the fault zone of Y direction is [3,3] and [4,4], and X-direction does not have fault zone.
Failure I is continuous in Y direction with failure II, and fault zone merges.The fault zone of 3rd row and the 4th row router
For [3,4].
3rd row router is in failure II propagation border, and in failure I propagation regions, therefore, backwards to failure II
Propagation regions, westwards fault propagation II link failure information;4th propagation border of the row in failure I, and failure II's
In propagation regions, therefore, backwards to failure I, eastwards fault propagation I link failure information, as shown in Figure 8.
The failure I and failure II that 2nd row, the 3rd row, the 4th row and the 5th row router know, fault zone is [3,4].
Because the 2nd row and the 5th row router are respectively on failure I and failure II propagation regions border, so link failure
Information is propagated so far.
In Y direction:0th row, the 1st row, the 6th row and the 7th row router are unaware of link failure information;2nd row, the 3rd
Row, the 4th row and the 5th row router know failure I and failure II, and fault zone is [3,4].There is no link failure in X-direction
Information.
When two router failures:Router (4,3) failure(Failure I)With router (2,4) failure(Failure II).It is false
The width for determining orthogonal direction propagation is 2, as shown in Fig. 9, Figure 10 and Figure 11.
Failure I occurs, and router (3,3), router (4,2), router (5,3) and router (4,4) detect failure,
New fault propagation is carried out, as shown in Figure 9.
Y direction:1st row, the 2nd row, the 3rd row, the 4th row and the 5th row router know failure I, fault zone for [3,
3];X-direction:2nd row, the 3rd row, the 4th row, the 5th row and the 6th row router know failure I, and fault zone is [4,4].
Failure II occurs, and router (Isosorbide-5-Nitrae), router (2,3), router (3,4) and router (2,5) detect event
Barrier, carries out new fault propagation, as shown in Figure 10.
Y direction:1st row router knows failure I, and the 2nd row, the 3rd row, the 4th row and the 5th row know failure I and failure
II, the 6th row router know failure III, and fault zone is [3,3] and [4,4].X-direction:0th row and the 1st row router are known
Know failure II, the 2nd row, the 3rd row and the 4th row router know failure I and failure II, and the 5th row and the 6th row router know failure
I, fault zone are [2,2] and [4,4].
Failure I and failure II is continuous and discontinuous in Y direction in X-direction, so two faulty sections in the X-axis direction
Domain can merge, and fault zone is constant in the Y-axis direction, update fault zone.2nd row, the 3rd row, the 4th row and the 5th row route
The fault zone of device is [3,4].
2nd row router is in failure II propagation border, and in failure I propagation regions, therefore, backwards to failure 1,
Fault propagation II failure message westwards;5th row router is in failure I propagation border, and in failure II propagation regions
It is interior, therefore, backwards to failure I, eastwards fault propagation I information, as shown in figure 11.
The failure I and failure II that 1st row, the 2nd row, the 3rd row, the 4th row, the 5th row and the 6th row router know, faulty section
Domain is [3,4].
Because the 2nd row and the 5th row router are respectively on failure I and failure II propagation regions border, so link failure
Information is propagated so far.
In Y direction:0th row and the 7th row router are unaware of fault message;1st row, the 2nd row, the 3rd row, the 4th row, the
5 row and the 6th row router know failure I and failure II, and fault zone is [3,4].In X-direction:7th row router is unaware of
Fault message;5th row and the 6th row router know failure I, and fault zone is [4,4];2nd row, the 3rd row and the 4th row router
Know failure 1 and failure II, fault zone is [2,2] and [4,4];0th row and the 1st row router know failure II, faulty section
Domain is [2,2].
When three link failures:From router (4,2) to the link failure of router (5,2)(Failure I), router
(3,4) link failure of router (4,4) is arrived(Failure II)The link failure of router (2,3) is arrived with router (1,3)(Failure
III).It is assumed that the width that orthogonal direction is propagated is 1, as shown in accompanying drawing 12, Figure 13, Figure 14, Figure 15 and Figure 16.
Failure I occurs, and router (4,2) and router (5,2) detect failure, carry out new fault propagation, such as Figure 12 institutes
Show.
1st row, the 2nd row and the 3rd row router know failure I, and the fault zone of Y direction is [2,2], and X-direction does not have
Faulty region.
Failure 2 occurs, and router (3,4) and router (4,4) detect failure, carry out new fault propagation, such as Figure 13 institutes
Show.
1st row and the 2nd row router know failure I, and the 3rd row router knows failure I and II, and the 4th row and the 5th row are route
Device knows failure II, and the fault zone of Y direction is [2,2] and [4,4], and X-direction does not have fault zone.
Failure I and failure II is 2 in the distance of X-direction, therefore two fault zones wouldn't merge.Now, the 1st row and
The fault zone of 2nd row router is [2,2], and the fault zone of the 3rd row router is [2,2] and [4,4], the 4th row and the 5th row
The fault zone of router is [4,4].
Failure III occurs, and router (1,3) and router (2,3) detect failure, carry out new link failure information biography
Broadcast, as shown in figure 14.
1st row router knows failure I, and the 2nd row router knows failure I and III, the 3rd row router know failure I,
Failure II and failure III, the 4th row router know failure II and failure III, and the 5th row router knows failure II, fault zone
For [2,2], [3,3] and [4,4].
Failure I is continuous in Y direction with failure III, and failure I+ failures III is merged into fault zone.Failure II and failure
III is continuous in Y direction, and failure II+ failures III is merged into fault zone.The fault zone of 2nd row router is [2,3], the
The fault zone of 3 row routers is [2,3] and [3,4], and the fault zone of the 4th row router is [3,4].
2nd row router is in failure III propagation border, and in failure I propagation regions, therefore, backwards to failure
The message of III propagation regions, westwards fault propagation III;Propagation border of the III row in failure I, and failure III's
In propagation regions, therefore, backwards to the information of failure I propagation regions, eastwards fault propagation I, as shown in figure 15.
Similarly, the 4th row router is in failure III propagation border, and in failure II propagation regions, therefore,
Backwards to failure III propagation regions, i.e. fault propagation III message eastwards;3rd row router is in failure II propagation border, and
In failure III propagation regions, therefore, backwards to failure II propagation regions, westwards fault propagation I fault message.Step f and
Step g can be carried out simultaneously or exchange sequence, as shown in figure 15.
1st row router knows failure I+ failure III, fault zone [2,3];2nd row, the 3rd row and the 4th row router are known
Know failure I+ failures III and failure II+ failure III, fault zone is [2,3] and [3,4];5th row router knows failure II+
Failure III, fault zone [3,4].
Because the 1st row and the 4th row are respectively on failure I and failure III propagation regions border, so fault propagation is to this
Only.Because the 2nd row and the 5th row are respectively on failure II and failure III propagation regions border, so fault propagation is so far.
Failure I+ failures III and failure II+ failures III are overlapped in Y direction, and fault zone can merge.2nd
The fault zone of row, the 3rd row and the 4th row is [2,3,4].
2nd row router is in failure II+ failures III propagation border, and in failure I+ failures III propagation regions
It is interior, therefore, backwards to failure II+ failure III propagation regions, i.e. westwards fault propagation II message;4th row router is in failure I+
Failure III propagation border, and in failure II+ failures III propagation regions, therefore, backwards to failure I+ failure III propagation zones
The fault message in domain, eastwards fault propagation I, as shown in figure 16.
Because the 1st row and the 5th row router are respectively on failure I+ failures III and failure I+ failures II propagation regions side
Boundary, so fault propagation is so far.
In Y direction:0th row, the 6th row and the 7th row router are unaware of fault message;1st row, the 2nd row, the 3rd row, the
4 row and the 5th row router know failure I, failure II and failure III, and fault zone is [2,4].There is no failure letter in X-direction
Breath.
Claims (5)
- A kind of 1. link failure information transmission method applied to Mesh topology network-on-chips, it is characterised in that:Including following step Suddenly:S1. the propagation regions of link failure information are set:Set the width that the orthogonal direction of link failure information is propagated, link event The width means for hindering the orthogonal direction propagation of information are N;For the router of link failure occurs, non-working port is failed router The port of link failure occurs, then the propagation regions of link failure information is on the non-working port place column or row of failed router All-router and the adjacent N column or row of failed router on all-router;S2. newly-increased link failure information is spread:Obtain newly-increased link failure information, by this increase newly link failure information travel to In propagation regions centered on faulty link;S3. current failure region is updated:Fault zone is the region that can produce the faulty link composition interfered between each other, Each router updates current failure region according to the newly-increased link failure information received;S4. supplement diffusion link failure information:The also Unrecorded link failure letter of each router is propagated in current failure region Breath;Each router is according to the mode in the newly-increased link failure information renewal current failure region received in step S3:When router previously no record fault message, newly-increased fault zone is obtained, using newly-increased fault zone as current event Hinder region;When router precedence record during fault message, newly-increased fault zone is obtained, by known fault region and newly-increased faulty section The union in domain is as current failure region.
- 2. a kind of link failure information transmission method applied to Mesh topology network-on-chips according to claim 1, its It is characterised by:The mode for propagating the also Unrecorded link failure information of each router in step S4 in current failure region is: Newly-increased link failure information is broadcast to the router in known fault region, known link fault message is broadcast to newly-increased event Hinder the router in region.
- 3. a kind of link failure information transmission method applied to Mesh topology network-on-chips according to claim 1, its It is characterised by:The faulty link interfered described in step S3 is failure chain adjacent in line direction or in a column direction Road.
- 4. a kind of link failure information transmission method applied to Mesh topology network-on-chips according to claim 1, its It is characterised by:Fault zone described in step S3 is:If only a row or column has faulty link, the row or column is defined as fault zone;It is continuous that if more row or column, which have faulty link and the row or column of faulty link be present,:Deposited per a line when consecutive rows of In faulty link, by the behavior fault zone between consecutive rows of initial row and end line, initial row and end line are defined as The border of fault zone;When each row of continuation column have link failure, by between the starting row and end column of continuation column Definition be fault zone, starting row and end column are defined as the border of fault zone;If there is faulty link and the row or column of faulty link be present to be discontinuous in more row or column:Then failure chain be present by all The union of the row and column on road is defined as fault zone.
- 5. a kind of link failure information transmission method applied to Mesh topology network-on-chips according to claim 1, its It is characterised by:Described known fault region and the acquisition pattern of the union of newly-increased fault zone are:To being in together on line direction Fault zone take union, union is taken to the fault zone being in together on column direction.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510436479.8A CN105024926B (en) | 2015-07-23 | 2015-07-23 | A kind of fault message transmission method applied to Mesh topology network-on-chips |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510436479.8A CN105024926B (en) | 2015-07-23 | 2015-07-23 | A kind of fault message transmission method applied to Mesh topology network-on-chips |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105024926A CN105024926A (en) | 2015-11-04 |
CN105024926B true CN105024926B (en) | 2018-02-27 |
Family
ID=54414648
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510436479.8A Expired - Fee Related CN105024926B (en) | 2015-07-23 | 2015-07-23 | A kind of fault message transmission method applied to Mesh topology network-on-chips |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105024926B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105591910B (en) * | 2016-03-04 | 2018-06-12 | 北京交通大学 | A kind of fault-tolerant Manhattan method for routing for route direction monotone variation network |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101834797A (en) * | 2010-05-06 | 2010-09-15 | 复旦大学 | Low-complexity and extensible fault-tolerant routing algorithm specific to network on chip |
CN104579951A (en) * | 2014-12-29 | 2015-04-29 | 合肥工业大学 | Fault-tolerance method in on-chip network under novel fault and congestion model |
-
2015
- 2015-07-23 CN CN201510436479.8A patent/CN105024926B/en not_active Expired - Fee Related
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101834797A (en) * | 2010-05-06 | 2010-09-15 | 复旦大学 | Low-complexity and extensible fault-tolerant routing algorithm specific to network on chip |
CN104579951A (en) * | 2014-12-29 | 2015-04-29 | 合肥工业大学 | Fault-tolerance method in on-chip network under novel fault and congestion model |
Non-Patent Citations (4)
Title |
---|
"Design of Fault-Tolerant and Reliable Networks-on-Chip";Junshi Wang等;《IEEE》;20150710;第545-550页 * |
"基于链路故障传递的保护方案设计与应用";韩亚雷等;《光通信技术》;20131215(第12期);第38-40页 * |
"片上网络无缓冲路由器关键技术研究";冯超超;《中国博士学位论文全文数据库信息科技辑》;20140415;第I136-25页 * |
"蜂窝网络上的虫孔容错路由算法";张伟文等;《重庆大学学报(自然科学版)》;20070415;第30卷(第4期);第102-107页 * |
Also Published As
Publication number | Publication date |
---|---|
CN105024926A (en) | 2015-11-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN101471693B (en) | Method and apparatus for sharing grid protection | |
Gopalan et al. | On identifying additive link metrics using linearly independent cycles and paths | |
US11489588B2 (en) | Method for supporting SNCP over packet network | |
US8614952B2 (en) | Efficient propagation of link state advertisements in densely interconnected OSPF networks | |
CN103685035B (en) | Route renewing method, interchanger and system | |
US9467332B2 (en) | Node failure detection for distributed linear protection | |
US9491097B2 (en) | Packet forwarding in a star stacking system | |
WO2016165142A1 (en) | Preserving method and device for virtual network | |
CN103067274A (en) | Spanning tree back-up port quick recovery method and device | |
CN108199955A (en) | A kind of method for routing foundation and device | |
CN103200097B (en) | A kind of route computing method and the network equipment | |
CN105024926B (en) | A kind of fault message transmission method applied to Mesh topology network-on-chips | |
CN110830394B (en) | Method for generating routing table based on RapidIO network | |
Taheri et al. | Advertiser elevator: A fault tolerant routing algorithm for partially connected 3D Network-on-Chips | |
CN107302500B (en) | Single-node fault protection method based on hop-by-hop mode | |
CN111092822A (en) | Method for balancing pw linear protection forwarding load of vpls | |
CN108667751A (en) | A kind of notifying method and device of Delay | |
CN106100722A (en) | A kind of virtual network survivability mapping method that can distinguish maximum sharing capability | |
WO2016161802A1 (en) | Method and system for realizing quick changeover of dni-pw | |
Karali et al. | Hybrid mocs for long-distance on-chip communications | |
CN110336744A (en) | A kind of fault tolerance rout ing method of area fault perception in wireless network-on-chip | |
Chakraborty et al. | A tree-based local repairing approach for increasing lifetime of query driven WSN | |
CN110213162A (en) | Fault-tolerant routing method for large-scale computer system | |
CN105704038B (en) | A kind of communication guarantee method for the electric power grid of guarantor temporarily | |
Zheng et al. | Cross-layer approach for minimizing routing disruption in IP networks |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180227 Termination date: 20180723 |
|
CF01 | Termination of patent right due to non-payment of annual fee |