CN114860511A

CN114860511A - Data processing method and device, chip, electronic equipment and medium

Info

Publication number: CN114860511A
Application number: CN202210473209.4A
Authority: CN
Inventors: 冷祥纶; 李冰; 赵月新; 王海生
Original assignee: Shanghai Power Tensors Intelligent Technology Co Ltd
Current assignee: Shanghai Power Tensors Intelligent Technology Co Ltd
Priority date: 2022-04-29
Filing date: 2022-04-29
Publication date: 2022-08-05
Also published as: WO2023207952A1

Abstract

The embodiment of the disclosure provides a data processing method and device, a chip, an electronic device and a medium, wherein the data processing device comprises a node array, and the node array comprises a plurality of node groups; wherein adjacent nodes in the same node group are connected, and each node in the same node group is connected with a plurality of nodes of other node groups; the plurality of nodes comprise default nodes and standby nodes, wherein the default node of one node is the standby node of at least one other node in the node group where the node is located; in the case that a default node of a first node in a node group is in a normal state, connection between the first node and the corresponding default node is enabled; in case a default node of a first node in a node group is in an abnormal state, connections between the first node and at least one second node in the same node group as the first node and a corresponding standby node are enabled.

Description

Data processing method and device, chip, electronic equipment and medium

Technical Field

The present disclosure relates to the field of data processing device technologies, and in particular, to a data processing method and device, a chip, an electronic device, and a medium.

Background

For a data processing apparatus including a plurality of nodes, when one of the nodes is in an abnormal state, it may cause the entire data processing apparatus to be rejected. For this reason, redundancy logic may be added to improve the usability of the data processing apparatus. However, the redundant logic in the related art often requires a large change in the topology of each node.

Disclosure of Invention

In a first aspect, an embodiment of the present disclosure provides a data processing apparatus, which includes a node array including a plurality of node groups; wherein adjacent nodes in the same node group are connected, and each node in the same node group is connected with a plurality of nodes of other node groups; the plurality of nodes comprise default nodes and standby nodes, wherein the default node of one node is the standby node of at least one other node in the node group where the node is located; in the case that a default node of a first node in a node group is in a normal state, connection between the first node and the corresponding default node is enabled; in case a default node of a first node in a node group is in an abnormal state, connections between the first node and at least one second node in the same node group as the first node and a corresponding standby node are enabled.

In some embodiments, each node in the array of nodes includes a processing core and a router connected to the processing core, the router of each node in one node group for connecting routers of a plurality of nodes of another node group; under the condition that a router and a processing core of a node are both in a normal state, the node is in the normal state; in the case where at least one of a router and a processing core of a node is in an abnormal state, the node is in an abnormal state.

In some embodiments, each node group comprises at least one redundant node and a working node other than the redundant node, the redundant node of one node group is a backup node of at least one working node of another node group, and the first node is the working node; in the case where each working node of a node group is in a normal state, the redundant node of the node group is disabled.

In some embodiments, in the case that one redundant node is provided for each node group, the standby node of each working node is a default node of the next working node in the group where the node is located; and responding to the fact that the default node of at least one working node is in an abnormal state, and enabling connection between the working node and the last working node of the node group where the working node is located and the corresponding standby node, wherein the number of the nodes in the abnormal state in any node group is smaller than or equal to the number of redundant nodes in the group, and the redundant nodes are the next nodes of the last working node in the group to which the redundant nodes belong.

In some embodiments, in a case that at least two redundant nodes are provided for each node group, and the at least two redundant nodes are distributed at two ends of the node group, in response to that a default node of at least one working node is in an abnormal state, connection between the working node and a previous working node of a target redundant node of the working node and a corresponding standby node is enabled, respectively, where the number of nodes in the abnormal state in any node group is less than or equal to the number of redundant nodes in the group, the target redundant node and the working node are in the same node group, and the default node of each working node between a next working node of the working node and the previous working node of the target redundant node is in a normal state.

In some embodiments, in a case where at least two redundant nodes are provided for each node group, and the at least two redundant nodes are distributed at one end of the node group, in response to that a default node of at least one working node is in an abnormal state, connection between the working node and a previous working node of the redundant nodes of the group where the working node is located and a corresponding standby node is enabled, respectively, where the number of nodes in the abnormal state in any node group is less than or equal to the number of redundant nodes in the group, and for each third node connected between the enabled node and the standby node, the default node of the third node is not adjacent to the standby node of the third node.

In some embodiments, the standby node for each working node is the default node for the next working node in the group in which the working node is located.

In some embodiments, the standby node of each working node is a default node of the working nodes which are arranged in the group where the working node is located and spaced from the working node.

In some embodiments, the redundant nodes of each node group include an nth node of the node group, a jth node of an ith node group is a default node of a jth node of an i-1 th node group and a jth node of an i +1 th node group, and a j +1 th node of the ith node group is a spare node of the jth node of the i-1 th node group and the jth node of the i +1 th node group; under the condition that the jth node of the ith node group is in an abnormal state, enabling connection between the vth node of the ith-1 node group and the v +1 th node of the ith node group, and enabling connection between the vth node of the ith +1 node group and the v +1 th node of the ith node group; j is more than or equal to 1 and less than N, v is more than or equal to j and less than N, v, i, j and N are positive integers, and N is the total number of nodes of each node group.

In some embodiments, the redundant nodes of each node group include the 1 st node and the nth node of the node group; the jth node of the ith node group is a default node of the jth node of the ith-1 node group and the jth node of the ith +1 node group, the jth +1 node of the ith node group and the jth-1 node of the ith node group are standby nodes of the jth node of the ith-1 node group, and the jth +1 node of the ith node group and the jth-1 node of the ith node group are standby nodes of the jth node of the ith +1 node group; under the condition that the jth node and the kth node of the ith node group are in abnormal states, enabling connection between the vth node of the ith-1 node group and the vth +1 node of the ith node group, enabling connection between the vth node of the ith +1 node group and the vth +1 node of the ith node group, enabling connection between the uth node of the ith-1 node group and the u-1 node of the ith node group, and enabling connection between the uth node of the ith +1 node group and the u-1 node of the ith node group; 1< j < k < N, k < v < N, 1< u < j, u, v, i, j, k and N are positive integers, and N is the total number of nodes in each node group.

In some embodiments, the redundant nodes of each node group include the nth-1 node and the nth node of the node group; the jth node of the ith node group is a default node of the jth node of the ith-1 node group and the jth node of the ith +1 node group, the jth +1 node of the ith node group and the jth +2 node of the ith node group are standby nodes of the jth node of the ith-1 node group, and the jth +1 node of the ith node group and the jth +2 node of the ith node group are standby nodes of the jth node of the ith +1 node group; under the condition that the jth node and the j +1 th node of the ith node group are in abnormal states, enabling connection between the vth node of the ith-1 node group and the v +2 th node of the ith node group, and enabling connection between the vth node of the ith +1 node group and the v +2 th node of the ith node group; j is more than or equal to 1 and less than N-1, v is more than or equal to j and less than or equal to N-1, v, i, j and N are positive integers, and N is the total number of nodes of each node group.

In some embodiments, the redundant nodes of each node group include an nth node of the node group, a jth node of an ith node group is a default node of a jth node of an i-1 th node group and a jth node of an i +1 th node group, and a j +1 th node of the ith node group is a spare node of the jth node of the i-1 th node group and the jth node of the i +1 th node group; under the condition that the jth node of the ith node group and the jth node of the (i + 1) th node group are in abnormal states, enabling connection between the vth node of the (i-1) th node group and the (v + 1) th node of the ith node group, and enabling connection between the vth node of the (i + 2) th node group and the (v + 1) th node of the (i + 1) th node group; j is more than or equal to 1 and less than N, v is more than or equal to j and less than N, v, i, j and N are positive integers, and N is the total number of nodes of each node group.

In some embodiments, in the case that the number of nodes in an abnormal state in any one node grouping is greater than the number of redundant nodes in the node grouping, the target node in each node grouping is bypassed, so that the number of nodes in an abnormal state which are not bypassed in any one node grouping is less than or equal to the number of redundant nodes in the node grouping; wherein the target node comprises the node in the abnormal state, and the target node in one node group is a default node of the target node in another node group.

