CN111866069A

CN111866069A - Data processing method and device

Info

Publication number: CN111866069A
Application number: CN202010502775.4A
Authority: CN
Inventors: 马传文; 范志刚; 卢涛
Original assignee: Xian Wanxiang Electronics Technology Co Ltd
Current assignee: Xian Wanxiang Electronics Technology Co Ltd
Priority date: 2020-06-04
Filing date: 2020-06-04
Publication date: 2020-10-30

Abstract

The disclosure provides a data processing method and device, relates to the field of data processing, and can solve the problem of low data processing efficiency. The specific technical scheme is as follows: receiving data information; determining a processing action on the data information; processing data information according to a preset processing action; sending a data packet to the target neighboring processing unit node, wherein the data packet comprises: the processing action identification and the target data, the processing action identification indicates the processing action of the target adjacent processing unit node on the target data in the data packet, and the target data comprises: data information, or processed data information. According to the invention, each processing unit node is connected with the adjacent processing unit node, and after the current processing unit node processes the data information according to the preset processing action, the processing action of the target adjacent processing unit node is also determined, so that the processing capacity of each processing unit node can be fully exerted by the processing mode, and the processing efficiency is improved.

Description

Data processing method and device

Technical Field

The present disclosure relates to the field of data processing, and in particular, to a data processing method and apparatus.

Background

A Processing Element (PE) based coding method is a common coding implementation method at present, and when coding is performed according to this method, the coding capability can be improved by increasing the number of PEs or increasing the number of cores of an Encoder (english: Encoder), but although this method can extend the coding processing capability of PEs, the PEs are independent, so that the processing capability of multiple PEs cannot be fully exerted, and the data processing efficiency is low.

Disclosure of Invention

The embodiment of the disclosure provides a data processing method and device, which can solve the problem of low data processing efficiency. The technical scheme is as follows:

according to a first aspect of the embodiments of the present disclosure, there is provided a data processing method, which is applied to a processing unit node in an encoder network, where the encoder network includes N × M processing unit nodes, each processing unit node is connected to its neighboring processing unit node, N is a positive integer greater than 1, M is a positive integer greater than 1: the method comprises the following steps:

receiving data information;

determining a processing action for the data information;

processing the data information according to the processing action;

Sending a data packet to a target neighboring processing unit node, the data packet comprising: a processing action identifier and target data, the processing action identifier indicating a processing action of the target neighboring processing unit node on the target data in the data packet, the target data including: the data information, or the processed data information.

The data processing method provided by the embodiment of the disclosure is applied to processing unit nodes in an encoder network, wherein the encoder network comprises N multiplied by M processing unit nodes, each processing unit node is connected with the adjacent processing unit node, N is a positive integer greater than 1, and M is a positive integer greater than 1; the data processing method comprises the following steps: receiving data information; determining a processing action on the data information; processing data information according to a preset processing action; sending a data packet to the target neighboring processing unit node, wherein the data packet comprises: the processing action identification and the target data, the processing action identification indicates the processing action of the target adjacent processing unit node on the target data in the data packet, and the target data comprises: data information, or processed data information. Because each processing unit node is connected with the adjacent processing unit node, the interconnection relation among the processing unit nodes is improved, and the processing action of the target adjacent processing unit node can be determined after the current processing unit node processes the data information according to the preset processing action, so that the processing capacity of each processing unit node can be fully exerted through the processing mode, and the processing efficiency is improved.

In one embodiment, the determining the processing action on the data information comprises:

judging whether the encoder core is idle or not;

if so, determining that the data information is processed according to a preset processing mode;

and if not, determining that the processing action on the data information is to find an available target adjacent processing unit node so as to enable the target adjacent processing unit node to process the data information according to the preset processing mode.

In one embodiment, the determining the processing action on the data information is to find an available target neighboring processing unit node, and the processing the data information according to the processing action includes:

acquiring the core utilization rate of encoders of all adjacent processing unit nodes;

detecting whether idle encoder cores exist in each adjacent processing unit node according to the encoder core utilization rate;

when detecting that the adjacent processing unit node has an idle encoder core, determining the adjacent processing unit node with the idle encoder core as the target adjacent processing unit node;

generating the data packet, wherein the processing action identifier included in the data packet indicates the processing action of the target neighboring processing unit node on the target data as processing the target data according to the preset processing mode, and the target data included in the data packet is the data information;

When detecting that no idle encoder core exists in all the adjacent processing unit nodes, selecting the target adjacent processing unit node from all the adjacent processing unit nodes according to a first preset rule;

generating the data packet, wherein the processing action identifier included in the data packet indicates that the processing action of the target neighboring processing unit node on the target data is to find an available target neighboring processing unit node, and the target data included in the data packet is the data information; so that the target neighboring processing unit node re-executes the step of finding an available target neighboring processing unit node.

In one embodiment, said determining that the neighboring processing unit node having an idle encoder core is the target neighboring processing unit node when it is detected that the neighboring processing unit node has an idle encoder core comprises:

if detecting that a plurality of adjacent processing unit nodes all have idle encoder cores, determining the adjacent processing unit node with the minimum utilization rate of the encoder cores as the target adjacent processing unit node;

and if detecting that a plurality of adjacent processing unit nodes with the minimum utilization rate of the encoder core exist, selecting one of the adjacent processing unit nodes as the target adjacent processing unit node according to a first preset rule.

In one embodiment, the obtaining the encoder core usage of all neighboring processing unit nodes includes:

receiving the encoder core utilization rates of all adjacent processing unit nodes;

storing the encoder core utilization rates of all the adjacent processing unit nodes to a local database;

and acquiring the encoder core utilization rates of all the adjacent processing unit nodes from the local database.

In one embodiment, determining that the data information is processed according to a preset processing mode includes:

sending the data information to a local idle encoder core;

controlling the local idle encoder core to process the data information according to the preset processing mode;

selecting the target adjacent processing unit node from all adjacent processing unit nodes according to a second preset rule;

and generating the data packet, wherein the processing action identifier included in the data packet indicates that the processing action of the target neighboring processing unit node on the target data is forwarding, and the target data included in the data packet is the processed data information.

In one embodiment, the method further comprises:

updating the core utilization rate of the encoder;

and sending the updated encoder core usage to the neighboring processing unit nodes.

In one embodiment, the first preset rule includes:

selecting the adjacent processing unit node with the minimum number as the target adjacent processing unit node;

alternatively, the first and second electrodes may be,

selecting the neighboring processing unit node with the largest number as the target neighboring processing unit node;

alternatively, the first and second electrodes may be,

one of the neighboring processing unit nodes is randomly selected as the target neighboring processing unit node.

In one embodiment, the second preset rule includes:

calculating an optimal path between a current processing unit node and a data output node;

determining a processing unit node adjacent to the current processing unit node in the optimal path as the target adjacent processing unit node.

According to a second aspect of the embodiments of the present disclosure, there is provided a data processing method applied to a processing unit node in an encoder network, where the encoder network includes N × M processing unit nodes, each processing unit node is connected to its neighboring processing unit nodes, N is a positive integer greater than 1, M is a positive integer greater than 1: the method comprises the following steps:

Receiving first data packets sent by other processing unit nodes; the first data packet comprises: processing the action identification and the target data; the processing action identifier indicates a processing action of a target neighboring processing unit node on target data in the data packet, the target data including: data information, or processed data information.

