An arbitration method, apparatus, device, storage medium, and computer program product

Abstract

The application discloses an arbitration method and device, equipment, a storage medium and a computer program product, and applies to the technical field of data transmission, and comprises the following steps: determining the maximum path congestion information based on the path congestion information corresponding to each resource request, and taking the routing path corresponding to the maximum path congestion information as a target routing path; determining the arbitration weight corresponding to each resource request based on the node fault condition of the target routing path corresponding to each resource request and the maximum path congestion information; and allocating resources to each resource request in turn according to the arbitration weight from large to small based on the arbitration weight corresponding to each resource request. The application considers the current node fault condition and the congestion information on the routing path when arbitrating, determines a weight value for each resource request based on the node fault condition and the congestion information, and finally allocates resources according to the weight value, thereby improving the arbitration accuracy and enabling the traffic to be more evenly distributed.

Description

Technical Field

[0001] This invention relates to the field of data transmission technology, and in particular to an arbitration method, apparatus, device, storage medium, and computer program product. Background Technology

[0002] Arbitrators are primarily designed to resolve the issue of multiple requests simultaneously vying for the same resource. They utilize an arbitration mechanism to strive for a fairer allocation of resources for each request, making them particularly suitable for devices like routers where information congestion and conflicts are highly probable. Traditional arbitrators arbitrate based on the results of the previous cycle or process resource requests according to a first-in, first-out (FIFO) principle. However, this approach suffers from technical issues regarding the inaccuracy of arbitration results.

[0003] It is evident that improving the accuracy of arbitration is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide an arbitration method, apparatus, device, storage medium and computer program product, which solves the technical problem of low arbitration accuracy in the prior art.

[0005] To address the aforementioned technical problems, this invention provides an arbitration method, comprising:

[0006] Determine the set of resource requests within the input port buffer;

[0007] A preset number of routing paths are determined for each resource request in the resource request set, and path congestion information for each routing path is determined based on the congestion information of each node on each routing path; wherein, the larger the value of the congestion information, the more idle the node.

[0008] Based on the path congestion information corresponding to each resource request, determine the maximum path congestion information and use the routing path corresponding to the maximum path congestion information as the target routing path.

[0009] Based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information, the arbitration weight corresponding to each resource request is determined.

[0010] Based on the arbitration weight corresponding to each resource request, resources are allocated to each resource request in descending order of arbitration weight.

[0011] On the one hand, a preset number of routing paths are determined for each resource request in the resource request set, and path congestion information for each routing path is determined based on the congestion information of each node on each routing path, including:

[0012] Determine the congestion information of each node on each routing path, and the difference between the target time and the current time corresponding to each congestion information;

[0013] Based on the difference, target weights are assigned to each node based on its congestion information. Based on the congestion information of each node and its corresponding target weight, path congestion information for each routing path is determined.

[0014] On the one hand, target weights are assigned to each node based on the difference in congestion information. Based on the congestion information of each node and its corresponding target weight, path congestion information for each routing path is determined, including:

[0015] Based on the target mapping table, the target weight corresponding to each difference is determined; wherein, the target mapping table is a table that includes the difference range and the weight mapping relationship.

[0016] On the one hand, based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information, the arbitration weight corresponding to each resource request is determined, including:

[0017] The arbitration weight for each resource request is determined based on the node failure status, the maximum path congestion information, and the previous arbitration result.

[0018] On the one hand, the arbitration weight for each resource request is determined based on the node failure status corresponding to each resource request, the maximum path congestion information, and the previous arbitration result, including:

[0019] Based on the previous arbitration result and the output direction corresponding to the previous resource request, determine the assignment corresponding to each output direction;

[0020] Based on the node failure status corresponding to the current resource request, the maximum path congestion information, and the value assigned to the output direction corresponding to the current resource request, the arbitration weight corresponding to the current resource request is determined.

[0021] On the one hand, based on the path congestion information corresponding to each resource request, the maximum path congestion information is determined, including:

[0022] If the path region corresponding to the current resource request is a 2×2 region, determine the intermediate nodes on each path;

[0023] Determine the intermediate node congestion information corresponding to the intermediate node, and take the largest intermediate node congestion information as the largest path congestion information.

[0024] On the one hand, when the congestion information of nodes on all routing paths is the same, the maximum path congestion information is determined based on the path congestion information corresponding to each resource request, including:

[0025] The congestion information of the shortest path corresponding to the current resource request is taken as the maximum path congestion information.

[0026] On the one hand, after determining the arbitration weight for each resource request based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information, the process also includes:

[0027] Determine whether the fault condition of the node corresponding to the node can be corrected;

[0028] When the fault of the node cannot be corrected, the path topology map corresponding to the current resource request is updated.

[0029] On the one hand, based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information, the arbitration weight corresponding to each resource request is determined, including:

[0030] Determine the predicted congestion information corresponding to each node, and based on the predicted congestion information corresponding to each node, determine the path predicted congestion information corresponding to each target routing path;

[0031] The arbitration weight for each resource request is determined based on the node failure status, the maximum path congestion information, and the path prediction congestion information.

[0032] On the one hand, before determining the predicted congestion information corresponding to each node, and before determining the path predicted congestion information corresponding to each target routing path based on the predicted congestion information corresponding to each node, the method further includes:

[0033] Analyze the characteristics of historical idle value changes based on the historical congestion information corresponding to each node;

[0034] The correlation characteristics of congestion information between nodes are determined based on the historical congestion information corresponding to each node.

[0035] A congestion information prediction model is trained based on the historical idle value change characteristics and the correlation characteristics of congestion information between nodes, so as to predict the predicted congestion information of each node based on the congestion information prediction model.

[0036] On the one hand, the aforementioned arbitration methods also include:

[0037] If the congestion information corresponding to the current node is 0, and the congestion information has not been updated for a preset maximum time, the current node will be designated as a faulty node.