In some embodiments, the number of redundant nodes in a node group is determined based on at least one of the following conditions: the area of the data processing device, the probability of the node being in an abnormal state, and the number of nodes in the node array.

In some embodiments, the connection between one node and the corresponding default node or backup node is enabled based on the corresponding preset identification information; each default node and each standby node of one node correspond to different preset identification information.

In some embodiments, the data processing apparatus further comprises a control unit for: acquiring the working state of each node in the node array; and setting the preset identification information of each node based on the working state of each node.

In some embodiments, the abnormal state comprises a first abnormal state caused by a process defect; the data processing apparatus further includes: the storage unit is used for storing first position information of the node in the first abnormal state, so that the control unit sets preset identification information of the node in the first abnormal state based on the first position information.

In some embodiments, the exception state comprises a second exception state caused by a work environment; the data processing apparatus further includes: and the detection unit is used for detecting second position information of the node in a second abnormal state in real time in the working process of the data processing device so that the control unit sets the preset identification information of the node in the second abnormal state based on the second position information.

In some embodiments, the control unit is further configured to: under the condition that at least one node is switched from a policy state to an abnormal state, suspending a task currently executed by each node in the node array before preset identification information of each node is set based on the working state of each node.

In some embodiments, the outputs of the nodes in each node group are connected to a multiplexer, and the inputs of the nodes in each node group are connected to a demultiplexer; the multiplexer of one node is used for outputting the output signal of the node to a default node or a spare node of the node through different channels; a demultiplexer of a node is used to input the output signals output to the node through different channels into the node.

In some embodiments, the data processing apparatus further comprises: and the interfaces are used for connecting the nodes of other data processing devices.

In some embodiments, each node includes a bypass element; and under the condition that the node is in an abnormal state, the bypass unit bypasses the node so as to directly connect two nodes adjacent to the node in the node group in which the node is positioned.

In a second aspect, an embodiment of the present disclosure provides a chip including the data processing apparatus according to any embodiment of the present disclosure.

In a third aspect, an embodiment of the present disclosure provides an electronic device including the data processing apparatus in any embodiment of the present disclosure.

In a fourth aspect, an embodiment of the present disclosure provides a data processing method, configured to adjust a connection relationship between nodes in a data processing apparatus according to any one of the first aspect of the present disclosure; the method comprises the following steps:

acquiring states of default nodes of all nodes, wherein the states comprise a normal state and an abnormal state;

adjusting the connection relation of a plurality of nodes in the node array based on the state of a default node of each node; wherein:

enabling a connection between a first node and a corresponding default node in a node group when the default node of the first node is in a normal state;

and enabling connection between the first node and at least one second node in the same node group with the first node and a corresponding standby node under the condition that a default node of the first node in one node group is in an abnormal state, wherein the default node of each second node is another second node or the standby node of the first node.

In a fifth aspect, an embodiment of the present disclosure provides a data processing apparatus, configured to adjust a connection relationship between nodes in the data processing apparatus according to any one of the first aspects of the present disclosure; the device comprises:

the acquisition module is used for acquiring the states of default nodes of all the nodes, wherein the states comprise a normal state and an abnormal state;

the adjusting module is used for adjusting the connection relation of a plurality of nodes in the node array based on the state of the default node of each node; wherein:

In the embodiment of the present disclosure, each node of the node group may be connected to a plurality of nodes including a default node and a standby node, so that, when the default node is in a normal state, only the connection with the default node may be enabled, and when the default node is in an abnormal state, the standby node may be used as a redundant node of the default node, thereby implementing a redundant logic. In addition, when the default node of the first node is in an abnormal state, the first node and the connection between at least one second node in the same node group with the first node and the corresponding standby node are enabled, and the default node of each second node is another second node or the standby node of the first node, so that the overall topology structure of the whole node array is kept the same as the original topology structure as much as possible, and the change of the topology structure of the nodes in the data processing device after adopting redundant logic is reduced.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the disclosure.

Fig. 1 is a schematic diagram of a data processing apparatus of an embodiment of the present disclosure.

Fig. 2A is a schematic diagram of a connection mode of nodes in a normal state.

Fig. 2B and 2C are schematic diagrams of different connection modes of nodes in an abnormal state, respectively.

Fig. 3 to 8 are schematic diagrams of the redundancy logic in different cases, respectively.

Fig. 9 and 10 are schematic diagrams of a node array including a processing core and a router, respectively.

Fig. 11 is a schematic diagram showing a connection mode of a node in an abnormal state in the node array shown in fig. 10.

Fig. 12 is a schematic diagram of a data processing apparatus including an interface.

Fig. 13 is a schematic diagram of a data processing apparatus including a multiplexer and a demultiplexer.

Fig. 14 is a diagram when the number of nodes in an abnormal state is larger than the number of redundant nodes.

Fig. 15 is a flow chart of a data processing method of an embodiment of the present disclosure.

Fig. 16 is a block diagram of a data processing apparatus of another embodiment of the present disclosure.

Detailed Description

Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, like numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the exemplary embodiments below are not intended to represent all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the disclosure, as detailed in the appended claims.

The terminology used in the present disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. As used in this disclosure and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items. In addition, the term "at least one" herein means any one of a plurality or any combination of at least two of a plurality.

It is to be understood that although the terms first, second, third, etc. may be used herein to describe various information, such information should not be limited to these terms. These terms are only used to distinguish one type of information from another. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information, without departing from the scope of the present disclosure. The word "if" as used herein may be interpreted as "at … …" or "when … …" or "in response to a determination", depending on the context.

In order to make the technical solutions in the embodiments of the present disclosure better understood and make the above objects, features and advantages of the embodiments of the present disclosure more comprehensible, the technical solutions in the embodiments of the present disclosure are described in further detail below with reference to the accompanying drawings.

For a data processing apparatus including a plurality of nodes, when one of the nodes is in an abnormal state, it may cause the entire data processing apparatus to be rejected. In some application scenarios, the data processing apparatus is applied in a chip. With the rapid development of artificial intelligence, the chip area is larger and larger. The larger the chip area, the lower the yield (yield). The yield is the ratio of the number of qualified chips to the total number of chips produced. In order to improve the yield of the chip, redundant logic is generally added to the chip, for example, in a chip including an X row and Y column node array, one row or one column of nodes may be additionally added as redundant nodes. Under normal conditions, the redundant node does not work; when a node in an abnormal state exists in the node array, the redundant node can replace the node in the abnormal state to work. However, the redundant logic in the related art often requires a large change in the topology of each node in the chip.

Based on this, the present disclosure provides a data processing apparatus, see fig. 1, comprising a node array 101, said node array 101 comprising a plurality of node groups 101 a;

wherein adjacent nodes in the same node group 101a are connected, and each node in the same node group 101a is connected with a plurality of nodes of other node groups 101 a; the plurality of nodes comprise default nodes and standby nodes, wherein the default node of one node is the standby node of at least one other node in the node group 101a where the node is located;

in a case where a default node of a first node in one node group 101a is in a normal state, connection between the first node and the corresponding default node is enabled;

in case the default node of a first node in one node group 101a is in an abnormal state, the connection between said first node and at least one second node in the same node group as the first node and the corresponding stand-by node is enabled.

In the data processing apparatus provided by the embodiment of the present disclosure, the node array 101 may include a plurality of nodes, and each node in each node group is used for data processing, and may also be referred to as a data processing node or a processing node. The plurality of nodes may form an array of P rows and Q columns, where P and Q are both positive integers. In the embodiment shown in fig. 1, each square represents a node, and P and Q both take the value 5. Of course, in practical application, P and Q may take other values, and the values of P and Q may not be equal. In fig. 1, the numbers in each block indicate the logical coordinates of the corresponding node in the node array 101, the preceding numbers in the parentheses indicate the row where the node is located, and the following numbers indicate the column where the node is located. For example, (0,0) represents a node of the 1 st row and 1 st column in the node array 101, and (0,1) represents a node of the 1 st row and 2 nd column in the node array 101. For convenience of description, the node is hereinafter represented by coordinates of the node, for example, the node having coordinates (0,0) is represented as a node (0,0), and other functional units are named similarly to the node.