And processing the target data according to the processing action corresponding to the processing action identifier.

The data processing method provided by the embodiment of the disclosure is applied to processing unit nodes in an encoder network, wherein the encoder network comprises N multiplied by M processing unit nodes, each processing unit node is connected with the adjacent processing unit node, N is a positive integer greater than 1, and M is a positive integer greater than 1; the data processing method comprises the following steps: receiving first data packets sent by other processing unit nodes; the first data packet includes: processing the action identification and the target data; the processing action identifier indicates a processing action of the target neighboring processing unit node on target data in the data packet, and the target data comprises: data information, or processed data information. And processing the target data according to the processing action corresponding to the preset processing action identifier. Because each processing unit node is connected with the adjacent processing unit node, the interconnection relation among the processing unit nodes is improved, and the previous processing unit node can inform the processing action of the next processing unit node, so that the processing capability of each processing unit node can be fully exerted through the processing mode, and the processing efficiency is improved.

In one embodiment, the processing action identifier indicates that the target neighboring processing unit node performs a processing action on the target data as processing the target data according to the preset processing mode, where the target data includes: the data information; the processing the target data according to the processing action corresponding to the processing action identifier includes:

processing the data information according to the preset processing mode:

detecting whether a current processing unit node is a data output node or not;

if the current processing unit node is a data output node, outputting the processed data information;

if the current processing unit node is not the data output node, selecting a target adjacent processing unit node from all adjacent processing unit nodes according to a second preset rule;

sending a second data packet to the target neighboring processing unit node, the second data packet comprising: processing the action identification and the target data; and the processing action identifier indicates the processing action of the target adjacent processing unit node on the target data as forwarding, and the target data is the processed data information.

In one embodiment, the processing action identifier indicates that the target neighboring processing unit node performs a processing action on the target data as forwarding, the target data is the processed data information, and the processing the target data according to the processing action corresponding to the processing action identifier includes:

detecting whether a current processing unit node is a data output node or not;

sending the first data packet to the target neighboring processing unit node.

In one embodiment, the processing action identifier indicates that the processing action of the target neighboring processing unit node on the target data is to find an available target neighboring processing unit node; the target data is the data information, and the processing the target data according to the processing action corresponding to the processing action identifier includes:

generating a third data packet, where the processing action identifier included in the third data packet indicates that the target neighboring processing unit node performs processing action on the target data as processing the target data according to the preset processing mode, and the target data included in the data packet is the data information;

generating a fourth data packet, where the processing action identifier included in the fourth data packet indicates that the processing action of the target neighboring processing unit node on the target data is to find an available target neighboring processing unit node, and the target data included in the data packet is the data information; so that the target neighboring processing unit node re-executes the step of finding an available target neighboring processing unit node.

According to a third aspect of the embodiments of the present disclosure, there is provided a data processing apparatus, the apparatus is applied to a processing unit node in an encoder network, the encoder network includes N × M processing unit nodes, each processing unit node is connected to its neighboring processing unit nodes, N is a positive integer greater than 1, M is a positive integer greater than 1: the device comprises:

the first receiving module is used for receiving data information;

the determining module is used for determining the processing action of the data information;

the first processing module is used for processing the data information according to the processing action;

a first sending module, configured to send a data packet to a target neighboring processing unit node, where the data packet includes: a processing action identifier and target data, the processing action identifier indicating a processing action of the target neighboring processing unit node on the target data in the data packet, the target data including: the data information, or the processed data information.

In one embodiment, the determining module comprises:

the judgment submodule is used for judging whether an idle encoder core exists or not;

the first determining submodule is used for determining that the data information is processed according to a preset processing mode if the judging submodule judges that the judging submodule has an idle encoder core;

And the second determining submodule is used for determining that the processing action of the data information is to search for an available target adjacent processing unit node if the judging submodule judges that no idle encoder core exists, so that the target adjacent processing unit node processes the data information according to the preset processing mode.

In one embodiment, the determining the processing action on the data information is to find an available target neighboring processing unit node, and the first processing module includes:

the first acquisition submodule is used for acquiring the encoder core utilization rates of all the adjacent processing unit nodes;

the first detection submodule is used for detecting whether idle encoder cores exist in each adjacent processing unit node according to the utilization rate of the encoder cores;

a third determining submodule, configured to determine, when it is detected that an idle encoder core exists in an adjacent processing unit node, that the adjacent processing unit node having the idle encoder core is the target adjacent processing unit node;

the first generation submodule is used for generating the data packet, the processing action identifier included in the data packet indicates the processing action of the target adjacent processing unit node on the target data to be used for processing the target data according to the preset processing mode, and the target data included in the data packet is the data information;

The first selection submodule is used for selecting the target adjacent processing unit node from all the adjacent processing unit nodes according to a first preset rule when detecting that no idle encoder core exists in all the adjacent processing unit nodes;

a second generation submodule, configured to generate the data packet, where the processing action identifier included in the data packet indicates that the processing action of the target neighboring processing unit node on the target data is to find an available target neighboring processing unit node, and the target data included in the data packet is the data information; so that the target neighboring processing unit node re-executes the step of finding an available target neighboring processing unit node.

In one embodiment, the third determination submodule includes:

a first determining subunit, configured to determine, if it is detected that all of the plurality of neighboring processing unit nodes have an idle encoder core, the neighboring processing unit node with the smallest encoder core utilization rate as the target neighboring processing unit node;

and the selecting subunit is configured to, if it is detected that there are a plurality of neighboring processing unit nodes with the smallest core utilization rate of the encoder, select one of the neighboring processing unit nodes as the target neighboring processing unit node according to a first preset rule.

In one embodiment, the first obtaining sub-module includes:

a receiving subunit, configured to receive the encoder core usage rates of all neighboring processing unit nodes;

the storage subunit is used for storing the encoder core utilization rates of all the adjacent processing unit nodes to a local database;

and the acquisition subunit is used for acquiring the encoder core utilization rates of all the adjacent processing unit nodes from the local database.

In one embodiment, the determining the processing action on the data information is to process the data information according to a preset processing mode, and the first processing module includes:

the first sending submodule is used for sending the data information to a local idle encoder core;

the first processing submodule is used for controlling the local idle encoder core to process the data information according to the preset processing mode;

a second selection submodule, configured to select the target neighboring processing unit node from all neighboring processing unit nodes according to a second preset rule;

a third generating sub-module, configured to generate the data packet, where the processing action identifier included in the data packet indicates that the processing action of the target neighboring processing unit node on the target data is forwarding, and the target data included in the data packet is the processed data information.

In one embodiment, the apparatus further comprises:

the updating module is used for updating the core utilization rate of the encoder;

and the second sending module is used for sending the updated utilization rate of the encoder core to the adjacent processing unit node.

