# US 2023/0267079 A1 · Processing apparatus, method and system for executing data processing on a plurality of channels

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 2023/0267079 A1
- Document ID: US20230267079A1
- Status: Published application
- Kind: A1
- Application number: 18/089,096
- Filed: 2022-12-27
- Published: 2023-08-24
- Earliest claimed: 2022-02-23
- Assignees: Beijing ESWIN Computing Technology Co., Ltd.
- Applicants: Beijing ESWIN Computing Technology Co., Ltd.
- Inventors: Shilin LUO
- Web page: https://patentgrep.com/patent/US20230267079A1

## CPC

- G06F 9/30145: {Instruction analysis, e.g. decoding, instruction word fields}
- G06F 12/0875: with dedicated cache, e.g. instruction or stack
- G06F 9/30032: {Movement instructions, e.g. MOVE, SHIFT, ROTATE, SHUFFLE}
- G06F 9/30043: {LOAD or STORE instructions; Clear instruction}
- G06F 9/3016: {Decoding the operand specifier, e.g. specifier format}
- G06F 9/345: of multiple operands or results {(addressing multiple banks G06F12/06)}
- G06F 9/383: {Operand prefetching (cache prefetching G06F12/0862)}
- G06F 9/3877: {using a secondary processor, e.g. coprocessor (peripheral processor G06F13/12)}
- G06F 13/28: using burst mode transfer, e.g. direct memory access {DMA}, cycle steal (G06F13/32 takes precedence)
- G06F 2212/452: Instruction code
- Y02D 10/00: Energy efficient computing, e.g. low power processors, power management or thermal management
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- Y02D: CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE

## Abstract

A processing apparatus, a method and a system for executing data processing on a plurality of channels are disclosed. The processing apparatus for executing data processing on a plurality of channels includes: a channel information acquiring circuit, configured to acquire channel information of the plurality of channels; a storing circuit, including a plurality of storage regions corresponding to the plurality of channels, in which the storage regions are configured to store data information for the plurality of channels; a data reading control circuit, configured to read target data information corresponding to the channel information from a target storage region among the plurality of storage regions of the storing circuit, according to the channel information; and a cache circuit, configured to pre-store the target data information read from the target storage region of the storing circuit, by the data reading control circuit, to wait for use in the data processing.

## Claims

### Claim 1

A processing apparatus for executing data processing on a plurality of channels, comprising:
a channel information acquiring circuit, configured to acquire channel information of the plurality of channels;
a storing circuit, comprising a plurality of storage regions corresponding to the plurality of channels, wherein the storage regions are configured to store data information for the plurality of channels;
a data reading control circuit, configured to read target data information corresponding to the channel information from a target storage region among the plurality of storage regions of the storing circuit, according to the channel information; and
a cache circuit, configured to pre-store the target data information read from the target storage region of the storing circuit, by the data reading control circuit, to wait for use in the data processing.

### Claim 2 (depends on 1)

The processing apparatus according to claim 1, wherein the cache circuit comprises a first-in-first-out queue based on combined logic, and
the first-in-first-out queue is configured to enable a request signal for the first-in-first-out queue and data information associated with the request signal to occur in a same clock cycle.

### Claim 3 (depends on 2)

The processing apparatus according to claim 2, wherein the cache circuit is configured to be capable of pre-storing at least two pieces of data information for use in the data processing.

### Claim 4 (depends on 1)

The processing apparatus according to claim 1, wherein the cache circuit comprises a first-in-first-out queue based on timing logic, and
the cache circuit is configured to be capable of pre-storing at least three pieces of data information for use in the data processing.

### Claim 5 (depends on 1)

The processing apparatus according to claim 1, wherein the channel information acquiring circuit is configured to query different channels in at least two adjacent times of Round Robin.

### Claim 6 (depends on 1)

The processing apparatus according to claim 1, wherein the processing apparatus is a direct memory access apparatus, and the data processing comprises a data moving operation.