The node array 101 may include a plurality of node packets 101a, which may also be referred to as packets for short. One node group 101a may be one row or one column in the node array 101, for example, in the case where each column in the node array 101 is taken as one node group 101a, the node (0,0), the node (1,0), the node (2,0), the node (3,0), and the node (4,0) may constitute one of the node groups 101 a. Adjacent nodes in the same node group are connected, for example, the node (0,0) is connected with the node (1,0), the node (2,0) is connected with the node (1,0) and the node (3,0), respectively, the solid line with the arrow represents the connection relation, and the arrow represents the data flow direction. Each node in the node group 101a is used to connect to a plurality of nodes in another node group 101a, taking a first node group in which a first column in the node array 101 is located and a second node group in which a second column in the node array 101 is located as an example, a node (0,0) in the first node group is used to connect to a node (0,1) and a node (1,1) in the second node group, and a node (1,0) in the first node group is used to connect to a node (0,1), a node (1,1) and a node (2,1) in the second node group.

The plurality of nodes include default nodes and standby nodes, and the number of the default nodes and the number of the standby nodes of the same node are both greater than or equal to 1. In the case of taking each column in the node array 101 as one node group 101a, for a node in one node group in the node array 101, an adjacent node in the other node group that is in the same row (i.e., the same row coordinate) as the node may be used as a default node of the node, and a node in the other node group that is in a different row (i.e., different row coordinate) from the node and has a connection relationship with the node may be used as a backup node of the node. For example, the default node of the node (0,0) is the node (0,1), and the default node of the node (0,1) includes the node (0,0) and the node (0, 2); the standby nodes of the nodes (0,0) are the nodes (1,1), and the standby nodes of the nodes (0,1) comprise the nodes (1,0) and the nodes (1, 2).

In addition to the situation shown in the figure, a node may have a greater number of standby nodes. For example, the standby nodes of the nodes (0,0) may include the nodes (1,1) and (2,1), and the standby nodes of the nodes (0,1) may include the nodes (1,0), the nodes (2,0), the nodes (1,2), and the nodes (2, 2). That is, a standby node for a node may include one or more nodes that are in the same node group as the default node for the node and have a row position below the default node for the node. In addition to the above, a stand-by node for a node may also include one or more nodes in the same node group as the default node for the node and having a row position above the default node for the node. For example, the backup node of the nodes (2,0) may be the node (1,1), or include the node (1,1) and the node (0, 1). Alternatively, the standby node of one node may include a node satisfying any one of the above situations. For example, the standby nodes of the nodes (2,0) may include the nodes (1,1) and the nodes (3,1), or the standby nodes of the nodes (2,0) may include the nodes (0,1), the nodes (1,1) and the nodes (3, 1). The above embodiment shows a case where the row distance between one node and the corresponding standby node is less than or equal to 2, and in other embodiments, the row distance between one node and the corresponding standby node may also be greater than 2, which is not described herein.

The default node of one node may be a backup node for at least one other node in the same node group, e.g., the default node of node (1,0) is node (1,1), and node (1,1) may be a backup node for node (0,0) and node (2, 0).

The above example exemplarily describes a case where each node group 101a includes one column in the node array 101, and in other examples, each node group 101a may also include one row in the node array 101. For example, in fig. 1, node (2,0), node (2,1), node (2,2), node (2,3), and node (2,4) may constitute one of the node groups 101 a. Adjacent nodes in the same node group are connected, for example, the node (2,0) is connected with the node (2,1), the node (2,2) is respectively connected with the node (2,1) and the node (2,3), the solid line with the arrow represents the connection relation, and the arrow represents the data flow direction. Taking a third node group in which a first row in the node array 101 is located and a fourth node group in which a second row in the node array 101 is located as an example, the node (0,0) in the third node group is used to connect the node (1,0) and the node (1,1) in the fourth node group, and the node (0,1) in the third node group is used to connect the node (1,0), the node (1,1) and the node (1,2) in the fourth node group.

Likewise, the plurality of nodes includes a default node and a standby node, and the number of the default node and the number of the standby nodes of the same node may be greater than or equal to 1. In the case of using each row in the node array 101 as one node group 101a, for a node in one node group in the node array 101, an adjacent node in the other node group in the same column (i.e., the same column coordinate) as the node may be used as a default node of the node, and a node in the other node group in a different column (i.e., different column coordinate) from the node and having a connection relationship with the node may be used as a standby node of the node. For example, the default node of the node (0,0) is the node (1,0), and the default node of the node (1,0) includes the node (0,0) and the node (2, 0); the standby nodes of the nodes (0,0) are the nodes (1,1), and the standby nodes of the nodes (1,0) comprise the nodes (0,1) and the nodes (2, 1).

In addition to the situation shown in the figure, a node may have a greater number of standby nodes. For example, the standby nodes of the nodes (0,0) may include the nodes (1,1) and (1,2), and the standby nodes of the nodes (1,0) may include the nodes (0,1), the nodes (0,2), the nodes (2,1) and the nodes (2, 2). That is, a standby node for a node may include one or more nodes that are in the same node group as the default node for the node and are located after the default node for the node. In addition to the above, the standby node of a node may also include one or more nodes that are in the same node group as the default node of the node and are located before the default node of the node. For example, the backup node of the nodes (0,2) may be the node (1,1), or include the node (1,1) and the node (1, 0). Alternatively, the standby node of one node may include a node satisfying any one of the above situations. For example, the standby nodes of the nodes (0,2) may include the nodes (1,1) and the nodes (1,3), or the standby nodes of the nodes (0,2) may include the nodes (1,0), the nodes (1,1) and the nodes (1, 3). The above embodiment shows a case where the column distance between one node and the corresponding standby node is less than or equal to 2, and in other embodiments, the column distance between one node and the corresponding standby node may also be greater than 2, which is not described herein.

The default node of one node may be a backup node for at least one other node in the same node group, e.g., the default node of node (0,1) is node (1,1), and node (1,1) may be a backup node for node (0,0) and node (0, 2).

In some embodiments, in the case where a node is in a normal state, the node is in a connected state with an adjacent node in the same group; in the case where a node is in an abnormal state, the node may be bypassed so that two nodes adjacent to the node in the same packet are in a connected state. In some embodiments, the connection between one node and the default node and each of the standby nodes may be enabled only one at a time. Further, the connection between the first node and the default node may be preferentially enabled, that is, the first node is preferentially connected to the corresponding default node as long as the default node of the first node is in the normal state, and the first node is connected to the corresponding standby node only when the default node of the first node is in the abnormal state. In this way, the standby node may be used as a redundant node of the default node to establish a connection with the corresponding first node when the default node is in an abnormal state, so as to ensure that the node array 101 is in a normal working state as a whole.

For example, in the case where each node group 101a includes one column in the node array 101, assuming that the default node of the nodes (1,0) is the node (1,1), and the standby nodes are the nodes (2,1) and the nodes (3,1), in the case where the node (1,1) is in the normal state, the connection between the node (1,0) and the node (1,1) is enabled, the connection between the node (1,0) and the node (2,1) is disabled, and the connection between the node (1,0) and the node (3,1) is also disabled. In the case where the node (1,1) is in an abnormal state, the connection between the node (1,0) and the node (1,1) is disabled, the connection between the node (1,0) and the node (2,1) is enabled, and the connection between the node (1,0) and the node (3,1) is disabled. Alternatively, in the case where the node (1,1) is in an abnormal state, the connection between the node (1,0) and the node (1,1) is disabled, the connection between the node (1,0) and the node (2,1) is disabled, and the connection between the node (1,0) and the node (3,1) is enabled. The connection between a specific enabled node and which standby node can be determined according to actual conditions.

In the case where each node group 101a includes one row in the node array 101, assuming that the default node of the nodes (0,1) is the node (1,1), and the spare nodes are the nodes (1,2) and the nodes (1,3), in the case where the node (1,1) is in the normal state, the connection between the node (0,1) and the node (1,1) is enabled, the connection between the node (0,1) and the node (1,2) is disabled, and the connection between the node (0,1) and the node (1,3) is also disabled. In the case where the node (1,1) is in an abnormal state, the connection between the node (0,1) and the node (1,1) is disabled, the connection between the node (0,1) and the node (1,2) is enabled, and the connection between the node (0,1) and the node (1,3) is disabled. Alternatively, in the case where the node (1,1) is in an abnormal state, the connection between the node (1,0) and the node (1,1) is disabled, the connection between the node (0,1) and the node (1,2) is disabled, and the connection between the node (0,1) and the node (1,3) is enabled. The connection between the specific enabled standby node and the standby node can be determined according to actual conditions.