In one embodiment, the first preset rule includes:

alternatively, the first and second electrodes may be,

In one embodiment, the second preset rule includes:

According to a fourth aspect of the embodiments of the present disclosure, there is provided a data processing apparatus, the apparatus is applied to a processing unit node in an encoder network, the encoder network includes N × M processing unit nodes, each processing unit node is connected to its neighboring processing unit nodes, N is a positive integer greater than 1, M is a positive integer greater than 1: the device comprises:

The second receiving module is used for receiving the first data packet sent by other processing unit nodes; the first data packet comprises: processing the action identification and the target data; the processing action identifier indicates a processing action of a target neighboring processing unit node on target data in the data packet, the target data including: data information, or processed data information.

And the second processing module is used for processing the target data according to the processing action corresponding to the processing action identifier.

In one embodiment, the processing action identifier indicates that the target neighboring processing unit node performs a processing action on the target data as processing the target data according to the preset processing mode, where the target data includes: the data information; the second processing module comprises:

the second processing submodule is used for processing the data information according to the preset processing mode;

the second detection submodule is used for detecting whether the current processing unit node is a data output node:

the first output submodule is used for outputting the processed data information if the current processing unit node is a data output node;

A third selection submodule, configured to select a target neighboring processing unit node from all neighboring processing unit nodes according to a second preset rule if the current processing unit node is not a data output node;

a second sending submodule, configured to send a second data packet to the target neighboring processing unit node, where the second data packet includes: processing the action identification and the target data; and the processing action identifier indicates the processing action of the target adjacent processing unit node on the target data as forwarding, and the target data is the processed data information.

In one embodiment, the processing action identifier indicates that the target neighboring processing unit node performs a processing action on the target data as forwarding, the target data is the processed data information, and the second processing module includes:

the third detection submodule is used for detecting whether the current processing unit node is a data output node or not;

a second output submodule, configured to output the processed data information if the current processing unit node is a data output node:

the fourth selection submodule is used for selecting a target adjacent processing unit node from all adjacent processing unit nodes according to a second preset rule if the current processing unit node is not the data output node;

And the third sending submodule is used for sending the first data packet to the target adjacent processing unit node.

In one embodiment, the processing action identifier indicates that the processing action of the target neighboring processing unit node on the target data is to find an available target neighboring processing unit node; the target data is the data information, and the second processing module includes:

the second acquisition submodule is used for acquiring the encoder core utilization rates of all the adjacent processing unit nodes;

the fourth detection submodule is used for detecting whether idle encoder cores exist in each adjacent processing unit node according to the utilization rate of the encoder cores;

a fourth determining submodule, configured to determine, when it is detected that an idle encoder core exists in an adjacent processing unit node, that the adjacent processing unit node having the idle encoder core is the target adjacent processing unit node;

a third generation sub-module, configured to generate a third data packet, where the processing action identifier included in the third data packet indicates that the target neighboring processing unit node performs a processing action on the target data as processing the target data according to the preset processing mode, and the target data included in the data packet is the data information;

A fifth selection submodule, configured to, when it is detected that there is no idle encoder core in all the neighboring processing unit nodes, select the target neighboring processing unit node from all the neighboring processing unit nodes according to a first preset rule;

a fourth generating sub-module, configured to generate a fourth data packet, where the processing action identifier included in the fourth data packet indicates that the processing action of the target neighboring processing unit node on the target data is to find an available target neighboring processing unit node, and the target data included in the data packet is the data information; so that the target neighboring processing unit node re-executes the step of finding an available target neighboring processing unit node.

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 network topology diagram of an encoder network provided by an embodiment of the present disclosure;

fig. 2 is a flowchart of a data processing method provided by an embodiment of the present disclosure;

Fig. 3 is a flowchart of a data processing method provided by an embodiment of the present disclosure;

fig. 4 is a topology diagram of an encoder network provided by an embodiment of the present disclosure;

FIG. 5 is an interaction diagram of a data processing method provided by an embodiment of the present disclosure;

FIG. 6 is a block diagram of a data processing apparatus provided in an embodiment of the present disclosure;

fig. 7 is a block diagram of a determination module in a data processing apparatus according to an embodiment of the present disclosure;

fig. 8 is a structural diagram of a first processing module in a data processing apparatus according to an embodiment of the present disclosure:

fig. 9 is a block diagram of a third determination submodule in the data processing apparatus according to the embodiment of the present disclosure;

fig. 10 is a structural diagram of a first obtaining sub-module in a data processing apparatus according to an embodiment of the present disclosure;

fig. 11 is a block diagram of a first processing module in a data processing apparatus according to an embodiment of the present disclosure;

fig. 12 is a block diagram of a data processing apparatus provided in an embodiment of the present disclosure;

fig. 13 is a block diagram of a data processing apparatus provided in an embodiment of the present disclosure;

fig. 14 is a block diagram of a second processing module in a data processing apparatus according to an embodiment of the present disclosure;

fig. 15 is a block diagram of a second processing module in a data processing apparatus according to an embodiment of the present disclosure;

Fig. 16 is a structural diagram of a second processing module in the data processing apparatus according to the embodiment of the 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 present disclosure, as detailed in the appended claims.

The present disclosure provides an encoder network with a mesh structure, fig. 1 is a network topology diagram of an encoder network provided in an embodiment of the present disclosure, where the encoder network is composed of N × M processing unit nodes (process elements, abbreviated as: PE) with encoding processing capability, where N is a positive integer greater than 1, and M is a positive integer greater than 1; n and M may be the same or different. Each PE of the N × M PEs may be a single-core PE or a multi-core PE. The N × M PEs are arranged in a mesh structure, and each PE is connected to its neighboring PE. Any one of the peripheral nodes in the encoder network (the node connected by the double arrow lines in fig. 1 is the peripheral node) may be used as the data input node and/or the data output node, and in practical applications, at least one peripheral node in the encoder network is set as the data input node and at least one peripheral node is set as the data output node.

The disclosed embodiment provides a data processing method, fig. 2 is a flowchart of the data processing method provided by the disclosed embodiment, the method is applied to processing unit nodes in an encoder network, the encoder network comprises N × M processing unit nodes, each processing unit node is connected with its adjacent processing unit node, N is a positive integer greater than 1, and M is a positive integer greater than 1; the data processing method comprises the following steps:

101. data information is received.

102. A processing action on the data information is determined.

103. And processing the data information according to the preset processing action.

104. Sending a data packet to a target adjacent processing unit node, wherein the data packet comprises a processing action identifier and target data, the processing action identifier indicates a processing action of the target adjacent processing unit node on the target data in the data packet, and the target data comprises: data information, or processed data information.

The method provided by this embodiment is applied to a processing unit node in an encoder network, where the processing unit node is a processing unit node that first receives original data information to be processed in the encoder network, and may also be understood as: the processing unit node is the processing unit node that first receives the original data information to be processed in the data input nodes in the encoder network.