[0038] This invention also provides an arbitration device, comprising:

[0039] The resource request set determination module is used to determine the resource request set in the input port buffer;

[0040] The path congestion information determination module is used to determine a preset number of routing paths corresponding to each resource request in the resource request set, and to determine the path congestion information corresponding to each routing path based on the congestion information of each node on each routing path; wherein, the larger the value of the congestion information, the more idle the node.

[0041] The maximum path congestion information determination module is used to determine the maximum path congestion information based on the path congestion information corresponding to each resource request, and to take the routing path corresponding to the maximum path congestion information as the target routing path.

[0042] The arbitration weight determination module is used to determine the arbitration weight corresponding to each resource request based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information.

[0043] The resource arbitration module is used to allocate resources to each resource request in descending order of the arbitration weight corresponding to each resource request.

[0044] This invention also provides an arbitration device, comprising:

[0045] Memory, used to store computer programs;

[0046] A processor for executing the computer program to implement the steps of the arbitration method described above.

[0047] This invention also provides a storage medium (i.e., a computer-readable storage medium) storing a computer program, which, when executed by a processor, implements the steps of the arbitration method described above.

[0048] This invention also provides a computer program product, including a computer program / instruction that, when executed by a processor, implements the steps of the arbitration method described above.

[0049] To address the aforementioned technical problems, embodiments of the present invention provide an arbitration method, which may include: determining a set of resource requests within an input port buffer; determining a preset number of routing paths corresponding to each resource request in the resource request set, and determining path congestion information corresponding to each routing path based on the congestion information of each node on each routing path; wherein, the larger the value of the congestion information, the more idle the node; determining the maximum path congestion information based on the path congestion information corresponding to each resource request, and using the routing path corresponding to the maximum path congestion information as the target routing path; determining the arbitration weight corresponding to each resource request based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information; and allocating resources to each resource request in descending order of arbitration weight based on the arbitration weight corresponding to each resource request.

[0050] As can be seen from the above technical solution, the beneficial effects of the present invention are as follows: Compared with the current arbitration of resource requests according to the first-in-first-out principle, the present invention considers the current node failure status and congestion information on the routing path during arbitration, determines a weight value for each resource request based on the node failure status and congestion information, and finally allocates resources according to the weight value. The present invention considers the overall network congestion situation for comprehensive evaluation during arbitration, thereby improving the accuracy of arbitration and enabling a more balanced distribution of traffic. Attached Figure Description

[0051] To more clearly illustrate the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0052] Figure 1 A flowchart of an arbitration method provided in an embodiment of the present invention;

[0053] Figure 2 A schematic diagram of a routing path provided for an embodiment of the present invention;

[0054] Figure 3 A schematic diagram of node congestion information provided in an embodiment of the present invention;

[0055] Figure 4 A schematic diagram of an arbitration framework provided in an embodiment of the present invention;

[0056] Figure 5 A schematic diagram of a router microarchitecture provided for an embodiment of the present invention;

[0057] Figure 6 A flowchart illustrating an arbitration method provided in an embodiment of the present invention;

[0058] Figure 7 A schematic diagram of the structural framework of an arbitration device provided in an embodiment of the present invention;

[0059] Figure 8 This is a schematic diagram of the structural framework of an arbitration device provided in an embodiment of the present invention. Detailed Implementation

[0060] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of the present invention.

[0061] The terms "comprising" and "having," and any variations thereof, in the specification and accompanying drawings of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may include steps or units not listed.

[0062] To enable those skilled in the art to better understand the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0063] Next, we will describe in detail an arbitration method provided by an embodiment of the present invention. Figure 1 A flowchart of an arbitration method provided in an embodiment of the present invention, the method may include:

[0064] S101, determine the set of resource requests in the input port buffer.

[0065] The steps in this embodiment can be performed by a designated electronic device, which may specifically be a server, portable terminal, node, or other form. The input port buffer in this embodiment is a buffer for receiving resource requests. The set of resource requests in this embodiment includes multiple resource requests.

[0066] S102, determine a preset number of routing paths corresponding to each resource request in the resource request set, and determine the path congestion information corresponding to each routing path based on the congestion information of each node on each routing path; wherein, the larger the value of the congestion information, the more idle the node.

[0067] In this embodiment, the multiple routing paths corresponding to each resource request are not the same. The routing paths in this embodiment consist of various nodes. The transmission of congestion information between nodes in this embodiment is achieved through credit backpressure; that is, when communication occurs between nodes, the credit is transmitted in reverse, transferring the node's flit (idle value) to the previous hop, such as... Figure 2 As shown, Figure 2 This is a schematic diagram of a routing path provided by an embodiment of the present invention. As shown, when node 1 sends a message to node 2, node 2 sends a credit message back to node 1, informing node 1 of node 2's credit, which is the free value (congestion information) of the flit (the amount of free space in the virtual channel within the buffer, measured by the size of the flit). The free flit calculation module involved here has an additional recording function compared to previous modules. It records the credit information transmitted from all nodes, forming a credit set, and continues to transmit it. When the message transmission reaches all nodes, it means that each node has a grasp of the credit information of all nodes, i.e., the congestion information. Each new message transmission automatically overwrites the old credit information, realizing the update of congestion information and ultimately achieving network-wide congestion awareness. However, during arbitration, only the congestion information of all nodes within the 3*3 area on the transmission path from the source node (node ​​1) to the destination node (node ​​11) is read. Figure 2 As shown in the dashed box, global information was not read. The reason for not using local congestion information perception is mainly that if congestion information of surrounding nodes were collected in real time, firstly, it would increase the waiting time, causing a delay in the source node arbitration; secondly, some nodes might not be exchanging messages at that moment and therefore could not provide congestion information, also causing a delay in arbitration or preventing arbitration altogether, indirectly increasing network overhead. In fact, this invention uses a combination of global congestion perception and local congestion information calculation. Global congestion perception and arbitration work in parallel and do not affect each other. During arbitration, only the congestion information of the required nodes is read. This embodiment does not limit the specific method for determining the path congestion information corresponding to each routing path. For example, this embodiment can directly add up all the congestion information on the path to determine the path congestion information, or it can determine the weight of each path node based on set rules, thereby using the weight of the node's congestion information as the weight, and thus determining the path congestion information based on the weights and congestion information of each node. For example, determine if any node on any possible path within a 3x3 area downstream of the input port buffer has a `flit` idle value of 0, indicating node congestion. If so, the routing path corresponding to the node with a `flit` idle value of 0 needs to be excluded from the feasible paths. Assume... Figure 2In this context, node 1 is the source node and node 11 is the destination node. The possible paths from node 1 to node 11 are node 1-2-7-10-11, 1-2-8-9-11, 1-2-8-10-11, 1-5-8-10-11, 1-5-8-9-11, and 1-5-6-9-11. Since the flit idle value of node 7 is 0, the path 1-2-7-10-11 is discarded. The sum of the flit idle values ​​for the remaining paths is used as the path congestion information for each routing path.