It will be understood by those skilled in the art that the case where each node group 101a includes one row in the node array 101 is equivalent to the case where each node group 101a includes one column in the node array 101, and the node array is rotated by 90 degrees, and for convenience of description, the case where each node group 101a includes one column in the node array 101 will be mainly described below as an example.

In addition to enabling a connection between a first node and a corresponding standby node, embodiments of the present disclosure enable a connection between at least one second node in the same node group as the first node and the corresponding standby node if a default node of the first node is in an abnormal state. Wherein the default node of each second node is a backup node of the first node or another second node. For example, for any one node a except the last row in fig. 1, it is assumed that the neighboring node in the same row as the node a in fig. 1 is the default node of the node a, the node next to the default node of the node a is the next node of the node a, and the node in the same column as the default node of the node a is the standby node of the node a.

In the case where each node is in a normal state, the connection between each node and the corresponding default node is enabled, and the connection manner in this case is as shown in fig. 2A. Since the connection between each node and its standby node is disabled, there is only a connection relationship between neighboring nodes in the same row and between neighboring nodes in the same column.

Assuming that the default node of the nodes (1,0) (i.e., the node (1,1)) is a node in an abnormal state, all of the nodes (1,0), the nodes (2,0), the nodes (3,0), the nodes (4,0), the nodes (1,2), the nodes (2,2), the nodes (3,2), and the nodes (4,2) are enabled and the connection between the node and the default node of each node is disabled, and the connection manner in this case is as shown in fig. 2B. It should be noted that, under the above redundancy logic, the last row of nodes does not have a corresponding backup node, and therefore, under this connection relationship, the nodes (4,0), the nodes (4,2), the nodes (4,3), and the nodes (4,4) in the last row of nodes may all be set to the inactive state, that is, the nodes (4,0), the nodes (4,2), the nodes (4,3), and the nodes (4,4) are deactivated.

By the above mode, the topological structure of each node in the node array can be kept basically unchanged, and only one row of nodes in the node array is reduced. For example, in the embodiment shown in fig. 2B, node (4,1) replaces node (3,1) as the node connecting node (3,0) and node (3,2), node (3,1) replaces node (2,1) as the node connecting node (2,0) and node (2,2), node (2,1) replaces node (1,1) as the node connecting node (1,0) and node (1,2), and the topology of the first 4 rows in the node array remains unchanged.

Of course, the above situation is only an exemplary illustration, and is not the only way to implement the solution of the present disclosure, for example, in the embodiment shown in fig. 2C, for any one node a except the last two rows in fig. 1, it is assumed that the neighboring node in the same row as the node a in fig. 1 is the default node of the node a, the second row below the default node of the node a, and the node in the same column as the default node of the node a is the standby node of the node a. In addition, in other embodiments, a node in the z-th row below the default node of the node a and in the same column as the default node of the node a may be a standby node of the node a, where z is an integer greater than 2. Alternatively, the nodes in the rows below the default node of node a and in the same column as the default node of node a may be the standby nodes of node a.

In some embodiments, each node group includes at least one redundant node and a working node other than the redundant node, the redundant node of one node group is a backup node of at least one working node of another node group, and the first node is the working node. For example, in the embodiment shown in fig. 2A and 2B, the last node of each node group may be taken as a redundant node, i.e., the last row of nodes in the node array 101 are all redundant nodes.

In the case where each working node of a node group is in a normal state, the redundant node of the node group may be disabled, that is, the redundant node of the node group is set to a non-working state. The redundant nodes of a node group are enabled only if there is a working node in an abnormal state in the node group. Still taking the embodiment shown in fig. 2B as an example, the last row of nodes in the node array are redundant nodes, which are normally in a non-operating state, and since the node (1,1) is in an abnormal state from a certain time, the node (4,1) in the redundant nodes is enabled to implement the redundant logic, and the other redundant nodes are still in the non-operating state. In this way, the topology of each working node can be kept unchanged. The present embodiment differs from the foregoing embodiments in that all of the nodes (4,0), the nodes (4,2), the nodes (4,3), and the nodes (4,4) are set to the non-operating state, in that in the foregoing embodiments, the nodes in the last row are all in the operating state, like other nodes, and only in the case where there is a node in an abnormal state, the nodes in the last row are set to the non-operating state.

The above-described embodiment shows the case where the redundant node includes a row of nodes in the node array 101, and besides, the redundant node may include a column of nodes in the node array 101, or include at least two rows and/or at least two columns of nodes in the node array 101. Taking the example that the redundant node includes two rows of nodes in the node array 101, the at least two rows of nodes may be two consecutive rows in the node array 101, for example, the 1 st row and the 2 nd row, or the last row and the 2 nd from last row; there may also be two non-consecutive rows in the node array 101, for example, row 1 and last row.

The number of redundant nodes in a node group is determined based on at least one of the following conditions: the area of the data processing device, the probability of the node being in an abnormal state, and the number of nodes in the node array. Generally, the area of the data processing device, the probability of the node being in an abnormal state, and the number of nodes in the node array are all positively correlated with the number of redundant nodes in one node group, and the larger the area of the data processing device is, the more nodes are included in one node group, so that the more nodes that may be in an abnormal state are included in the node group. Therefore, a greater number of redundant nodes need to be provided in the node group. Similarly, the greater the probability that a node is in an abnormal state, or the greater the number of nodes in the node array 101, the more nodes in a node group that may be in an abnormal state, and therefore, the greater the number of redundant nodes in the node group.

The following illustrates possible scenarios for the number and location of redundant nodes, and the redundancy logic in these scenarios.

The first condition is as follows: each node grouping includes a redundant node. In this case, in response to that a default node of at least one working node is in an abnormal state, the connections between the working node and the last working node of the node group where the working node is located and the corresponding standby node are enabled, respectively, where the number of nodes in the abnormal state in any node group is less than or equal to the number of redundant nodes in the group, and the redundant node is a node next to the last working node in the group to which the redundant node belongs. Optionally, the standby node of each working node is a default node of a next working node in the group where the node is located; or, optionally, the standby node of each working node is a default node of the working node which is located behind the working node in the group where the node is located and has a distance greater than or equal to 2 from the working node.

For example, assume that the default node of node a is in an abnormal state, the group in which node a is located is a group T, the nodes in the group T are { node B, node C, node a, node D, node E, node F, redundant node }, the standby node of node a is the default node of the next working node of node a (i.e., node D), the standby node of node D is the default node of the next working node of node D (i.e., node E), and so on. The last working node from the working node to the node group where the working node is located includes node a, node D, node E and node F. The connections between the node A, D, E, F and the corresponding standby node are all enabled.

In some embodiments, the redundant nodes of each node group include an nth node of the node group, a jth node of an ith node group is a default node of a jth node of an ith-1 node group and a jth node of an ith +1 node group, and a jth +1 node of the ith node group is a spare node of the jth node of the ith-1 node group and the jth node of the ith +1 node group.

Under the condition that the jth node of the ith node group is in an abnormal state, enabling connection between the vth node of the ith-1 node group and the v +1 node of the ith node group, and enabling connection between the vth node of the ith +1 node group and the v +1 node of the ith node group; j is more than or equal to 1 and less than N, j is more than or equal to v and less than N, v, j and N are positive integers, and N is the total number of nodes of each node group.

This embodiment is similar to the embodiment shown in fig. 2B, and the difference is only that the redundant node is in the disabled state when there is no working node in the abnormal state in the node group to which the redundant node belongs.

It should be noted that the redundant node of each node group may also be replaced by the 1 st node of the node group, in this case, the jth node of the ith node group is a default node of the jth node of the ith-1 st node group and the jth node of the (i + 1) th node group, and the jth-1 st node of the ith node group is a standby node of the jth node of the ith-1 st node group and the jth node of the (i + 1) th node group. Under the condition that the jth node of the ith node group is in an abnormal state, enabling connection between the vth node of the ith-1 node group and the vth-1 node of the ith node group, and enabling connection between the vth node of the (i + 1) th node group and the vth-1 node of the ith node group; wherein, j is more than 1 and less than or equal to N, v is more than or equal to 1 and less than or equal to j, v, i, j and N are positive integers, and N is the total number of nodes of each node group. This case corresponds to the data processing apparatus being flipped upside down, similarly to the case where the redundant node is the nth node in the node group.