According to the Encoder network, the encoding capacity is improved by increasing the number of PEs or increasing the number of cores of encoders (English: encoders), and meanwhile, each processing unit node is connected with the adjacent processing unit nodes to achieve information intercommunication between the adjacent processing unit nodes, so that the processing capacities of the PEs can be fully exerted, and the data processing efficiency is improved. In the new encoder network, a new scheduling policy of processing unit nodes is proposed, when a processing unit node receives data information, it first determines a processing action on target data included in the data information, after determining the processing action on the target data, it processes the target data according to the processing action, and after processing the data information according to a preset processing action, it sends a data packet to a target neighboring processing unit node, where the data packet includes: the processing action identification and the target data, the processing action identification indicates the processing action of the target adjacent processing unit node on the target data in the data packet, and the target data comprises: and the data information or the processed data information enables the target adjacent processing unit node to execute corresponding processing on the target data in the data packet according to the processing action identifier carried in the data packet.

For example, the data format of the data packet may be as shown in table 1, where the data packet is composed of a header portion head and a data portion payload, and the header portion is added with a processing action identifier; payload is added with the target data.

TABLE 1

head

payload

In an alternative embodiment, the head may contain two bits of data, and the meaning represented by the different values is as follows:

00: means none;

01: representing the target data to be processed according to a preset processing mode;

10: indicating forwarding;

11: indicating finding an available target neighboring processing unit node.

Therefore, the target neighboring processing unit node that receives the corresponding data packet can determine whether its own processing action on the target data in the payload is to be processed in a preset processing manner, or to be forwarded, or to find an available target neighboring processing unit node, according to the value contained in the head.

Since the data information received by the processing unit node in this embodiment is original data information, at this time, for example: which may be data received directly from memory, then at this point the above step 102 comprises the following sub-steps a 1-A3:

at a1, it is determined whether or not there is a free encoder core.

In a2, if any, the processing action on the data information is determined as processing the data information in a preset processing manner.

In a3, if not, the processing action on the data information is determined as finding an available target neighboring processing unit node, so that the target neighboring processing unit node processes the data information according to the preset processing mode.

After the current processing unit node receives the data information, whether an idle Encoder (English: Encoder) core exists or not (namely, whether the data information can be processed or not) is judged, if the idle Encoder core exists, the processing action of the current processing unit node on the data information is determined to be processing the data information according to a preset processing mode, if the idle Encoder core does not exist, an available target adjacent processing unit node is searched for the data information from adjacent processing unit nodes adjacent to the current processing unit node, and the target adjacent processing unit node processes the data information according to the preset processing mode.

If the data information can be processed by the processing unit, the processing unit processes the data information by the processing unit, and if the data information cannot be processed by the processing unit, the target adjacent processing unit node which can process the data information can be actively searched, so that the problem that the data information cannot be timely and effectively processed is solved.

In one implementation, the step A3 of determining a processing action for the data information as finding an available target neighboring processing unit node, at which point the data information is processed according to a preset processing action, includes the following sub-steps a31-a 36:

and A31, acquiring the encoder core utilization rate of all the adjacent processing unit nodes.

In order to find an available target neighboring processing unit node, a neighboring processing unit node having a free Encoder core needs to be found, and then, the Encoder core usage rates of all neighboring processing unit nodes need to be acquired at this time.

Before or after step a31, the following sub-steps are included to obtain the encoder core usage of all neighboring processing unit nodes:

and A311, receiving the encoder core utilization rates of all the adjacent processing unit nodes.

And A312, storing the encoder core utilization rates of all the adjacent processing unit nodes to a local database.

At this time, step a31 includes the following sub-steps:

and A313, acquiring the encoder core utilization rates of all the adjacent processing unit nodes from the local database.

In an implementation manner, before executing the method of the present disclosure, the Encoder core usage rates of all neighboring processing unit nodes adjacent to the current processing unit node may be obtained, and then the Encoder core usage rates of all neighboring processing unit nodes are stored in the local database.

One way to achieve the encorder core usage of all neighboring processing unit nodes neighboring the current processing unit node is: the processing unit node actively sends a utilization rate acquisition request instruction to an adjacent processing unit node adjacent to the processing unit node, and after receiving the utilization rate acquisition request instruction, the adjacent processing unit node sends the corresponding Encoder core utilization rate; the other realization mode is as follows: and the processing unit node actively sends the corresponding Encoder core utilization rate to all adjacent processing unit nodes.

By the mode, each processing unit node can report the Encoder core utilization rate of the processing unit node to the adjacent processing unit nodes; thus, each processing unit node is aware of the Encoder core usage of all its neighboring processing unit nodes.

Specifically, the Encoder core usage rate can be calculated by the following formula:

Q＝N_{use of}/N_{General assembly}×100％；

Wherein Q is Encoder core usage; n is a radical of_{Use of}Refers to the number of Encoder cores, N, currently on the processing unit node_{General assembly}Refers to the number of total Encoder cores on the current processing unit node.

In one embodiment, since the usage rate of the Encoder core of each processing unit node is changed, after the usage rate of the Encoder core of the processing unit node is updated, the processing unit node actively sends the updated usage rate of the Encoder core to the neighboring processing unit node, so that the neighboring processing unit node can update the information stored in its local database in time.

A32, detecting whether there is idle encoder core in each adjacent processing unit node according to the utilization rate of the encoder core.

Since the Encoder core usage rate is determined according to the Encoder core usage number of the neighboring processing unit nodes and the total Encoder core number of the neighboring processing unit nodes, it is possible to know which neighboring processing unit nodes have idle Encoder cores based on the Encoder core usage rate.

A33, when detecting that the adjacent processing unit node has an idle encoder core, determining the adjacent processing unit node having the idle encoder core as the target adjacent processing unit node.

After detecting that there is an idle encoder core in the neighboring processing unit nodes, the neighboring processing unit nodes of the idle encoder core may process the data information, and at this time, it is determined that there is an idle encoder core as the target neighboring processing unit node.

In one embodiment, if it is detected that all of the plurality of neighboring processing unit nodes have an idle encoder core, the neighboring processing unit node with the smallest encoder core utilization rate is determined as the target neighboring processing unit node.

Reading Encoder core utilization rates of all adjacent processing unit nodes from a local database, and if detecting that one adjacent processing unit node has an idle Encoder core, determining the adjacent processing unit node as a target adjacent processing unit node; if it is detected that a plurality of neighboring processing unit nodes all have an idle Encoder core, a target neighboring processing unit node is further determined from the neighboring processing unit nodes, and specifically, the neighboring processing unit node with the lowest Encoder core utilization rate is found as the target neighboring processing unit node.

By determining the adjacent processing unit node with the lowest Encoder core utilization rate as the target adjacent processing unit node, the processing of each processing unit node can be equalized.

In yet another embodiment, if a plurality of neighboring processing unit nodes with the minimum usage rate of the encoder core is detected, one of the neighboring processing unit nodes is selected as the target neighboring processing unit node according to a first preset rule.

If the Encoder core utilization rate is minimum, one of the adjacent processing unit nodes can be randomly selected as a target adjacent processing unit node, or the adjacent processing unit node with the minimum number is selected as the target adjacent processing unit node; or, the neighboring processing unit node with the largest number is selected as the target neighboring processing unit node among the neighboring processing unit nodes, and so on.

And A34, generating a data packet, wherein the processing action identifier included in the data packet indicates that the processing action of the target adjacent processing unit node on the target data is to process the target data according to a preset processing mode, and the target data included in the data packet is data information.