[0068] It should be further explained that, in order to improve the accuracy of path congestion information determination, the above-mentioned determination of a preset number of routing paths corresponding to each resource request in the resource request set, and the determination of path congestion information corresponding to each routing path based on the congestion information of each node on each routing path, may include:

[0069] S1021, determine the congestion information of each node on each routing path, and the difference between the target time and the current time corresponding to each congestion information.

[0070] In this embodiment, congestion information is primarily based on the flit idle value (congestion information) of the input port buffer within the node. Each input buffer can be configured with 4 virtual channels, each virtual channel being 4 flits in size, for a total of 16 flits. A larger flit idle value indicates that the node is not congested in that direction, while a smaller value indicates varying degrees of congestion. The node will adjust the priority of response requests accordingly. When the flit idle value is 0, it means congestion has occurred, and the node's request will remain in place, re-applying for resources in the next round to alleviate congestion. Alternatively, an adaptive routing algorithm can be used to replan the route path for resource requests, such as bypassing congested nodes. It is understood that each message with a flit idle value has a timestamp recording the message's sending time, and when reading the flit idle value, it is compared with the current time.

[0071] S1022, assign target weights to each node based on the difference in congestion information, and determine the path congestion information for each routing path based on the congestion information of each node and its corresponding target weight.

[0072] This embodiment can determine the weight corresponding to each difference based on the difference, and then use this weight as the target weight for congestion information. This allows for the determination of the actual congestion information of nodes based on the differences and congestion information, and the determination of path congestion information based on the actual congestion information of each node on the routing path. When determining path congestion information, this embodiment considers the time interval between the congestion information and the current time. This is because the Flit idle value is passed back through credit after a message enters a node. If two nodes have no message exchange for a long time, there may be a lag in the Flit idle value message. Adding a target weight here helps to mitigate the impact of message lag on the Flit idle value, thereby improving the accuracy of path congestion information determination.

[0073] It should be further explained that, in order to improve the accuracy of target weight determination, the above-mentioned allocation of target weights based on the difference value of each node’s congestion information, and determination of the path congestion information corresponding to each routing path based on the congestion information of each node and its corresponding target weight, may include: determining the target weight corresponding to each difference value based on the target mapping table; wherein, the target mapping table is a table that includes the difference range and the weight mapping relationship.

[0074] In this embodiment, the mapping table is a table that includes the weight mapping relationships corresponding to each difference range. For example, if the target time corresponding to the congestion information is less than 20 time points from the current time, then a (target weight) = 1; if the difference is greater than 20 time points but less than 50 time points, then a is 90-80%; if the difference is greater than 50 time points but less than 100 time points, a is 60-50%; and if the difference is greater than 100 time points, a is 20-10%. Figure 3 As shown, Figure 3 This is a schematic diagram of node congestion information provided in an embodiment of the present invention. The small numbers within each node indicate the idle flit value of each input buffer. At this time, the idle flit values ​​of all nodes within the 3*3 area are concentrated in the idle flit calculation module of node 1. Based on the difference between the time recorded upon entry (target time) and the current time, a suitable target weight parameter 'a' is selected. For example, the idle flit value of node 2 on the path node 1-2-8-9-11 has not been updated for 30 time intervals, so 'a' is 80%; the idle flit value of node 8 has not been updated for 70 time intervals, so 'a' is 50%; and the idle flit value of node 9 has not been updated for 110 time intervals, so 'a' is 10%. The path congestion information C at this time is C = 10*80% + 10*50% + 10*10% = 14. This process is repeated for all possible paths to calculate C. When path congestion information is determined based on the difference, the path congestion information C can be expressed as: The mapping relationship in this embodiment is the optimal mapping relationship obtained through experiments. The method of determining the target weight based on the mapping relationship improves the accuracy of the target weight determination.

[0075] S103: Based on the path congestion information corresponding to each resource request, determine the largest path congestion information and use the routing path corresponding to the largest path congestion information as the target routing path.

[0076] This embodiment determines the path congestion information corresponding to the routing path for each resource request, and then uses the routing path with the largest path congestion information as the target routing path.

[0077] It should be further explained that, based on any of the above embodiments, determining the maximum path congestion information from the path congestion information corresponding to each resource request may include: if the path region corresponding to the current resource request is a 2×2 region, determining the intermediate nodes on each path; determining the intermediate node congestion information corresponding to the intermediate nodes, and using the maximum intermediate node congestion information as the maximum path congestion information. When the path region corresponding to the resource request is 2×2, it is only necessary to compare the flit idle values ​​of the intermediate nodes, while also taking into account the issue of congestion information lag. This embodiment can reduce the congestion information of the nodes participating in the calculation, thereby improving the efficiency of congestion information calculation.

[0078] It should be further explained that, based on any of the above embodiments, when the congestion information of nodes on all routing paths is the same, determining the maximum path congestion information based on the path congestion information corresponding to each resource request can include: using the congestion information of the shortest path corresponding to the current resource request as the maximum path congestion information. Since both are on the same dimension, if there are no faults or congestion on the path, it can be directly transmitted to the destination node according to the shortest path principle. This embodiment improves the efficiency of calculating the maximum path congestion information.