Case two: each node group is provided with at least two redundant nodes, and the at least two redundant nodes are distributed at two ends of the node group. In this case, in response to that a default node of at least one working node is in an abnormal state, connection between a working node and a corresponding standby node that is before a target redundant node of the working node and is connected to the corresponding standby node is enabled, where the number of nodes in the abnormal state in any node group is less than or equal to the number of redundant nodes in the group, the target redundant node and the working node are in the same node group, and the default node of each working node between a working node that is after the working node and a working node that is before the target redundant node is in a normal state.

Optionally, the standby node of each working node is a default node of a next working node in the group where the node is located; or, optionally, the standby node of each working node is a default node of the working node which is located behind the working node in the group where the node is located and has a distance greater than or equal to 2 from the working node.

For example, assuming that default nodes of the node a and the node B are in an abnormal state, the group in which the node a is located is a group T, and nodes in the group T are { redundant node 1, node C, node D, node a, node E, node B, node F, redundant node 2}, then a target redundant node of the node a is redundant node 1, and a target redundant node of the node B is redundant node 2. Thus, for node a, the working node immediately preceding the target redundant node of the working node includes node C, node D, and node a, and for node B, the working node immediately preceding the target redundant node of the working node includes node B, and node F.

Taking the number of redundant nodes in each node group equal to 2 as an example, assuming that the redundant nodes of each node group include the 1 st node and the nth node of the node group; the jth node of the ith node group is a default node of the jth node of the ith-1 node group and the jth node of the ith +1 node group, the jth +1 node of the ith node group and the jth-1 node of the ith node group are standby nodes of the jth node of the ith-1 node group, and the jth +1 node of the ith node group and the jth-1 node of the ith node group are standby nodes of the jth node of the ith +1 node group.

Under the condition that the jth node and the kth node of the ith node group are in abnormal states, enabling connection between the vth node of the ith-1 node group and the vth +1 node of the ith node group, enabling connection between the vth node of the ith +1 node group and the vth +1 node of the ith node group, enabling connection between the uth node of the ith-1 node group and the u-1 node of the ith node group, and enabling connection between the uth node of the ith +1 node group and the u-1 node of the ith node group; wherein 1< j < k < N, k < v < N, 1< u < j, u, v, i, j, k and N are positive integers, and N is the total number of nodes grouped by each node.

Referring to fig. 3, it is assumed that the node array 101 includes 6 rows and 5 columns, where a row of nodes with row coordinates of 0 and a row of nodes with row coordinates of 5 are redundant nodes. In the case where the node (4,1) is in the abnormal state, the connection between the node (4,0) and the node (5,1) may be enabled, and the connection between the node (4,2) and the node (5,1) may be enabled. In the case where the node (2,1) is in the abnormal state, the connection between the node (2,0) and the node (1,1) may be enabled, the connection between the node (2,2) and the node (1,1) may be enabled, the connection between the node (1,0) and the node (0,1) may be enabled, and the connection between the node (1,2) and the node (0,1) may be enabled.

Case three: each node group is provided with at least two redundant nodes, and the at least two redundant nodes are distributed at one end of the node group. In this case, in response to that a default node of at least one working node is in an abnormal state, connections between a working node and a corresponding standby node before the working node and the redundant node of a group in which the working node is located may be enabled, respectively, where the number of nodes in the abnormal state in any node group is less than or equal to the number of redundant nodes in the group.

In some embodiments, for each third node that enables a connection with a standby node, the default node of the third node is not adjacent to the standby node of the third node. For example, the standby node of each working node is a default node of the working nodes which are arranged in the group where the working node is located and spaced from the working node. The two node interval setting may include a case where one or more nodes are included between the two nodes.

For example, assuming that the default nodes of the node a and the node B are in an abnormal state, the group in which the node a is located is the group T, and the nodes in the group T are { node C, node D, node a, node B, node E, node F, redundant node 1, redundant node 2}, the node a takes the node E as a redundant node, and the nodes between the working node and the previous working node of the redundant node in the group in which the working node is located include the node a, node B, node E, and node F. The standby node of the node a is the node E, and the default node of the node B is also included between the default node of the node a and the standby node of the node a, so that the default node of the node a is not adjacent to the standby node of the node a. Similarly, the standby node of the node B is the node F, and the default node of the node E is also included between the default node of the node B and the standby node of the node B, so that the default node of the node B is not adjacent to the standby node of the node B.

For example, the redundant nodes of each node group include the nth-1 node and the nth node of the node group; the jth node of the ith node group is a default node of the jth node of the ith-1 node group and the jth node of the ith +1 node group, the jth +1 node of the ith node group and the jth +2 node of the ith node group are standby nodes of the jth node of the ith-1 node group, and the jth +1 node of the ith node group and the jth +2 node of the ith node group are standby nodes of the jth node of the ith +1 node group.

Under the condition that the jth node and the j +1 th node of the ith node group are in abnormal states, enabling connection between the vth node of the ith-1 node group and the v +2 th node of the ith node group, and enabling connection between the vth node of the ith +1 node group and the v +2 th node of the ith node group; j is more than or equal to 1 and less than N-1, v is more than or equal to j and less than N-1, v, i, j and N are positive integers, and N is the total number of nodes of each node group.

Referring to fig. 4, it is assumed that the node array 101 includes 6 rows and 5 columns, where a row of nodes with row coordinates of 4 and a row of nodes with row coordinates of 5 are redundant nodes. In the case where both the node (2,1) and the node (3,1) are in the abnormal state, the connection between the node (2,0) and the node (4,1) may be enabled, the connection between the node (2,2) and the node (4,1) may be enabled, the connection between the node (3,0) and the node (5,1) may be enabled, and the connection between the node (3,2) and the node (5,1) may be enabled.

Case four: one node and the default node of the node are in an abnormal state. For example, the redundant nodes of each node group comprise the Nth node of the node group, the jth node of the ith node group is a default node of the jth node of the ith-1 node group and the jth node of the ith +1 node group, and the jth +1 node of the ith node group is a spare node of the jth node of the ith-1 node group and the jth node of the ith +1 node group.

Under the condition that the jth node of the ith node group and the jth node of the (i + 1) th node group are in abnormal states, enabling connection between the vth node of the (i-1) th node group and the (v + 1) th node of the ith node group, and enabling connection between the vth node of the (i + 2) th node group and the (v + 1) th node of the (i + 1) th node group; j is more than or equal to 1 and less than N, j is more than or equal to v and less than N, v, i, j and N are positive integers, and N is the total number of nodes of each node group.

Case four is actually a special manifestation of case one or case two or case three, and in the case where each node grouping includes one redundant node, case four is a special manifestation of case one; under the condition that each node group comprises two redundant nodes and the redundant nodes are distributed at two ends of the node group, the fourth case is a special expression form of the second case; in case each node group comprises two redundant nodes, and the redundant nodes are distributed at one end of the node group, case four is a special manifestation of case three. Referring to fig. 5, taking a case where each node group includes one redundant node as an example, it is assumed that the node array 101 includes 5 rows and 5 columns, where one row of nodes with row coordinates of 5 is a redundant node. In the case where both the node (2,1) and the node (2,2) are in the abnormal state, the connection between the node (2,0) and the node (3,1) may be enabled, the connection between the node (2,3) and the node (3,2) may be enabled, the connection between the node (3,0) and the node (4,1) may be enabled, and the connection between the node (3,3) and the node (4,2) may be enabled.

In addition, two or more of the above cases may be combined, and fig. 6 and 7 show two combining manners, respectively. As shown in fig. 6, is a combination of case three and case four, above, in which case the redundant nodes comprise the last two rows of nodes in the node array 101. Since both the node (3,1) of the second column and the node (3,2) of the third column are in an abnormal state, the node of the first column looks for a spare node in the second column, and the node of the fourth column looks for a spare node in the third column. In addition, since the second column and the third column each include two nodes in an abnormal state, and the redundant nodes are two consecutive rows in the node array 101, the row distances between the node (2,0), the node (3,0), the node (2,3), and the node (3,3) and the respective standby nodes are all 2, for example, the standby node of the node (2,0) is the node (4,1), and the row distance between the node (2,0) and the respective standby node is 4-2 ═ 2.