After the target neighboring processing unit node is selected, the data information needs to be sent to the target neighboring processing unit node, and at this time, the target neighboring processing unit node needs to be notified of the processing action of the target neighboring processing unit node on the data information, so that the target neighboring processing unit node can execute the corresponding processing action on the data information.

In order to inform the target adjacent processing unit node of the processing action of the data information, the current processing unit node generates a data packet, the processing action identifier included in the data packet indicates that the target adjacent processing unit node processes the target data according to a preset processing mode, the target data included in the data packet is the data information, and then the data packet is sent to the target adjacent processing unit node.

Taking table 1 as an example, the processing action flag at this time is 01.

And A35, when detecting that no idle encoder core exists in all the adjacent processing unit nodes, selecting a target adjacent processing unit node from all the adjacent processing unit nodes according to a first preset rule.

A36, generating a data packet, wherein the processing action identifier included in the data packet indicates that the processing action of the target neighboring processing unit node on the target data is to find an available target neighboring processing unit node, and the target data included in the data packet is data information: so that the target neighboring processing unit node re-executes the step of finding an available target neighboring processing unit node.

When detecting that there is no idle encoder core in all the neighboring processing unit nodes, it is required to select a target neighboring processing unit node from all the neighboring processing unit nodes according to a first preset rule, so that the target neighboring processing unit node re-executes the step of searching for an available target neighboring processing unit node. At this time, a data packet is also generated, wherein a processing action identifier included in the data packet indicates that the processing action of the target adjacent processing unit node on the target data is to find an available target adjacent processing unit node, and the target data included in the data packet is data information; so that the target neighboring processing unit node re-executes the step of finding an available target neighboring processing unit node.

Wherein selecting a target neighboring processing unit node from all neighboring processing unit nodes according to a first preset rule comprises:

1. one of the target neighboring processing unit nodes may be randomly selected from all neighboring processing unit nodes.

For example: the neighboring processing unit nodes adjacent to processing unit node 1 are neighboring processing unit node 2, neighboring processing unit node 3, and neighboring processing unit node 4; at this time, one may be randomly selected from among the neighboring processing unit node 2, the neighboring processing unit node 3, or the neighboring processing unit node 4.

2. Selecting the neighboring processing unit node with the smallest number as the target neighboring processing unit node:

continuing with the above example, a neighboring processing unit node 2 may be selected.

3. The neighboring processing unit node with the largest number is selected as the target neighboring processing unit node, and the like.

Continuing with the above example, a neighboring processing unit node 4 may be selected.

In step a35, since the current processing unit node does not find a neighboring processing unit node with a free encoder core, a neighboring processing unit node adjacent to the current processing unit node needs to be made available to continue searching for an available target neighboring processing unit node.

Taking table 1 as an example, the processing action identifier in the packet at this time is 11.

After receiving the data packet, the adjacent processing unit node determines that an available target adjacent processing unit node needs to be found for the target data according to the value 11 in the head, and then the adjacent processing unit node locally obtains the Encoder core utilization rates of all adjacent nodes, finds out the PE with the lowest Encoder core utilization rate as the target adjacent processing unit node, and sends the data packet to the target adjacent processing unit node, wherein the processing action identifier included in the data packet indicates that the target adjacent processing unit node performs processing on the target data according to a preset processing mode, the target data included in the data packet is data information, and then sends the data packet to the target adjacent processing unit node.

If all the adjacent processing unit nodes adjacent to the adjacent processing unit node do not have the adjacent processing unit node with the idle Encoder core, the steps are continuously executed, and the like is repeated until an available target adjacent processing unit node is found.

In this embodiment, a target neighboring processing unit node capable of processing target data is searched from neighboring processing unit nodes neighboring a current processing unit node based on the Encoder core utilization, and if the target neighboring processing unit node is not found, one neighboring processing unit node is controlled to continue to search for the target neighboring processing unit node capable of processing target data from neighboring processing unit nodes until the target neighboring processing unit node is found.

In one embodiment, if it is determined that the data information is processed according to a preset processing mode, the step 103 includes the following sub-steps:

b1, sending the data information to a local idle encoder core;

b2, controlling a local idle encoder core to process the data information according to a preset processing mode;

b3, selecting a target adjacent processing unit node from all adjacent processing unit nodes according to a second preset rule;

for example, the second preset rule includes:

and determining the processing unit node adjacent to the current processing unit node in the optimal path as a target adjacent processing unit node.

After any processing unit node finishes processing the target data according to a preset processing mode, the processing unit node calculates an optimal path from the processing unit node to the data output node, and determines a processing unit node adjacent to the current processing unit node in the optimal path as a target adjacent processing unit node.

And B4, generating a data packet, wherein the processing action identifier included in the data packet indicates the processing action of the target adjacent processing unit node on the target data as forwarding, and the target data included in the data packet is processed data information.

After determining that the processing unit node adjacent to the current processing unit node in the optimal path is the target adjacent processing unit node, sending a data packet to the target adjacent processing unit node, and after receiving the data packet, determining that target data in the data packet needs to be forwarded, thus locally calculating the optimal path between the data packet and the data output node, continuously sending the data packet to the processing unit node adjacent to the current processing unit node on the optimal path, and so on until the target data is output through the data output node.

It should be noted that: since all data output nodes are located in the encoder network, all data output nodes share a memory and can address any unit in the memory, and therefore, one path of data packet can be output through one data output node or different data output nodes.

It is worth noting that:

before, during or after the local idle encoder core processes the data information, the method further comprises the following sub-steps:

updating the core utilization rate of the encoder;

The updated core utilization rate of the encoder can be sent to the adjacent processing unit nodes in time, so that the latest core utilization rate of the encoder is stored in the local database of each processing unit node, and the accuracy of searching the target adjacent processing unit node is improved.

Fig. 3 is a flowchart of a data processing method, where the method is applied to processing unit nodes in an encoder network, where the encoder network includes N × M processing unit nodes, each processing unit node is connected to its neighboring processing unit nodes, N is a positive integer greater than 1, and M is a positive integer greater than 1: the method comprises the following steps:

201. receiving first data packets sent by other processing unit nodes; the first data packet includes: processing the action identification and the target data; the processing action identifier indicates a processing action of the target neighboring processing unit node on target data in the data packet, and the target data comprises: data information, or processed data information.

202. And processing the target data according to the processing action corresponding to the preset processing action identifier.

The method provided by the embodiment is applied to the processing unit node in the encoder network, and the processing unit node receives the data sent by other processing unit nodes, but not receives the data sent by the memory or other external devices.

After the current processing node receives the first data packet sent by the other processing unit nodes, the current processing node processes the target data according to the processing action corresponding to the processing action identifier included in the first data packet.

In one embodiment, the processing action identifier indicates that the target neighboring processing unit node processes the target data in a preset processing manner, where the target data includes: data information; the above step 202 comprises the following sub-steps C1-C5:

at C1, the data information is processed according to a preset processing method.

When the processing action identifier in the first data packet received by the current processing unit node indicates that the processing action of the target adjacent processing unit node on the target data is to process the target data according to the preset processing mode, it indicates that the previous processing unit node does not process the target data, and at this time, the current processing unit node processes the data information according to the preset processing mode.

At C2, it is detected whether the current processing element node is a data output node.