[0079] S104. Based on the node failure status and maximum path congestion information of the target routing path corresponding to each resource request, determine the arbitration weight corresponding to each resource request.

[0080] In this embodiment, the arbitration weight for each resource request is determined based on both the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information. The node failure status in this embodiment is primarily determined by the fault detection / calculation module in the router. This module sends a probe signal; if no response is received within a certain time (e.g., after 110 seconds), the node is considered faulty. This module also has the function of receiving signals and providing feedback to neighboring nodes. The module monitors the changes in the `flit` idle value within the node's buffer. If the `flit` idle value changes across the four input ports, it returns a signal; the faster the change, the faster the feedback. If the `flit` idle value remains unchanged for a long period, it waits indefinitely without providing feedback to other nodes. After a certain time, other nodes can consider this node to be faulty. If the `flit` idle value is 0 and remains unchanged for a long period, it may also indicate congestion. This scheme treats this situation as a fault and incorporates this result into the arbitration weight determination method. It should be noted that a node failure on the routing path has a veto power over the arbitration result; if a failure occurs, the arbitration weight corresponding to that routing path is 0. In this embodiment, the node fault request can be the fault status of the next-hop node in the input port buffer, or the fault status of all downstream nodes. That is, if the congestion information corresponding to the current node is 0, and the congestion information has not been updated for a preset maximum time, the current node is considered a faulty node. This embodiment of the invention improves the accuracy of fault status determination.

[0081] It should be further explained that, based on any of the above embodiments, in order to improve the accuracy of the arbitration weight determination, the above determination of the arbitration weight corresponding to each resource request based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information may include: determining the arbitration weight corresponding to each resource request based on the node failure status of each resource request, the maximum path congestion information and the previous arbitration result.

[0082] In this embodiment, the previous arbitration result can refer to the arbitration weights of the previous resource requests being sorted from largest to smallest. Based on this sorting result and the exit direction corresponding to the previous resource request, a value is assigned to each exit direction, and this assigned value becomes the previous arbitration result. Alternatively, in this embodiment, the previous arbitration result can be used when processing all previous resource requests. If the resource becomes unavailable at the last round, the previous resource request is moved to this round, thus increasing its processing weight based on the previous arbitration result. This embodiment also considers the previous arbitration result during arbitration, thereby improving the accuracy of determining the arbitration weights.

[0083] It should be further explained that, to ensure the accuracy of determining the arbitration weight based on the previous arbitration result, the aforementioned determination of the arbitration weight for each resource request based on the node failure status, maximum path congestion information, and the previous arbitration result may include: determining the assignment corresponding to each output direction based on the previous arbitration result and the output direction corresponding to the previous resource request; and determining the arbitration weight corresponding to the current resource request based on the node failure status, maximum path congestion information, and the assignment of the output direction corresponding to the current resource request. The arbitration framework corresponding to this embodiment is shown in [link to relevant documentation]. Figure 4 , Figure 4 This is a schematic diagram of an arbitration framework provided in an embodiment of the present invention. Figure 4 As the arbitration strategy framework indicates, a next-hop routing failure (i.e., a failure of a downstream node in the routing path) has a veto power over the arbitration result. If a failure occurs, the request to acquire resources is halted, and the process awaits the next round of arbitration to alleviate congestion, or a reconfigurable adaptive routing algorithm is used to select an alternative transmission path. In this arbitration framework, the key factor is still the downstream node congestion situation (path congestion information). The previous round of arbitration results, while providing supplementary information on downstream node congestion, should have a reduced weight in the formula.

[0084] Understandably, routers typically have five ports for incoming requests, such as... Figure 5 As shown, Figure 5This invention provides a schematic diagram of a router microarchitecture. Compared to traditional router designs, the router microarchitecture of this invention includes two additional modules: a fault detection / calculation module and a flit idle value calculation module. These two modules are primarily responsible for transmitting fault and congestion information to surrounding nodes, recording and performing simple calculations, and providing information for arbitration of resource requests within the input buffers (East, West, South, North, and Local). Output ports do not have buffers; resource requests in the input buffers (input port buffers) are directly allocated to the corresponding output ports by the crossbar switch. The crossbar switch consists of 5 input ports (East, West, South, North, and Local) and 5 output ports (East, West, South, North, and Local). The resource request ports are East, West, South, North, and Local (flit input0 ~ 4). Some routers located in corners or on the edge may only have 3 or 4 input ports. This case only uses the common 5-port example for illustration; routers with 3 or 4 ports can handle the situation as if a certain direction input port is missing. If the destination of an incoming input (East, West, South, North input port) request is local, its priority is set to the highest at that moment, and it is forcibly processed in FIFO (First-In-First-Out) order with the output order of East, West, South, North, etc., to release resources as quickly as possible. If the request is a local input, the arbitration strategy and weight calculation formula are treated the same as for incoming input requests. The arbitration results of each input buffer are independent, and the arbitration in this round will refer to the arbitration results of the same output direction in the previous round within the same input buffer. The output direction of the request is obtained by reading the message header flit from the RC module (routing calculation module) when the message enters the router. The process of determining the arbitration weight in this embodiment of the invention may include: comparing the path congestion information of all possible routing paths for each pair of source nodes and destination nodes, and substituting the largest path congestion information Cmax into the arbitration formula to participate in the calculation of the arbitration weight Sd. The weight coefficient Sd calculation formula can be expressed as: In Sd, d represents a certain output direction, representing e, w, s, n (east, west, south, north). After completing one round of arbitration calculation, d is replaced accordingly, representing the arbitration result in a certain direction at that moment; r indicates whether there is a fault in the nodes on the possible path, i.e., r = 1 when the next node of the source node has no faults, and r = 0 when there are faults; Sld represents the value of the previous arbitration result of this node after reassignment, i.e., the previous arbitration result Sd is sorted from largest to smallest, and only the first occurrence of Sd in different directions is assigned a value to form Sld, which is used to prepare for the next arbitration. For example, in the previous arbitration, there were 5 requests in the south input port buffer. After arbitration, their output order is east, west, north, and south. Then the first occurrence of the eastward output result Se and the first occurrence of the westward output result Sw are assigned the values ​​1 and 2 respectively, and the northward output result Sn is assigned the value 3. This can be understood as assigning a value only once for the same output direction. If a direction that does not appear is needed, it will be forcibly assigned the values ​​4, 5, 6, and 7 in the order of east, south, west, and north respectively. For example, if there are 4 requests in the south input port buffer of the previous arbitration round, and the arbitration result output order is east, west, east, and west respectively, then in this round of arbitration, the Sle corresponding to the east-facing output Se will be assigned the value 1, and the Slw corresponding to the west-facing output Sw will be assigned the value 2. Since there cannot be a south-facing output (here, the south input port), only the Sln parameter corresponding to the north-facing output is missing. According to the above rules, the north-facing parameter Sln in this round of arbitration is defined as 7. Because Sld plays an auxiliary role in the formula to prevent the weight values ​​of each direction from being the same, a coefficient of 11% is added, which is an empirical value. Within a certain input buffer, if there are multiple requests for the same target node in a certain round, resources are allocated according to the FIFO principle. Within the cross switch, if there are multiple requests for the same target node, resources are allocated sequentially according to the FIFO principle. If the results are still the same, resources are forcibly allocated according to the input port order of east, west, south, and north. Sld needs a table to record the arbitration results of the previous round. Finally, Sd for each direction in this round is calculated and labeled as Se, Sw, Ss, Sn, and resources are obtained in descending order of Sd value.