As shown in fig. 7, which is a combination of case two and case four, the redundant node in this case includes a first row and a last row in the node array 101. Since both the node (2,1) of the second column and the node (2,2) of the third column are in an abnormal state, the node of the first column looks for a spare node in the second column, and the node of the fourth column looks for a spare node in the third column. In addition, since the second column and the third column each include two nodes in an abnormal state, and the redundant nodes are two discontinuous rows in the node array 101, the nodes (1,0) and (1,3) look up for the standby nodes, and the nodes (2,0), (3,0), (2,3) and (3,3) look down for the standby nodes.

As shown in fig. 8, this is the case where redundant nodes are provided in both the row direction and the column direction of the node array 101. In this case, when the node (3,1) is in an abnormal state, the node (3,1) may be replaced with a node (4,1) among the redundant nodes set in the row direction; when the node (1,3) is in an abnormal state, the node (1,3) may be replaced with the node (1,4) among the redundant nodes arranged in the column direction.

In addition to the above-listed situations, the location and number of redundant nodes, and the implementation manner of the redundant logic may be adjusted to other situations according to actual needs, and are not listed here.

In some embodiments, each node in the array of nodes includes a processing core and a router connected to the processing core, the router of each node in one node group for connecting routers of a plurality of nodes of another node group. That is, in the above-described embodiment, the connection between the nodes is realized by the connection between the routers between the nodes. The router may be configured to implement transmission of data between nodes, and may also be configured to send data received by a node to a processing core connected to the node, so that the processing core processes the data received by the node, and the router may also receive data returned by the processing core.

As shown in fig. 9 and 10, each block represents a router, and each ellipse represents a processing core. The coordinates in the box represent the coordinates of the router and the coordinates in the ellipse represent the coordinates of the processing core. For simplicity, only the connections between each router and neighboring routers are shown in fig. 9 and 10, and the connections between the respective routers and non-neighboring routers are omitted. Each of the dashed boxes of fig. 9 and 10 represents a node, and it can be seen that in fig. 9, each node includes a router and a processing core; each processing core is connected to a router, and the processing cores to which different routers are connected are different. In fig. 10, each node of the first and last columns includes one router and one processing core, the other nodes include one router and two processing cores, and adjacent nodes may share one processing core.

In some embodiments, a node is in a normal state in the case where both a router and a processing core of the node are in a normal state; in the case where at least one of a router and a processing core of a node is in an abnormal state, the node is in an abnormal state. For example, in the embodiment shown in fig. 9, assuming that the node (0,0) includes the router (0,0) and the processing core (0,0), in the case where both the router (0,0) and the processing core (0,0) are in the normal state, the node (0,0) may be considered to be in the normal state; in the case where one of the router (0,0) or the processing core (0,0) is in an abnormal state, the node (0,0) can be considered to be in an abnormal state. In the embodiment shown in fig. 10, any processing core or router belonging to a node is in an abnormal state, and the node is considered to be in the abnormal state. For example, in a case where the processing cores (1,1) are in an abnormal state, a node including the processing cores (1,1), the routers (1,2), and the processing cores (1,2) is determined to be in an abnormal state, and a node including the processing cores (1,0), the routers (1,1), and the processing cores (1,1) is determined to be in an abnormal state. A node is considered to be in a normal state only if all processing cores and routers belonging to the node are in a normal state.

In the embodiment shown in fig. 9, each processing core is connected to only one router, so that when a node is in an abnormal state, the connection mode of the processing core does not need to be adjusted, and only the connection mode of the router needs to be adjusted. In the embodiment shown in fig. 10, since one processing core can connect two routers, when a node is in an abnormal state, the connection mode between the router and the processing core that connect the node needs to be adjusted. As shown in fig. 11, the routers in the last row and the processing cores connected thereto are respectively redundant routers and redundant processing cores, as indicated by white squares and white ovals in the figure. The black squares represent routers in an abnormal state, the black ovals represent processing cores in an abnormal state, and the gray squares and gray ovals represent routers in a normal state and processing cores in a normal state, respectively. It can be seen that, because the router (3,1) is in an abnormal state, the connection modes of the router (3,0) and the processing core (3,0) connecting the router (3,1) need to be adjusted, and the adjusted router (3,0) and the processing core (3,0) are connected to the router (4, 1). Similarly, the connection modes of the routers (3,3) and the processing cores (3,2) connected with the routers (3,2) need to be adjusted, and the adjusted routers (3,3) and the processing cores (3,2) are connected with the routers (4, 2). Since the adjusted routers (4,1), routers (4,2), and processing cores (4,1) are all in a normal state as backup nodes, the routers (4,1), routers (4,2), and processing cores (4,1) are represented as gray squares and gray ellipses, respectively, in the figure.

In some embodiments, the data processing apparatus further comprises a plurality of interfaces for connecting nodes of other data processing apparatuses. The interfaces may be provided in plural numbers on the periphery of the data processing apparatus, one arrangement being as shown in fig. 12. In practical application, an interface can be respectively set for each node; in other embodiments, multiple nodes may share an interface. The interface may transmit data output from the other data processing apparatus to the router in the data processing apparatus, or may output data transmitted from the router in the data processing apparatus to the other data processing apparatus. In some embodiments, the interface may employ serdes, GPIO bus interface, I ² C interface, etc.

In some embodiments, the connection between one node a and the corresponding default node or backup node is enabled based on the preset identification information of the node (i.e., the node a). The preset identification information may be a string of binary numbers including a plurality of data bits, the number of bits of the data bits being determined based on the total number of the default node and the spare node of one node. For example, in the case where the total number of the default node and the spare node does not exceed 4, the number of bits of the data bit is 2; in the case where the total number of the default node and the spare node is greater than 4 and does not exceed 8, the number of bits of the data bit is 3.

And each default node and each standby node of one node correspond to different preset identification information. For example, when one node includes one default node and two standby nodes, the identification information corresponding to the default node may be set to 00, the identification information of one of the standby nodes may be set to 01, and the identification information of the other standby node may be set to 11. In this way, connection of each node to its default or standby node may be selectively enabled based on different identification information. For example, when the identification information of a node is 00, connection of the node to its default node is enabled.

In some embodiments, the data processing apparatus further includes a control unit, configured to obtain an operating state of each node in the node array; and setting the preset identification information of each node based on the working state of each node.

In some embodiments, the abnormal state comprises a first abnormal state caused by a process defect. The abnormal state caused by the process defect is generally fixed and irreversible, and therefore, the position of the node in the first abnormal state can be determined before the data processing apparatus is shipped from the factory. Specifically, the data processing apparatus may further include a storage unit configured to store first position information of the node in the first abnormal state, so that the control unit sets preset identification information of the node in the first abnormal state based on the first position information.

The node in the first abnormal state may be determined by Design for Testability (DFT) detection, and the memory cell may be a one-time programmable memory such as an electrically programmable fuse (efuse). After the data processing device is powered on and started, the defect core position pre-stored in the efuse can be read through a Micro-Controller Unit (MCU) or dedicated hardware, a replacement core strategy is selected according to the number and the position of the defect cores, and the preset identification information of the corresponding node is configured according to the replacement strategy, so that the reconstruction of the node array is completed.

In some embodiments, the exception state comprises a second exception state caused by the operating environment. The abnormal state caused by the working environment (e.g., high temperature, high pressure) is often uncertain, and can be either reversible or irreversible. Therefore, the position of the node in the second abnormal state cannot be directly stored in the storage unit. In order to solve the problem, a detection unit may be provided in the data processing apparatus, for detecting second position information of the node in the second abnormal state in real time during operation of the data processing apparatus, so that the control unit sets the preset identification information of the node in the second abnormal state based on the second position information.

The detection unit may be a failure detection circuit or detection software, for example, the failure detection circuit may be implemented by one or more sensors.

In some embodiments, the control unit is further configured to, in a case where at least one node switches from a normal state to an abnormal state, suspend a task currently executed by each node in the node array before setting preset identification information of each node based on a working state of each node. That is to say, when a new node is in an abnormal state, the task currently executed by each node may be suspended, the replacement node may be determined again, and the preset identification information may be configured according to the determined replacement node, so as to complete the reconstruction of the node array.