At C3, if the current processing unit node is a data output node, the processed data information is output.

At C4, if the current processing element node is not the data output node, a target neighboring processing element node is selected from all neighboring processing element nodes according to a second predetermined rule.

At C5, a second packet is sent to the target neighboring processing unit node, the second packet including: processing the action identification and the target data; the processing action identifier indicates the processing action of the target adjacent processing unit node on the target data as forwarding, and the target data is processed data information.

After the current processing unit node processes the data information according to a preset processing mode, if the current processing unit node is a data output node, the processed data information is directly output, for example, to a memory; if the node is not a data output node, selecting a target adjacent processing unit node from all adjacent processing unit nodes according to a second preset rule, and then sending a second data packet to the selected target adjacent processing unit node, wherein the second data packet comprises: processing the action identification and the target data; the processing action identifier indicates the processing action of the target adjacent processing unit node on the target data as forwarding, and the target data is processed data information.

The second preset rule in this embodiment is similar to the second preset rule in the above embodiment, and is not described herein again.

In one embodiment, the processing action identifier indicates a processing action of the target neighboring processing unit node on the target data as forwarding, the target data being processed data information, the step 202 includes the following sub-steps D1-D4:

in D1, it is detected whether the current processing element node is a data output node.

In D2, if the current processing unit node is a data output node, the processed data information is output.

At D3, if the current processing element node is not the data output node, a target neighboring processing element node is selected from all neighboring processing element nodes according to a second predetermined rule.

At D4, a first packet is sent to the target neighboring processing unit node.

When the processing action identifier in the first data packet received by the current processing unit node indicates that the target neighboring processing unit node performs processing action on the target data as forwarding, the previous processing unit node is indicated to process the data information according to a preset processing mode, and at this time, the current processing unit node only needs to forward. Before forwarding, whether the data output node is a data output node or not needs to be detected, and if the data output node is the data output node, the processed data information is directly output, for example, the data information is output to a memory; if the processing action identifier included in the first data packet indicates that the target processing unit node processes the target data as forwarding, and the target data included in the first data packet is processed data information.

In one embodiment, the processing action identification indicates that the processing action of the target neighboring processing unit node on the target data is to find an available target neighboring processing unit node; the target data is data information, and the step 202 includes the following sub-steps E1-E6:

in E1, the encoder core usage for all neighboring processing unit nodes is obtained.

In E2, whether there is a free encoder core in each neighboring processing unit node is detected according to the encoder core usage.

In E3, when it is detected that the neighboring processing unit node has an idle encoder core, the neighboring processing unit node having an idle encoder core is determined to be the target neighboring processing unit node.

In E4, a third data packet is generated, where the processing action identifier included in the third data packet indicates that the processing action of the target neighboring processing unit node on the target data is to process the target data according to the preset processing manner, and the target data included in the data packet is data information.

In E5, when it is detected that there is no free encoder core in all neighboring processing unit nodes, a target neighboring processing unit node is selected from all neighboring processing unit nodes according to a first preset rule.

In E6, a fourth data packet is generated, where the processing action identifier included in the fourth data packet indicates that the processing action of the target neighboring processing unit node on the target data is to find an available target neighboring processing unit node, and the target data included in the data packet is data information; so that the target neighboring processing unit node re-executes the step of finding an available target neighboring processing unit node.

The implementation manner of each step in this embodiment is similar to that in the above embodiments, and is not described here again.

In an implementation manner, the processing Unit node may be a Central Processing Unit (CPU).

Fig. 4 is a topological diagram of an encoder network according to an embodiment of the present disclosure, as shown in fig. 4, the encoder network includes 12 PE nodes, where PE1, PE2, PE3, PE4, PE5, PE8, PE9, PE10, PE11, and PE12 at edge positions are peripheral PE nodes. In this example, PE1, PE5, and PE9 are data input nodes, and PE4, PE8, and PE12 are data output nodes.

It should be noted that the input/output node setting manner in fig. 4 is only an exemplary setting manner, and in practical applications, any one or more peripheral PE nodes may be selected as the data input node, and one or more peripheral PE nodes may be selected as the data output node.

It can be seen that the encoder network provided by the present invention has very good scalability, and specifically, new PE nodes can be added around peripheral nodes, so that the range of the entire encoder network will be continuously expanded, and the encoding processing capability can be continuously enhanced.

Taking the preset processing mode as an example of encoding, it can be known from the above description of each implementation that the encoding network provided by the present invention has at least the following characteristics:

1) each data input node can receive one or more paths of data at the same time;

2) each data output node can simultaneously output one or more paths of data coding code streams;

3) each data input node is capable of sending data to its neighboring nodes;

4) each PE node can report the Encoder core utilization rate to the adjacent PE nodes; thus, each PE node is aware of the Encoder core usage of all its neighboring PE nodes.

5) Each path of data is encoded on an Encoder core, and one Encoder core also encodes one path of data:

after encoding the data information, each PE node has the capability of calculating the optimal path, and can transmit the encoded data information to the next PE node adjacent to the optimal path; the optimal path refers to a shortest path from the current PE node to the corresponding data output node.

The following describes the processing principle of the encoder network provided by the present invention by taking fig. 4 as an example and taking target data as image data as an example, as shown in fig. 5, in conjunction with the above embodiments.

Because the encoding process of each path of image data in the encoder network is similar, assuming that one path of image data is input through the PE 1 node, the encoding processing steps for the image data are as follows:

301. the PE 1 node receives image data.

302. The PE 1 node firstly judges whether an idle Encoder core exists according to the current Encoder core utilization rate of the PE 1 node, if the idle Encoder core exists, the PE 1 node sends image data to the local idle Encoder core for processing, and meanwhile the PE 1 node carries out the following processing: binding the image data with the selected Encoder core, updating the utilization rate of the Encoder core per se, and informing the updated Encoder core utilization rate to an adjacent PE node; if no idle Encoder core exists, the Encoder core utilization rates of all adjacent PE nodes are read locally, the PE node with the lowest Encoder core utilization rate is found out, a third data packet is generated (the processing action identifier included in the third data packet indicates the processing action of a target adjacent processing unit node on image data as coding, the target data included in the third data packet is image data), the third data packet is sent to the PE node with the lowest found Encoder core utilization rate, and if a plurality of PE nodes with the lowest Encoder core utilization rate exist, one PE node can be selected according to a first preset rule to serve as the target adjacent PE node for sending the data packet;

Assuming that the PE node with the lowest Encoder core utilization rate found by PE 1 in this example is PE 2, the processing is continued according to step 303.

Assuming that all PE nodes adjacent to PE 1 have no idle cores in this example, the process continues according to step 304 and step 305.

303. The third packet is sent to PE 2.

Specifically, the data format of the third packet is as shown in table 1 above.

Therefore, the PE node receiving the third data packet can determine whether to encode the target data or forward the target data according to the value in the head, or to search for an available target neighboring PE node for the target data by calculation.

304. The PE 1 generates a fourth data packet, wherein a processing action identifier included in the fourth data packet indicates that a target adjacent processing unit node searches for an available target adjacent processing unit node for the processing action of the target data, and the target data included in the data packet is image data; so that the target neighboring processing unit node re-executes the step of finding an available target neighboring processing unit node.