[0085] It should be further explained that, in order to improve the accuracy of subsequent path congestion information calculation, after determining the arbitration weight corresponding to each resource request based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information, it may also include: Step 1: Determine whether the node failure status corresponding to the node can be corrected; Step 2: When the node failure cannot be corrected, determine to update the path topology map corresponding to the current resource request.

[0086] In this embodiment, when a node's failure cannot be corrected, the path topology graph is updated promptly. This allows the next round of arbitration to be based directly on the updated topology graph, without considering the failed node. In this embodiment, when a node is correctable, it is corrected directly using the corresponding correction method.

[0087] It should be further explained that, based on any of the above embodiments, in order to improve the accuracy of the arbitration weight determination, the determination of the arbitration weight corresponding to each resource request based on the node failure status and the maximum path congestion information of the target routing path corresponding to each resource request may include: determining the predicted congestion information corresponding to each node; determining the path predicted congestion information corresponding to each target routing path based on the predicted congestion information corresponding to each node; and determining the arbitration weight corresponding to each resource request based on the node failure status, the maximum path congestion information, and the path predicted congestion information. This embodiment does not limit the method for determining the predicted congestion information corresponding to the target path of each resource request. For example, this embodiment may use a spatiotemporal convolutional prediction method to predict the predicted congestion information; or this embodiment may use a path entropy weight evaluation method to determine the predicted congestion information; or this embodiment may use a reinforcement learning dynamic prediction method to predict the predicted congestion information. In this embodiment, when determining the arbitration weight, the predicted congestion information is considered, thereby correcting the maximum path congestion information based on the predicted congestion information, improving the accuracy of path congestion determination, and thus improving the accuracy of arbitration weight determination.

[0088] It should be further explained that, based on any of the above embodiments, before determining the predicted congestion information corresponding to each node and the path predicted congestion information corresponding to each target routing path based on the predicted congestion information corresponding to each node, the method further includes: analyzing the historical idle value change characteristics based on the historical congestion information corresponding to each node; determining the correlation characteristics of congestion information between nodes based on the historical congestion information corresponding to each node; training a congestion information prediction model based on the historical idle value change characteristics and the correlation characteristics of congestion information between nodes, so as to predict the predicted congestion information of each node based on the congestion information prediction model. This embodiment analyzes the linear changes in the congestion information of each node to obtain the historical idle value change characteristics, that is, this embodiment can analyze the historical congestion information change patterns of each node and identify periodic fluctuation characteristics. Based on the correlation between nodes, the idle value correlation is determined, and thus a congestion information prediction model is constructed based on the correlation of idle values ​​between nodes and the linear changes in congestion information, thereby predicting congestion information based on the congestion information prediction model. That is, this embodiment establishes a gridded topology graph between nodes to capture the state linkage effect between adjacent nodes. When a node shows a congestion trend, the chain reaction of its surrounding nodes is predicted. This embodiment can train a machine learning algorithm based on historical idle value change characteristics, idle value correlation between nodes, and corresponding congestion information to obtain a congestion information prediction model.

[0089] It should be further explained that, in order to improve the accuracy of prediction, this embodiment can also consider nonlinear features. In addition to the linear features of congestion information changes of nodes determined based on historical congestion information and the correlation features of congestion information between nodes, it is also necessary to consider the nonlinear features of congestion information changes, thereby improving the accuracy of congestion information prediction (at this time, the congestion information prediction model is trained based on the historical idle value change features and the correlation features of congestion information between nodes, including: training the congestion information prediction model based on the historical idle value change features, the correlation features of congestion information between nodes and historical nonlinear features, so that the congestion information prediction model can predict nonlinear changes). This embodiment can set weights for predicting congestion information based on the impact of predicting congestion information on improving arbitration accuracy.

[0090] S105, based on the arbitration weight corresponding to each resource request, resources are allocated to each resource request in descending order of arbitration weight.

[0091] This embodiment considers external congestion during arbitration, thereby improving the accuracy of the arbitration. In this embodiment, different resource requests correspond to different exit points, resulting in different directions. Therefore, the direction corresponding to each resource request can be determined, and the arbitration weight for each direction can be established. This allows for the allocation of resources to resource requests in different directions.