As shown in FIG. 13, the Multiplexer (Multiplexer) is denoted as MUX, and the demultiplexer (De-Multiplexer) is denoted as DMUX. Each square represents a node, the left side of the dotted line represents the signal flow direction when each node is in a normal state, the right side of the dotted line represents the signal flow direction when the node (1,1) is in an abnormal state, and the solid line with an arrow represents the connection relationship between the nodes. It can be seen that in the left side node array, connection between nodes (1,0), nodes (1,1) and nodes (1,2) is enabled, as shown by the thicker solid lines in the left side node array; in the node array on the right, the connection between nodes (1,0), nodes (0,1) and nodes (1,2) is enabled, as shown by the thicker solid lines in the node array on the right. Similarly, the connection between the node (0,0) and the previous node of the nodes (0,1) is enabled, and the connection between the node (0,2) and the previous node of the nodes (0,1) is also enabled.

In the above embodiment, each of the multiplexers and each of the demultiplexers corresponds to one of a set of preset identification information. In the above-described embodiment, the nodes (1,1) are in an abnormal state, and the preset identification information of each multiplexer and each demultiplexer connected by a thick solid line can be configured. For example, in the above-described embodiment, one multiplexer and one demultiplexer are involved in a three-way signal, and thus, a set of preset identification information may include 00, 01, and 11, where 00 denotes a node connected to the same row, 01 denotes a node connected to the upper row, and 11 denotes a node connected to the next row. In this case, the preset identification information corresponding to the MUX connected to the node (1,0), the DMUX on the left side of the node (0,1), the MUX on the right side of the node (0,1), the DMUX connected to the node (1,2), and the MUX connected to the node (1,2) is respectively set as: 01,01,11,11,11,11,01,01.

Taking the embodiment shown in fig. 2B as an example, the nodes are not only connected to each other in the horizontal direction but also connected to each other in the vertical direction. In case the node (1,1) is in an abnormal state, the node (1,1) needs to be bypassed by a bypass unit in the node (1,1) so that the node (0,1) is directly connected to the node (2,1) in the vertical direction.

In some embodiments, in the case that the number of nodes in an abnormal state in any one node grouping is greater than the number of redundant nodes in the node grouping, the target node in each node grouping is bypassed, so that the number of nodes in an abnormal state which are not bypassed in any one node grouping is less than or equal to the number of redundant nodes in the node grouping; and the target node in each node group comprises the node in the abnormal state, and the target node in one node group is a default node of the target node in the other node group. For example, assuming that the node group includes a group 1, a group 2, a group 3, and a group 4, and target nodes in the four node groups are sequentially denoted as a target node 1, a target node 2, a target node 3, and a target node 4, the target node 1 is a default node of the target node 2, the target node 2 is a default node of the target node 3, and the target node 3 is a default node of the target node 4.

In some embodiments, a node group includes a column of a node array, and a default node of a node in a group is in the same row as the node, then the destination node 1, the destination node 2, the destination node 3, and the destination node 4 are in the same row of nodes in the node array. In other embodiments, a node group includes a row of a node array, the default node of the node in a group is in the same column as the node, and the destination node 1, the destination node 2, the destination node 3, and the destination node 4 are in the same column of the node array. That is, in the above embodiment, once the number of nodes in an abnormal state in a certain node group is greater than the number of redundant nodes in the node group, one row or one column of the node array including the node in the abnormal state is removed.

As shown in fig. 14, assuming that the last row of nodes is a redundant node, since there is only one redundant node in each node group, but there are two nodes in an abnormal state in the node group where the second column is located, namely node (0,1) and node (1,1), at this time, the redundant node is not enough to provide redundant logic for the node in the abnormal state. Thus, the individual nodes in the row in which node (0,1) is located, i.e., the dashed box, may be removed to ensure that the number of redundant nodes is sufficient to provide redundant logic for the nodes in the abnormal state.

In some embodiments, the present disclosure also provides an electronic device including the data processing apparatus according to any one of the embodiments of the present disclosure.

In some embodiments, as shown in fig. 15, the present disclosure further provides a data processing method, configured to adjust a connection relationship between nodes in a data processing apparatus according to any embodiment of the present disclosure; the method comprises the following steps:

step 1501: acquiring states of default nodes of all nodes, wherein the states comprise a normal state and an abnormal state;

step 1502: adjusting the connection relation of a plurality of nodes in the node array based on the state of a default node of each node; wherein:

step 15021: enabling a connection between a first node and a corresponding default node in a node group when the default node of the first node is in a normal state;

step 15022: and enabling connection between the first node and at least one second node in the same node group with the first node and a corresponding standby node under the condition that a default node of the first node in one node group is in an abnormal state, wherein the default node of each second node is another second node or the standby node of the first node.

The method of the disclosed embodiment can be executed by a processing unit such as an MCU, a CPU or the like or special processing hardware. The manner of enabling the connection between the nodes is described in detail in the foregoing embodiments of the data processing apparatus, and is not described herein again.

In some embodiments, as shown in fig. 16, the present disclosure further provides a data processing apparatus, configured to adjust a connection relationship between nodes in the data processing apparatus according to any embodiment of the present disclosure; the device comprises:

an obtaining module 1601, configured to obtain states of default nodes of each node, where the states include a normal state and an abnormal state;

an adjusting module 1602, configured to adjust connection relationships of multiple nodes in the node array based on states of default nodes of the nodes; wherein:

In some embodiments, functions of or modules included in the apparatus provided in the embodiments of the present disclosure may be used to execute the method described in the above method embodiments, and specific implementation thereof may refer to the description of the above method embodiments, and for brevity, no further description is given.

The embodiments of the present disclosure also provide a computer-readable storage medium, on which a computer program is stored, which when executed by a processor implements the method of any of the foregoing embodiments.

Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), Digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.

From the above description of the embodiments, it is clear to those skilled in the art that the embodiments of the present disclosure can be implemented by software plus necessary general hardware platform. Based on such understanding, the technical solutions of the embodiments of the present specification may be essentially or partially implemented in the form of a software product, which may be stored in a storage medium, such as a ROM/RAM, a magnetic disk, an optical disk, etc., and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments of the present specification.

The systems, devices, modules or units illustrated in the above embodiments may be implemented by a computer data processing device or entity, or by an article of manufacture with certain functionality. A typical implementation device is a computer, which may take the form of a personal computer, laptop computer, cellular telephone, camera phone, smart phone, personal digital assistant, media player, navigation device, email messaging device, game console, tablet computer, wearable device, or a combination of any of these devices.

The embodiments in the present specification are described in a progressive manner, and the same and similar parts among the embodiments are referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the apparatus embodiment, since it is substantially similar to the method embodiment, it is relatively simple to describe, and reference may be made to some descriptions of the method embodiment for relevant points. The above-described apparatus embodiments are merely illustrative, and the modules described as separate components may or may not be physically separate, and the functions of the modules may be implemented in one or more software and/or hardware when implementing the embodiments of the present disclosure. And part or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of the embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.

The foregoing is only a specific embodiment of the embodiments of the present disclosure, and it should be noted that, for those skilled in the art, a plurality of modifications and decorations can be made without departing from the principle of the embodiments of the present disclosure, and these modifications and decorations should also be regarded as the protection scope of the embodiments of the present disclosure.

Claims

1. A data processing apparatus, characterized in that the data processing apparatus comprises a node array comprising a plurality of node packets;

wherein adjacent nodes in the same node group are connected, and each node in the same node group is connected with a plurality of nodes of other node groups; the plurality of nodes comprise default nodes and standby nodes, wherein the default node of one node is the standby node of at least one other node in the node group where the node is located;

in the case that a default node of a first node in a node group is in a normal state, connection between the first node and the corresponding default node is enabled;

in case a default node of a first node in a node group is in an abnormal state, connections between the first node and at least one second node in the same node group as the first node and a corresponding standby node are enabled.

2. The data processing apparatus of claim 1, wherein each node in the array of nodes comprises a processing core and a router connected to the processing core, the router of each node in one node group being for connecting routers of a plurality of nodes of another node group;

under the condition that a router and a processing core of a node are both in a normal state, the node is in the normal state;

in the case where at least one of a router and a processing core of a node is in an abnormal state, the node is in an abnormal state.

3. The data processing apparatus according to claim 1 or 2, wherein each node group includes at least one redundant node and a working node other than the redundant node, the redundant node of one node group being a backup node of at least one working node of another node group, the first node being the working node;

in the case where each working node of a node group is in a normal state, the redundant nodes of the node group are disabled.

4. The data processing apparatus according to claim 3, wherein, in a case where one redundant node is provided for each node group, in response to a default node of at least one working node being in an abnormal state, connection between the working node and a last working node of the node group where the working node is located and a corresponding standby node is enabled, respectively, wherein the number of nodes in the abnormal state in any node group is less than or equal to the number of redundant nodes in the group, and the redundant node is a node next to the last working node in the group to which the redundant node belongs.

5. The data processing apparatus according to claim 3, wherein at least two redundant nodes are provided for each node group, and in a case where the at least two redundant nodes are distributed at both ends of the node group, in response to a default node of at least one working node being in an abnormal state, connection between the working node and a previous working node of a target redundant node of the working node is enabled, respectively, wherein the number of nodes in the abnormal state in any node group is less than or equal to the number of redundant nodes in the group, the target redundant node and the working node are in the same node group, and the default node of each working node between a next working node of the working node and a previous working node of the target redundant node is in a normal state.

6. The data processing apparatus according to claim 3, wherein at least two redundant nodes are provided for each node group, and the at least two redundant nodes are distributed at one end of the node group, and in response to a default node of at least one working node being in an abnormal state, connection between the working node and a previous working node of the redundant nodes of the group in which the working node is located and a corresponding standby node is enabled, respectively, wherein the number of nodes in the abnormal state in any node group is less than or equal to the number of redundant nodes in the group, and for each third node connected between the enabled and standby nodes, the default node of the third node is not adjacent to the standby node of the third node.

7. A data processing apparatus as claimed in claim 4 or 5, wherein the standby node for each working node is the default node for the next working node in the group in which the working node is located.

8. The data processing apparatus according to claim 6, wherein the backup node of each working node is a default node of the working nodes spaced apart from the working node in the group in which the working node is located.

9. The data processing apparatus according to claim 3 or 4, wherein the redundant nodes of each node group comprise an nth node of the node group, a jth node of an ith node group is a default node of a jth node of an i-1 th node group and a jth node of an i +1 th node group, and a jth +1 node of the ith node group is a spare node of the jth node of the i-1 th node group and the jth node of the i +1 th node group;

under the condition that the jth node of the ith node group is in an abnormal state, enabling connection between the vth node of the ith-1 node group and the v +1 node of the ith node group, and enabling connection between the vth node of the ith +1 node group and the v +1 node of the ith node group;

j is more than or equal to 1 and less than N, v is more than or equal to j and less than N, v, i, j and N are positive integers, and N is the total number of nodes of each node group.

10. The data processing apparatus of claim 3 or 5, wherein the redundant nodes of each node group comprise the 1 st node and the nth node of the node group; the jth node of the ith node group is a default node of the jth node of the ith-1 node group and the jth node of the ith +1 node group, the jth +1 node of the ith node group and the jth-1 node of the ith node group are standby nodes of the jth node of the ith-1 node group, and the jth +1 node of the ith node group and the jth-1 node of the ith node group are standby nodes of the jth node of the ith +1 node group;

under the condition that the jth node and the kth node of the ith node group are in abnormal states, enabling connection between the vth node of the ith-1 node group and the vth +1 node of the ith node group, enabling connection between the vth node of the ith +1 node group and the vth +1 node of the ith node group, enabling connection between the uth node of the ith-1 node group and the u-1 node of the ith node group, and enabling connection between the uth node of the ith +1 node group and the u-1 node of the ith node group;

1< j < k < N, k < v < N, 1< u < j, u, v, i, j, k and N are positive integers, and N is the total number of nodes in each node group.

11. The data processing apparatus according to claim 3 or 6 or 8, wherein the redundant nodes of each node group comprise the N-1 st node and the nth node of the node group; the jth node of the ith node group is a default node of the jth node of the ith-1 node group and the jth node of the ith +1 node group, the jth +1 node of the ith node group and the jth +2 node of the ith node group are standby nodes of the jth node of the ith-1 node group, and the jth +1 node of the ith node group and the jth +2 node of the ith node group are standby nodes of the jth node of the ith +1 node group;

under the condition that the jth node and the j +1 th node of the ith node group are in abnormal states, enabling connection between the vth node of the ith-1 node group and the v +2 th node of the ith node group, and enabling connection between the vth node of the ith +1 node group and the v +2 th node of the ith node group;

j is more than or equal to 1 and less than N-1, v is more than or equal to j and less than or equal to N-1, v, i, j and N are positive integers, and N is the total number of nodes of each node group.

12. The data processing apparatus according to claim 3 or 4, wherein the redundant nodes of each node group comprise the nth node of the node group, the jth node of the ith node group is a default node of the jth node of the ith-1 node group and the jth node of the ith +1 node group, and the jth +1 node of the ith node group is a standby node of the jth node of the ith-1 node group and the jth node of the ith +1 node group;

under the condition that the jth node of the ith node group and the jth node of the (i + 1) th node group are in abnormal states, enabling connection between the vth node of the (i-1) th node group and the (v + 1) th node of the ith node group, and enabling connection between the vth node of the (i + 2) th node group and the (v + 1) th node of the (i + 1) th node group;

13. The data processing apparatus according to any one of claims 3 to 12, wherein in a case where the number of nodes in an abnormal state in any one node group is greater than the number of redundant nodes in the node group, the target node in each node group is bypassed so that the number of nodes in an abnormal state that are not bypassed in any one node group is less than or equal to the number of redundant nodes in the node group;

and the target node in each node group comprises the node in the abnormal state, and the target node in one node group is a default node of the target node in the other node group.

14. The data processing apparatus according to any of claims 3 to 13, wherein the number of redundant nodes in a node group is determined based on at least one of the following conditions: the area of the data processing device, the probability of the node being in an abnormal state, and the number of nodes in the node array.

15. The data processing apparatus of any of claims 1 to 14, wherein a connection between a node and a corresponding default node or standby node is enabled based on preset identification information of the node; and each default node and each standby node of one node correspond to different preset identification information.

16. The data processing apparatus of claim 15, further comprising a control unit configured to:

acquiring the working state of each node in the node array;

and setting the preset identification information of each node based on the working state of each node.

17. The data processing apparatus of claim 16, wherein the abnormal condition comprises a first abnormal condition caused by a process defect; the data processing apparatus further includes:

the storage unit is used for storing first position information of the node in the first abnormal state, so that the control unit sets preset identification information of the node in the first abnormal state based on the first position information.

18. A data processing apparatus as claimed in claim 16 or 17, wherein said exception state comprises a second exception state caused by a working environment; the data processing apparatus further includes:

and the detection unit is used for detecting second position information of the node in a second abnormal state in real time in the working process of the data processing device so that the control unit sets the preset identification information of the node in the second abnormal state based on the second position information.

19. The data processing apparatus of any of claims 16 to 18, wherein the control unit is further configured to:

under the condition that at least one node is switched from a policy state to an abnormal state, suspending a task currently executed by each node in the node array before preset identification information of each node is set based on the working state of each node.

20. A data processing apparatus as claimed in any one of claims 1 to 19, characterized in that the outputs of the nodes in each node group are connected to a multiplexer and the inputs of the nodes in each node group are connected to a demultiplexer;

the multiplexer of one node is used for outputting the output signal of the node to a default node or a spare node of the node through different channels;

a demultiplexer of a node is used to input the output signals output to the node through different channels into the node.

21. The data processing apparatus according to any one of claims 1 to 20, characterized in that the data processing apparatus further comprises:

and the interfaces are used for connecting the nodes of other data processing devices.

22. A data processing apparatus as claimed in any one of claims 1 to 21, characterized in that each node comprises a bypass element;

and under the condition that the node is in an abnormal state, the bypass unit bypasses the node so as to directly connect two nodes adjacent to the node in the node group in which the node is positioned.

23. A chip comprising a data processing device according to any one of claims 1 to 22.

24. An electronic device, characterized in that it comprises a data processing device according to any one of claims 1 to 22 or a chip according to claim 23.

25. A data processing method for adjusting a connection relationship of each node in the data processing apparatus according to any one of claims 1 to 22; the method comprises the following steps:

26. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the method of claim 25.