305. And sending the fourth data packet to a PE node adjacent to the fourth data packet, such as PE 5, selected according to the first preset rule.

After receiving the fourth packet, PE 5, according to the value in the head: 11, determining that an available target adjacent PE node needs to be found for the target data, then the PE 5 locally obtains the encor core usage rates of all adjacent nodes, finds out an adjacent PE node with the lowest encor core usage rate as the determined target adjacent PE node, generates a third data packet, and sends the third data packet to the target adjacent PE node (for example, PE 9), and after receiving the third data packet, the PE 9, according to the value in the head: 01 determines that the target data needs to be encoded, and therefore, selects an available Encoder core to encode the target data.

And if no available PE node exists in the adjacent PE nodes of the PE 5, generating a fourth data packet, routing the fourth data packet to one adjacent PE node, continuously searching for an available target adjacent PE node for the target data by the adjacent PE node, and so on until the available target adjacent PE node is found.

After the image data on any one PE node is coded, the corresponding PE node calculates an optimal path from the node to a data output node, generates a second data packet (the second data packet comprises a processing action identifier and target data, the processing action identifier indicates the processing action of a target adjacent processing unit node on the target data as forwarding, the target data is the processed image data), sends the generated second data packet to a next PE node adjacent to the path on the path, and after receiving the second data packet, the next PE node according to the value of the head: 10, determining that the second data packet needs to be forwarded, locally calculating an optimal path from the node to the data output node, and sending the second data packet to a next PE node adjacent to the second data packet on the path, and so on, until the second data packet is output through the data output node.

It should be noted that: since all the image output nodes share the memory and any unit in the memory can be addressed, the data packet forming one path of encoded image data can be output by one image output node or different image output nodes.

Based on the data processing method described in the embodiment corresponding to fig. 2, the following is an embodiment of the apparatus of the present disclosure, which can be used to execute an embodiment of the method of the present disclosure.

The disclosed embodiment provides a data processing apparatus, which is applied to processing unit nodes in an encoder network, wherein the encoder network comprises N × M processing unit nodes, each processing unit node is connected with its adjacent processing unit node, N is a positive integer greater than 1, and M is a positive integer greater than 1: as shown in fig. 6, the apparatus includes:

a first receiving module 41, configured to receive data information;

a determining module 42, configured to determine a processing action on the data information;

a first processing module 43, configured to process the data information according to the processing action;

a first sending module 44, configured to send a data packet to a target neighboring processing unit node, where the data packet includes: a processing action identifier and target data, the processing action identifier indicating a processing action of the target neighboring processing unit node on the target data in the data packet, the target data including: the data information, or the processed data information.

In one embodiment, as shown in fig. 7, the determining module 42 includes:

the judging submodule 421 is configured to judge whether there is an idle encoder core;

the first determining submodule 422 is configured to determine that the data information is processed according to a preset processing mode if the determining submodule determines that the determining submodule has an idle encoder core;

the second determining submodule 423 is configured to determine that the processing action on the data information is to find an available target neighboring processing unit node if the determining submodule determines that the determining submodule does not have an idle encoder core, so that the target neighboring processing unit node processes the data information according to the preset processing mode.

In one embodiment, as shown in fig. 8, the determining the processing action on the data information is to find an available target neighboring processing unit node, and the first processing module 43 includes:

a first obtaining submodule 431, configured to obtain encoder core usage rates of all neighboring processing unit nodes;

a first detection submodule 432, configured to detect whether there is an idle encoder core in each neighboring processing unit node according to the encoder core usage rate;

A third determining submodule 433, configured to determine, when it is detected that an idle encoder core exists in an adjacent processing unit node, that the adjacent processing unit node having the idle encoder core is the target adjacent processing unit node;

the first generating sub-module 434 is configured to generate the data packet, where the processing action identifier included in the data packet indicates that the processing action of the target neighboring processing unit node on the target data is to process the target data according to the preset processing mode, and the target data included in the data packet is the data information;

a first selection submodule 435, configured to, when it is detected that there is no idle encoder core in all neighboring processing unit nodes, select the target neighboring processing unit node from all neighboring processing unit nodes according to a first preset rule;

a second generating submodule 436, configured to generate the data packet, where the processing action identifier included in the data packet indicates that the processing action of the target neighboring processing unit node on the target data is to find an available target neighboring processing unit node, and the target data included in the data packet is the data information; so that the target neighboring processing unit node re-executes the step of finding an available target neighboring processing unit node.

In one embodiment, as shown in FIG. 9, the third determination submodule 433 includes:

a first determining subunit 4331, configured to determine, if it is detected that all of the multiple neighboring processing unit nodes have an idle encoder core, the neighboring processing unit node with the smallest encoder core usage rate as the target neighboring processing unit node;

a selecting subunit 4332, configured to, if it is detected that there are multiple neighboring processing unit nodes with the smallest usage rate of the encoder core, select one of the neighboring processing unit nodes as the target neighboring processing unit node according to a first preset rule.

In one embodiment, as shown in fig. 10, the first obtaining submodule 431 includes:

a receive subunit 4311, configured to receive encoder core usage of all neighboring processing unit nodes;

a saving subunit 4312, configured to save the encoder core usage rates of all the neighboring processing unit nodes to a local database;

an obtaining subunit 4313, configured to obtain, from the local database, the encoder core usage rates of all neighboring processing unit nodes.

In an embodiment, as shown in fig. 11, the determining that the processing action on the data information is to process the data information according to a preset processing manner, the first processing module 43 includes:

A first sending submodule 437, configured to send the data information to a local idle encoder core:

the first processing sub-module 438 is configured to control the local idle encoder core to process the data information according to the preset processing manner;

a second selection submodule 439, configured to select the target neighboring processing unit node from all neighboring processing unit nodes according to a second preset rule;

a third generating sub-module 4310, configured to generate the data packet, where the processing action identifier included in the data packet indicates that the processing action of the target neighboring processing unit node on the target data is forwarding, and the target data included in the data packet is the processed data information.

In one embodiment, as shown in fig. 12, the apparatus further comprises:

an updating module 45, configured to update the encoder core usage rate of the encoder;

and a second sending module 46, configured to send the updated encoder core usage rate to the neighboring processing unit node.

In one embodiment, the first preset rule includes:

Alternatively, the first and second electrodes may be,

alternatively, the first and second electrodes may be,

In one embodiment, the second preset rule includes:

Based on the data processing method described in the embodiment corresponding to fig. 3, the following is an embodiment of the apparatus of the present disclosure, which may be used to execute an embodiment of the method of the present disclosure.

The disclosed embodiment provides a data processing apparatus, which is applied to processing unit nodes in an encoder network, wherein the encoder network comprises N × M processing unit nodes, each processing unit node is connected with its adjacent processing unit node, N is a positive integer greater than 1, and M is a positive integer greater than 1: as shown in fig. 13, the apparatus includes:

a second receiving module 51, configured to receive a first data packet sent by another processing unit node; the first data packet comprises: processing the action identification and the target data; the processing action identifier indicates a processing action of a target neighboring processing unit node on target data in the data packet, the target data including: data information, or processed data information.