### Claim 7 (depends on 6)

The processing apparatus according to claim 6, wherein the data moving operation comprises moving of the data information or moving of target data indicated by the data information, the target data is data required by a neural-network processor for executing a neural network operation, and the neural-network processor comprises at least one processing unit array and a plurality of global buffer clusters; and
the data information indicates address information of the target data, the address information is obtained based on a mapping relationship between the target data and at least one processing unit in the processing unit array, and each of the global buffer clusters is configured to receive the moved data information or the moved target data on one piece of channel information among the plurality of pieces of channel information.

### Claim 8 (depends on 6)

The processing apparatus according to claim 6, further comprising:
an interface control circuit, configured to acquire the target data information from the cache circuit, generate a data moving request according to the target data information, and execute the data moving operation according to the data moving request.

### Claim 9 (depends on 6)

The processing apparatus according to claim 6, wherein the channel information acquiring circuit comprises:
a Round Robin scheduling circuit, configured to perform Round Robin and determine a plurality of pieces of channel information to execute the data processing.

### Claim 10 (depends on 9)

The processing apparatus according to claim 9, wherein the Round Robin scheduling circuit is further configured to,
in response to completion of the data moving operation on a current channel information and the data information in the storing circuit being non-empty, execute a next time of Round Robin to determine a next piece of channel information to execute the data moving operation.

### Claim 11 (depends on 1)

A computer system, comprising:
the processing apparatus according to claim 1; and
a neural-network processor, wherein the neural-network processor is electrically coupled to the processing apparatus, and the neural-network processor comprises at least one processing unit array and a plurality of global buffer clusters; and
the data information indicates address information of target data, the address information is obtained based on a mapping relationship between the target data and at least one processing unit in the processing unit array, and each of the global buffer clusters is configured to receive the moved data information or the moved target data indicated by the data information on one piece of channel information among the plurality of pieces of channel information.

### Claim 12 (depends on 11)

The computer system according to claim 11, further comprising:
an interface decoder, configured to receive an initial instruction and decode the initial instruction to obtain the data information, and supply the data information decoded to the processing apparatus.

### Claim 13 (depends on 12)

The computer system according to claim 12, further comprising:
an auxiliary processor, configured to supply the initial instruction to the interface decoder.

### Claim 14 (depends on 13)

The computer system according to claim 13, wherein the auxiliary processor is coupled to the interface decoder through a second data transmission protocol,
the neural-network processor is coupled to the processing apparatus through a first data transmission protocol, and
the first data transmission protocol is different from the second data transmission protocol.

### Claim 15

A scheduling method for executing data processing on a plurality of channels, comprising:
acquiring channel information of the plurality of channels;
reading target data information corresponding to the channel information from a target storage region among a plurality of storage regions corresponding to the plurality of channels, according to the channel information; and
pre-storing the read target data information corresponding to the channel information into a cache circuit to wait for use in the data processing.

### Claim 16 (depends on 15)

The scheduling method according to claim 15, wherein acquiring the channel information of the plurality of channels comprises:
querying different channels in at least two adjacent times of Round Robin.

### Claim 17 (depends on 15)

The scheduling method according to claim 15, wherein the scheduling method is used in a direct memory access apparatus, and the data processing comprises a data moving operation.

### Claim 18 (depends on 17)

The scheduling method according to claim 17, wherein the data moving operation comprises moving of the data information or moving of target data indicated by the data information, the target data is data required by a neural-network processor for executing a neural network operation, and the neural-network processor comprises at least one processing unit array and a plurality of global buffer clusters; and
the data information indicates address information of the target data, the address information is obtained based on a mapping relationship between the target data and at least one processing unit in the processing unit array, and each of the global buffer clusters is configured to receive the moved data information or the moved target data on one piece of channel information among the plurality of pieces of channel information.

### Claim 19 (depends on 17)

The scheduling method according to claim 17, wherein acquiring the channel information of the plurality of channels comprises:
performing Round Robin and determining a plurality of pieces of channel information to execute the data processing.

### Claim 20 (depends on 19)

The scheduling method according to claim 19, wherein performing Round Robin and determining the plurality of pieces of channel information to execute the data processing comprises:
executing a next time of Round Robin to determine a next piece of channel information to execute the data moving operation, in response to completion of the data moving operation on a current channel information and data information in a storage region being non-empty.