[0092] An arbitration method provided by this invention may include: S101, determining a set of resource requests within an input port buffer; S102, determining a preset number of routing paths corresponding to each resource request in the resource request set, and determining path congestion information corresponding to each routing path based on the congestion information of each node on each routing path; wherein, the larger the value of the congestion information, the more idle the node; S103, determining the largest path congestion information based on the path congestion information corresponding to each resource request, and using the routing path corresponding to the largest path congestion information as the target routing path; S104, determining the arbitration weight corresponding to each resource request based on the node failure status of the target routing path corresponding to each resource request and the largest path congestion information; S105, allocating resources to each resource request in descending order of arbitration weight based on the arbitration weight corresponding to each resource request. Compared to the current first-in-first-out arbitration of resource requests, this invention considers the current node failure status and congestion information on the routing path (i.e., external information) during arbitration. Based on the node failure status and path congestion information, an arbitration weight is determined for each resource request. Finally, resources are allocated according to the weight value, thereby improving the accuracy of arbitration and enabling a more balanced distribution of traffic.

[0093] For a clearer understanding of this invention, please refer to the following details. Figure 6 , Figure 6A flowchart illustrating an arbitration method provided in this embodiment of the invention may specifically include:

[0094] S201, Get the set of resource requests in the port buffer.

[0095] S202, determine a preset number of routing paths corresponding to each resource request in the resource request set; wherein the preset number is greater than 1.

[0096] S203, based on the idle value of the nodes on the routing path corresponding to each resource request, determine the congestion information corresponding to each node, and determine the target time corresponding to each idle value.

[0097] S204, determine the weight of each congestion information based on the difference between the target time and the current time for each idle value, determine the actual congestion information corresponding to each node based on the weight and the congestion information, and determine the path congestion information of the routing path corresponding to each resource request based on the actual congestion information.

[0098] S205: Based on the path congestion information corresponding to each resource request, determine the largest path congestion information and use the routing path corresponding to the largest path congestion information as the target routing path.

[0099] S206, determine the assignment corresponding to the output direction of each resource request based on the ranking result of the previous round of arbitration weights.

[0100] S207. The arbitration weight for each resource request is determined based on the node failure status of the target routing path corresponding to each resource request, the maximum path congestion information, and the assignment corresponding to the same output direction.

[0101] The arbitration strategy of this invention mainly takes the congestion status of all possible paths as a reference, and uses the failure status of the routing path and the round-robin arbitration results in the same direction of the node as auxiliary factors to obtain the arbitration weight.

[0102] S208: Based on the arbitration weight corresponding to each resource request, resources are allocated to each resource request in descending order of arbitration weight.

[0103] This invention incorporates congestion information from surrounding nodes into the arbitration process, helping the network achieve load balancing as much as possible, improving information transmission speed while reducing overload power consumption, and preventing congestion and failures. Furthermore, to support the arbitrator, a fault detection / calculation module (to determine node failure status) and an idle flit calculation module (to determine congestion information) can be added to the router. These two modules can respectively detect faults in neighboring routing nodes and aggregate their congestion information, ultimately helping the on-chip network achieve more balanced load and improved overall performance.

[0104] The arbitration device provided in the embodiments of the present invention will be described below. The arbitration device described below can be referred to in correspondence with the arbitration method described above.

[0105] Figure 7 A schematic diagram of the structural framework of an arbitration device provided in an embodiment of the present invention may include:

[0106] Resource request set determination module 100 is used to determine the resource request set in the input port buffer;

[0107] The path congestion information determination module 200 is used to determine a preset number of routing paths corresponding to each resource request in the resource request set, and to determine the path congestion information corresponding to each routing path based on the congestion information of each node on each routing path; wherein, the larger the value of the congestion information, the more idle the node.

[0108] The maximum path congestion information determination module 300 is used to determine the maximum path congestion information based on the path congestion information corresponding to each resource request, and to take the routing path corresponding to the maximum path congestion information as the target routing path.

[0109] The arbitration weight determination module 400 is used to determine the arbitration weight corresponding to each resource request based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information.

[0110] The resource arbitration module 500 is used to allocate resources to each resource request in descending order of the arbitration weight corresponding to each resource request.

[0111] Furthermore, based on the above embodiments, the path congestion information determination module 200 may include:

[0112] The difference determination module is used to determine the congestion information of each node on each routing path, as well as the difference between the target time and the current time corresponding to each congestion information.

[0113] The path congestion information determination unit is used to allocate target weights for each node based on the difference, and to determine the path congestion information corresponding to each routing path based on the congestion information of each node and its corresponding target weight.

[0114] Furthermore, based on any of the above embodiments, the path congestion information determination unit includes:

[0115] The target weight determination unit is used to determine the target weight corresponding to each difference based on the target mapping table; wherein the target mapping table is a table that includes the difference range and the weight mapping relationship.

[0116] Furthermore, based on any of the above embodiments, the arbitration weight determination module 400 may include:

[0117] The arbitration weight determination unit is used to determine the arbitration weight corresponding to each resource request based on the node failure status corresponding to each resource request, the maximum path congestion information, and the previous arbitration result.

[0118] Furthermore, based on the above embodiments, the arbitration weight determination unit may include:

[0119] The assignment unit corresponding to the output direction is used to determine the assignment corresponding to each output direction based on the previous arbitration result and the output direction corresponding to the previous resource request.

[0120] The arbitration weight determination unit is used to determine the arbitration weight corresponding to the current resource request based on the node failure status corresponding to the current resource request, the maximum path congestion information, and the assignment of the output direction corresponding to the current resource request.

[0121] Furthermore, based on any of the above embodiments, the maximum path congestion information determination module 300 may include:

[0122] The intermediate node determination unit is used to determine the intermediate node on each path when the path area corresponding to the current resource request is a 2×2 area.

[0123] The first maximum path congestion information determination unit is used to determine the intermediate node congestion information corresponding to the intermediate node, and to take the maximum intermediate node congestion information as the maximum path congestion information.

[0124] Furthermore, based on any of the above embodiments, the maximum path congestion information determination module 300 includes:

[0125] The second largest path congestion information determination unit is used to determine the congestion information of the shortest path corresponding to the current resource request as the largest path congestion information.

[0126] Furthermore, based on any of the above embodiments, the arbitration device may further include:

[0127] The judgment module is used to determine whether the fault condition of the node corresponding to the node can be corrected.

[0128] The path topology update module is used to determine and update the path topology corresponding to the current resource request when the fault of the node cannot be corrected.

[0129] Furthermore, based on any of the above embodiments, the arbitration weight determination module 400 may include:

[0130] The path prediction congestion information determination unit is used to determine the prediction congestion information corresponding to each node, and to determine the path prediction congestion information corresponding to each target routing path based on the prediction congestion information corresponding to each node.

[0131] The arbitration weight determination unit based on path prediction congestion information is used to determine the arbitration weight corresponding to each resource request based on the node failure status, the maximum path congestion information, and the path prediction congestion information.

[0132] Furthermore, based on any of the above embodiments, the arbitration device may further include:

[0133] The historical idle value change characteristic determination module is used to analyze the historical idle value change characteristics based on the historical congestion information corresponding to each node.

[0134] The association feature determination module is used to determine the association features of congestion information between nodes based on the historical congestion information corresponding to each node.

[0135] The congestion information prediction model training module is used to train a congestion information prediction model based on the historical idle value change characteristics and the correlation characteristics of congestion information between nodes, so as to predict the predicted congestion information of each node based on the congestion information prediction model.

[0136] Furthermore, based on any of the above embodiments, the arbitration device may further include:

[0137] The fault node determination module is used to identify the current node as a fault node if the congestion information corresponding to the current node is 0 and the congestion information has not been updated for a preset maximum time.

[0138] It should be noted that the order of the modules and units in the above arbitration device can be changed without affecting the logic.

[0139] Figure 7 For a description of the features in the corresponding embodiments, please refer to Figure 7 The relevant descriptions of the corresponding embodiments will not be repeated here.

[0140] An arbitration device provided in this embodiment of the invention may include: a resource request set determination module 100, used to determine a set of resource requests in an input port buffer; a path congestion information determination module 200, used to determine a preset number of routing paths corresponding to each resource request in the resource request set, and to determine path congestion information corresponding to each routing path based on the congestion information of each node on each routing path; wherein, the larger the value of the congestion information, the more idle the node; a maximum path congestion information determination module 300, used to determine the maximum path congestion information based on the path congestion information corresponding to each resource request, and to take the routing path corresponding to the maximum path congestion information as the target routing path; an arbitration weight determination module 400, used to determine the arbitration weight corresponding to each resource request based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information; and a resource arbitration module 500, used to allocate resources to each resource request in descending order of arbitration weight according to the arbitration weight corresponding to each resource request. Compared to the current first-in-first-out arbitration of resource requests, this invention considers the current node failure status and congestion information on the routing path (i.e., external information) during arbitration. Based on the node failure status and path congestion information, an arbitration weight is determined for each resource request. Finally, resources are allocated according to the weight value, thereby improving the accuracy of arbitration and enabling a more balanced distribution of traffic.

[0141] The following describes an arbitration device provided by an embodiment of the present invention. The arbitration device described below can be referred to in correspondence with the arbitration method described above.

[0142] Figure 8 This is a schematic diagram of the structural framework of an arbitration device provided in an embodiment of the present invention, as shown below. Figure 8 As shown, the arbitration device includes: a memory 60 for storing computer programs;

[0143] The processor 61 is used to implement the steps of the arbitration method as described in the above embodiments when executing a computer program.

[0144] The arbitration equipment provided in this embodiment may include, but is not limited to, smartphones, tablets, laptops, or desktop computers.

[0145] The processor 61 may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor 61 may be implemented using at least one hardware form selected from Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), and Programmable Logic Array (PLA). The processor 61 may also include a main processor and a coprocessor. The main processor, also known as the Central Processing Unit (CPU), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor 61 may integrate a Graphics Processing Unit (GPU), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor 61 may also include an Artificial Intelligence (AI) processor, which handles computational operations related to machine learning.

[0146] The memory 60 may include one or more computer-readable storage media, which may be non-transitory. The memory 60 may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In this embodiment, the memory 60 is used to store at least the following computer program 601, which, after being loaded and executed by the processor 61, is capable of implementing the relevant steps of the arbitration method disclosed in any of the foregoing embodiments. In addition, the resources stored in the memory 60 may also include an operating system 602 and data 603, and the storage method may be temporary or permanent storage. The operating system 602 may include Windows, Unix, Linux, etc. The data 603 may include, but is not limited to, data required by the arbitration method.

[0147] In some embodiments, the arbitration device may further include a display screen 62, an input / output interface 63, a communication interface 64, a power supply 65, and a communication bus 66.

[0148] Those skilled in the art will understand that Figure 8 The structure shown does not constitute a limitation on the arbitration device and may include more or fewer components than illustrated.

[0149] It is understood that if the arbitration method in the above embodiments is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the current technology, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and executes all or part of the steps of the methods in the various embodiments of the present invention. The aforementioned storage medium includes: USB flash drive, mobile hard drive, read-only memory (ROM), random access memory (RAM), electrically erasable programmable ROM, register, hard disk, removable disk, CD-ROM, magnetic disk or optical disk, and other media capable of storing program code.

[0150] Based on this, embodiments of the present invention also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the arbitration method described above.

[0151] The above provides a detailed description of an arbitration method provided by an embodiment of the present invention. The various embodiments are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0152] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0153] The above provides a detailed description of the arbitration method, apparatus, device, storage medium, and computer program product provided by this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and core ideas of this invention. It should be noted that those skilled in the art can make various improvements and modifications to this invention without departing from its principles, and these improvements and modifications also fall within the protection scope of this invention.

Claims

1. An arbitration method, characterized in that, include: Determine the set of resource requests within the input port buffer; Determine a preset number of routing paths corresponding to each resource request in the resource request set, determine the congestion information of each node on each routing path, and the difference between the target time and the current time corresponding to each congestion information; Based on the difference, target weights are assigned to each node based on its congestion information. Based on the congestion information of each node and its corresponding target weight, path congestion information for each routing path is determined. The larger the value of the congestion information, the more idle the node. Based on the path congestion information corresponding to each resource request, determine the maximum path congestion information and use the routing path corresponding to the maximum path congestion information as the target routing path. Based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information, the arbitration weight corresponding to each resource request is determined; if a node in the target routing path fails, the arbitration weight is 0. Based on the arbitration weight corresponding to each resource request, resources are allocated to each resource request in descending order of arbitration weight; Specifically, based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information, the arbitration weight corresponding to each resource request is determined, including: The arbitration weight for each resource request is determined based on the node failure status corresponding to each resource request, the maximum path congestion information, and the previous arbitration result. Alternatively, determine the predicted congestion information corresponding to each node, and determine the path predicted congestion information corresponding to each target routing path based on the predicted congestion information corresponding to each node; determine the arbitration weight corresponding to each resource request based on the node failure status, the maximum path congestion information, and the path predicted congestion information.

2. The arbitration method according to claim 1, characterized in that, Based on the difference, target weights are assigned to each node based on its congestion information. Based on the congestion information of each node and its corresponding target weight, path congestion information for each routing path is determined, including: Based on the target mapping table, the target weight corresponding to each difference is determined; wherein, the target mapping table is a table that includes the difference range and the weight mapping relationship.

3. The arbitration method according to claim 1, characterized in that, The arbitration weight for each resource request is determined based on the node failure status corresponding to each resource request, the maximum path congestion information, and the previous arbitration result, including: Based on the previous arbitration result and the output direction corresponding to the previous resource request, determine the assignment corresponding to each output direction; Based on the node failure status corresponding to the current resource request, the maximum path congestion information, and the value assigned to the output direction corresponding to the current resource request, the arbitration weight corresponding to the current resource request is determined.

4. The arbitration method according to claim 1, characterized in that, Based on the path congestion information corresponding to each resource request, the maximum path congestion information is determined, including: If the path region corresponding to the current resource request is a 2×2 region, determine the intermediate nodes on each path; Determine the intermediate node congestion information corresponding to the intermediate node, and take the largest intermediate node congestion information as the largest path congestion information.

5. The arbitration method according to claim 1, characterized in that, When the congestion information of nodes on all routing paths is the same, the maximum path congestion information is determined based on the path congestion information corresponding to each resource request, including: The congestion information of the shortest path corresponding to the current resource request is taken as the maximum path congestion information.

6. The arbitration method according to claim 1, characterized in that, After determining the arbitration weight for each resource request based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information, the method further includes: Determine whether the fault condition of the node corresponding to the node can be corrected; When the fault of the node cannot be corrected, the path topology map corresponding to the current resource request is updated.

7. The arbitration method according to claim 1, characterized in that, Before determining the predicted congestion information corresponding to each node, and before determining the path predicted congestion information corresponding to each target routing path based on the predicted congestion information corresponding to each node, the process also includes: Analyze the characteristics of historical idle value changes based on the historical congestion information corresponding to each node; The correlation characteristics of congestion information between nodes are determined based on the historical congestion information corresponding to each node. A congestion information prediction model is trained based on the historical idle value change characteristics and the correlation characteristics of congestion information between nodes, so as to predict the predicted congestion information of each node based on the congestion information prediction model.

8. The arbitration method according to claim 1, characterized in that, Also includes: If the congestion information corresponding to the current node is 0, and the congestion information has not been updated for a preset maximum time, the current node will be designated as a faulty node.

9. An arbitration device, characterized in that, include: The resource request set determination module is used to determine the resource request set in the input port buffer; A path congestion information determination module is used to determine a preset number of routing paths corresponding to each resource request in the resource request set, and to determine the path congestion information corresponding to each routing path based on the congestion information of each node on each routing path; wherein, the larger the value of the congestion information, the more idle the node; wherein, the path congestion information determination module includes: The difference determination module is used to determine the congestion information of each node on each routing path, as well as the difference between the target time and the current time corresponding to each congestion information. The path congestion information determination unit is used to allocate target weights based on the difference as the congestion information of each node, and to determine the path congestion information corresponding to each routing path based on the congestion information of each node and its corresponding target weight. The maximum path congestion information determination module is used to determine the maximum path congestion information based on the path congestion information corresponding to each resource request, and to take the routing path corresponding to the maximum path congestion information as the target routing path. The arbitration weight determination module is used to determine the arbitration weight corresponding to each resource request based on the node failure status of the target routing path corresponding to each resource request and the maximum path congestion information; if a node in the target routing path fails, the arbitration weight is 0. The resource arbitration module is used to allocate resources to each resource request in descending order of the arbitration weight corresponding to each resource request. The arbitration weight determination module includes: The arbitration weight determination unit is used to determine the arbitration weight corresponding to each resource request based on the node failure status corresponding to each resource request, the maximum path congestion information, and the previous arbitration result. Alternatively, the arbitration weighting determination module includes: The path prediction congestion information determination unit is used to determine the prediction congestion information corresponding to each node, and to determine the path prediction congestion information corresponding to each target routing path based on the prediction congestion information corresponding to each node. The arbitration weight determination unit based on path prediction congestion information is used to determine the arbitration weight corresponding to each resource request based on the node failure status, the maximum path congestion information, and the path prediction congestion information.

10. An arbitration device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the steps of the arbitration method as described in any one of claims 1 to 8.

11. A storage medium, characterized in that, The storage medium stores a computer program that, when executed by a processor, implements the steps of the arbitration method as described in any one of claims 1 to 8.

12. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instruction is executed by the processor, it implements the steps of the arbitration method according to any one of claims 1 to 8.

Images