And the second processing module 52 is configured to process the target data according to the processing action corresponding to the processing action identifier.

In an embodiment, as shown in fig. 14, the processing action identifier indicates that the target neighboring processing unit node performs a processing action on the target data as processing the target data according to the preset processing manner, where the target data includes: the data information; the second processing module 52 includes:

the second processing submodule 521 is configured to process the data information according to the preset processing mode;

a second detection submodule 522, configured to detect whether a current processing unit node is a data output node;

a first output submodule 523, configured to output the processed data information if the current processing unit node is a data output node;

a third selecting submodule 524, configured to select a target neighboring processing unit node from all neighboring processing unit nodes according to a second preset rule if the current processing unit node is not a data output node;

a second sending submodule 525, configured to send a second data packet to the target neighboring processing unit node, where the second data packet includes: processing the action identification and the target data; and the processing action identifier indicates the processing action of the target adjacent processing unit node on the target data as forwarding, and the target data is the processed data information.

In one embodiment, as shown in fig. 15, the processing action identifier indicates a processing action of the target neighboring processing unit node on the target data as forwarding, the target data is the processed data information, and the second processing module 52 includes:

a third detection submodule 526, configured to detect whether the current processing unit node is a data output node;

a second output submodule 527, configured to output the processed data information if the current processing unit node is a data output node;

a fourth selection submodule 528, configured to select a target neighboring processing unit node from all neighboring processing unit nodes according to a second preset rule if the current processing unit node is not a data output node;

a third sending submodule 529, configured to send the first data packet to the target neighboring processing unit node.

In one embodiment, as shown in FIG. 16, the processing action identifier indicates that the processing action of the target neighboring processing unit node on the target data is to find an available target neighboring processing unit node; the target data is the data information, and the second processing module 52 includes:

A second obtaining submodule 5210, configured to obtain the encoder core usage rates of all neighboring processing unit nodes;

a fourth detection submodule 5211, configured to detect whether there is an idle encoder core in each neighboring processing unit node according to the encoder core usage rate;

a fourth determining submodule 5212, configured to determine, when it is detected that an idle encoder core exists in a neighboring processing unit node, that the neighboring processing unit node having the idle encoder core is the target neighboring processing unit node;

a third generating submodule 5213, configured to generate a third data packet, where the processing action identifier included in the third data packet indicates that the target neighboring processing unit node performs processing action on the target data as processing the target data according to the preset processing mode, and the target data included in the data packet is the data information;

a fifth selecting submodule 5214, configured to, when it is detected that there is no idle encoder core in all the neighboring processing unit nodes, select the target neighboring processing unit node from all the neighboring processing unit nodes according to a first preset rule;

a fourth generating submodule 5215, configured to generate a fourth data packet, where the processing action identifier included in the fourth data packet indicates that the processing action of the target neighboring processing unit node on the target data is to find an available target neighboring processing unit node, and the target data included in the data packet is the data information; so that the target neighboring processing unit node re-executes the step of finding an available target neighboring processing unit node.

Based on the data processing method described in the embodiment corresponding to fig. 2, an embodiment of the present disclosure further provides a computer-readable storage medium, for example, the non-transitory computer-readable storage medium may be a Read Only Memory (ROM), a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like. The storage medium stores computer instructions for executing the data transmission method described in the embodiment corresponding to fig. 2, which is not described herein again.

Based on the data processing method described in the embodiment corresponding to fig. 3, an embodiment of the present disclosure further provides a computer-readable storage medium, for example, the non-transitory computer-readable storage medium may be a Read Only Memory (ROM), a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like. The storage medium stores computer instructions for executing the data transmission method described in the embodiment corresponding to fig. 3, which is not described herein again.

Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.

It will be understood that the present disclosure is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A data processing method is characterized in that the method is applied to processing unit nodes in an encoder network, the encoder network comprises N multiplied by M processing unit nodes, each processing unit node is connected with the adjacent processing unit node, N is a positive integer larger than 1, and M is a positive integer larger than 1; the method comprises the following steps:

receiving data information;

determining a processing action for the data information;

processing the data information according to the processing action;

2. The method of claim 1, wherein the determining the processing action for the data information comprises:

Judging whether the encoder core is idle or not;

if the encoder core is idle, determining that the data information is processed according to a preset processing mode;

and if the encoder core is not idle, determining that the processing action of the data information is to find an available target adjacent processing unit node so as to enable the target adjacent processing unit node to process the data information according to the preset processing mode.

3. The method of claim 2, wherein determining the processing action on the data information to find an available target neighboring processing unit node comprises:

detecting whether idle encoder cores exist in each adjacent processing unit node according to the encoder core utilization rate:

4. The method of claim 3, wherein determining that the neighboring processing unit node having an idle encoder core is the target neighboring processing unit node when detecting that the neighboring processing unit node has an idle encoder core comprises:

5. The method of claim 3, wherein obtaining the encoder core utilization for all neighboring processing unit nodes comprises:

6. The method according to claim 2, wherein determining the processing action on the data information is to process the data information according to a preset processing mode, and the processing the data information according to the processing action includes:

sending the data information to a local idle encoder core;

7. The method of claim 6, further comprising:

updating the core utilization rate of the encoder;

8. The method according to claim 3 or 4, wherein the first preset rule comprises:

alternatively, the first and second electrodes may be,

9. The method according to claim 6 or 7, wherein the second preset rule comprises:

10. A data processing method is characterized in that the method is applied to processing unit nodes in an encoder network, the encoder network comprises N multiplied by M processing unit nodes, each processing unit node is connected with the adjacent processing unit node, N is a positive integer larger than 1, and M is a positive integer larger than 1; the method comprises the following steps:

11. The method of claim 10, wherein the processing action identifier indicates the processing action of the target neighboring processing unit node on the target data as processing the target data according to the preset processing manner, and wherein the target data comprises: the data information; the processing the target data according to the processing action corresponding to the processing action identifier includes:

Processing the data information according to the preset processing mode;

detecting whether a current processing unit node is a data output node or not;

12. The method according to claim 10, wherein the processing action identifier indicates a processing action of the target neighboring processing unit node on the target data as forwarding, the target data is the processed data information, and the processing the target data according to the processing action corresponding to the processing action identifier includes:

detecting whether a current processing unit node is a data output node or not;

sending the first data packet to the target neighboring processing unit node.

13. The method of claim 10, wherein the processing action identifier indicates that the processing action of the target neighboring processing unit node on the target data is to find an available target neighboring processing unit node; the target data is the data information, and the processing the target data according to the processing action corresponding to the processing action identifier includes:

14. A data processing apparatus for use in a processing element node in an encoder network, the encoder network comprising N x M processing element nodes, each processing element node being connected to its neighbouring processing element nodes, N being a positive integer greater than 1, M being a positive integer greater than 1: the device comprises:

The first receiving module is used for receiving data information;

15. A data processing apparatus for use in a processing element node in an encoder network, the encoder network comprising N x M processing element nodes, each processing element node being connected to its neighbouring processing element nodes, N being a positive integer greater than 1, M being a positive integer greater than 1: the device comprises: