Data migration method and system, computer readable storage medium and electronic equipment
By analyzing and processing data transfer task information on the control chip, generating requests that comply with preset protocols, the problem of inefficiency of existing data transfer circuits is solved, and efficient data transfer is achieved to adapt to the chip requirements of high integration and complexity , and reduces costs.
Patent Information
- Application Number
- CN202411204079.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-29
- Publication Date
- 2024-12-06
AI Technical Summary
The existing data transfer circuit scheme is inefficient and requires the individual design of the circuit structure according to different needs, resulting in high time consumption and low efficiency.
Design a data transfer method and system, through the control chip, receive data transfer instructions sent by the superior engine, analyze task information, generate data transfer requests that comply with preset protocols, and send them to the subordinate engine to read the target data from the host It is stored in the local storage area, adopting a multi-function multiplexed circuit structure to reduce chip area and cost and improve resource utilization.
It improves data transfer efficiency, adapts to the needs of high integration, complexity and large-scale chips, and provides a circuit structure that is configurable and multi-functional multiplexed, which greatly reduces chip area and cost and improves resource utilization.
Smart Images

Figure CN119088306A_ABST
Abstract
Description
Technical Field
[0001] The embodiments of the present application relate to the computer field, and more specifically, to a data migration method, system, computer-readable storage medium, and electronic device. Background Art
[0002] The rapid development of social science and technology has also promoted the advancement of computing and storage technology. As people's demand for data storage increases, it has also spawned more and more diverse functions such as data acceleration, data storage and data migration.
[0003] At present, the core idea of computing storage architectures such as general computing acceleration architectures is to use microcode-driven methods to optimize the division of software and hardware interfaces through micro-architecture innovation, so as to take into account the performance of hardware execution, the flexibility and scalability of software programmability. It can be seen that in hardware storage acceleration, the computing storage architecture needs to hardwareize software operations and proximalize remote operations to maximize the release of CPU resources, and the hardware circuits related to data movement are becoming more and more important for improving system performance. The current data movement circuit solution needs to design circuit structures according to different requirements to complete the scene function. For example, the traditional NVME-based storage technology specifically performs corresponding functions at the software layer, but it has to design a separate circuit for each function, and the entire data movement scheduling will consume a lot of time and is very inefficient.
[0004] Therefore, in the related technology, there exists a technical problem of low data migration efficiency.
[0005] There is currently no effective solution to this problem. Summary of the invention
[0006] The embodiments of the present application provide a data migration method, system, computer-readable storage medium and electronic device to at least solve the technical problem of low data migration efficiency in related technologies.
[0007] According to one embodiment of the present application, a data migration method is provided, which is applied to a control chip, wherein the control chip belongs to a data migration system, and the data migration system includes at least an upper-level engine, the control chip, a lower-level engine and a host, including: receiving a data migration instruction sent from the upper-level engine, wherein the data migration instruction carries a migration task; parsing the task information of the migration task to obtain a parsing result; generating a data migration request that complies with a preset protocol according to the parsing result, and sending the data migration request to the lower-level engine, so that the lower-level engine reads the target data from the host and stores the target data in a local storage area.
[0008] In an exemplary embodiment, the control chip includes at least a task information parsing processing unit; the task information of the migration task is parsed to obtain a parsing result, including: parsing the task information by the task information parsing processing unit to obtain a task mode, a host address list first address pointer value, a local address list first address pointer value, a host address list offset value, a local list address offset value, a data migration length value, a host offset value, a local offset value and a split input and output address offset value; determining the parsing result according to the task mode, the host address list first address pointer value, the local address list first address pointer value, the host address list offset value, the local list address offset value, the data migration length value, the host offset value, the local offset value and the split input and output address offset value.
[0009] In an exemplary embodiment, the control chip also includes a host query head pointer unit and a host address splitting and assembly control unit; after parsing the task information of the relocation task to obtain the parsing result, the method also includes: obtaining the host address list first address pointer obtained by the host query head pointer unit from the parsing result; obtaining the current host address list pointer information corresponding to the host address list first address pointer by recursively searching the host address list; sending the current host address list pointer information to the host address splitting and assembly control unit; using the host address splitting and assembly control unit to obtain the host address list first address pointer, host data offset information and the first address number required to execute the relocation task; and generating first relocation information based on the host address list first address pointer, the host data offset information and the first address number required to execute the relocation task, wherein the first relocation information represents the host address single relocation address task information; generating a data relocation request that complies with the preset protocol based on the parsing result, including: generating a data relocation request that complies with the preset protocol based on the first relocation information.
[0010] In an exemplary embodiment, the control chip also includes a local query head pointer unit and a local address splitting and assembly control unit; based on the first movement information, a data movement request that complies with the preset protocol is generated, including: obtaining the local address list first address pointer obtained by the local query head pointer unit from the parsing result; obtaining the current local address list pointer information corresponding to the local address list first address pointer by recursively searching the local address list; sending the current local address list pointer information to the local address splitting and assembly control unit; using the local address splitting and assembly control unit to obtain the local address list first address pointer, local data offset information and the number of second addresses required to execute the movement task; and generating second movement information based on the local address list first address pointer, the local data offset information and the number of second addresses required to execute the movement task, wherein the second movement information represents the local address single movement address task information; generating a data movement request that complies with the preset protocol based on the first movement information and the second movement information.
[0011] In an exemplary embodiment, the control chip also includes an arbitrator unit; based on the first transfer information and the second transfer information, a data transfer request that complies with a preset protocol is generated, including: polling and arbitrating a first task request corresponding to the first transfer information and a second task request corresponding to the second transfer information by the arbitrator unit to obtain an arbitration result; determining a target task request from the arbitration result; and generating a data transfer request of a preset protocol based on the target task request.
[0012] In an exemplary embodiment, the control chip also includes an address information PCIE interface extraction unit, an AXI interface extraction unit, a data splitting and assembling control unit, an internal local address cache unit, and an internal host address cache unit; the data migration system also includes a local information external cache unit, wherein the AXI interface extraction unit and the local information external cache unit are connected; a data migration request of a preset protocol is generated based on the target task request, including: obtaining a bus request that complies with the AXI protocol specification and is generated by the AXI interface extraction unit based on the target task request; reading local address information from the local information external cache unit according to the bus request that complies with the AXI protocol specification; storing the local address information in the internal local address cache unit; receiving a third task request from the host address splitting and assembling control unit; storing the third task request The invention relates to a method for transmitting the address information to the PCIE interface extraction unit; obtaining a bus request that complies with the PCIE protocol specification and is generated by the address information PCIE interface extraction unit based on the third task request; reading host address information from the host according to the bus request that complies with the PCIE protocol specification; storing the host address information in the internal host address cache unit; obtaining the local address information from the internal local address cache unit through the data splitting and assembling control unit, and obtaining the host address information from the internal host address cache unit through the data splitting and assembling control unit; and obtaining a data moving task that is generated by the data splitting and assembling control unit based on the local address information, the host address information, and the data moving information in the parsing result; and generating a data moving request that complies with the preset protocol according to the task type corresponding to the data moving task.
[0013] In an exemplary embodiment, the control chip also includes a direction conversion control unit and a data information PCIE interface extraction unit; generating a data transfer request that complies with the preset protocol according to the transfer task type corresponding to the data transfer task, including: determining a fourth task request corresponding to the transfer task type through the direction conversion control unit, wherein the transfer task type includes one of the following: a task type of moving data from the host to the local storage area, and a task type of moving data from the local storage area to the host; sending the fourth task request to the data information PCIE interface extraction unit; obtaining a bus request that complies with the PCIE protocol specification generated by the data information PCIE interface extraction unit based on the fourth task request, wherein the data transfer request of the preset protocol at least includes the bus request that complies with the PCIE protocol specification; sending the data transfer request to the lower-level engine, including: when determining that the transfer task type is a task type of moving data from the host to the local storage area, sending the bus request that complies with the PCIE protocol specification to the lower-level engine.
[0014] According to another embodiment of the present application, a data migration system is provided, including: an upper-level engine, a control chip, a lower-level engine and a host; wherein the control chip is used to implement the steps in any of the above method embodiments.
[0015] According to another embodiment of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when run.
[0016] According to another embodiment of the present application, an electronic device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.
[0017] According to another embodiment of the present application, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.
[0018] Through the present application, since a data moving system including hardware such as an upper engine, a control chip, a lower engine and a host is pre-set, the control chip of the system can receive a data moving instruction sent from the upper engine, wherein the data moving instruction carries a moving task; the task information of the moving task is parsed to obtain a parsing result; a data moving request that complies with a preset protocol is generated according to the parsing result, and the data moving request is sent to the lower engine, so that the lower engine reads the target data from the host and stores the target data in a local storage area. The present application can cope with the increasingly high integration of chip design, the increasing complexity of chips, the increasing size of chips, the increasing number of integrated modules, the increasing overall performance requirements, and the increasingly fast changing application scenarios. The circuit structure with configurable directions and multi-function multiplexing is provided, and the multi-function multiplexing structure greatly reduces the area of the chip, reduces the cost of the chip, and uses a smaller area to design a more functional data moving circuit, thereby improving the utilization rate of resources. Therefore, the technical problem of low data moving efficiency can be solved, and the effect of improving data moving efficiency can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Figure 1 It is a hardware structure block diagram of a server device of a data migration method according to an embodiment of the present application;
[0020] Figure 2 is a flow chart of a data migration method according to an embodiment of the present application;
[0021] Figure 3 is a schematic diagram of the hardware structure of a data migration system according to an embodiment of the present application;
[0022] Figure 4 is a schematic diagram of a process of multifunctional multiplexed data migration according to an embodiment of the present application;
[0023] Figure 5 This is a schematic diagram of the principle of a host read data migration mode according to an embodiment of the present application (I);
[0024] Figure 6 This is a schematic diagram of the principle of a host read data migration mode according to an embodiment of the present application (II);
[0025] Figure 7 This is a schematic diagram of the principle of a host read-only migration list mode according to an embodiment of the present application (I);
[0026] Figure 8 This is a schematic diagram of the principle of a host read-only migration list mode according to an embodiment of the present application (II);
[0027] Fig. 9This is a schematic diagram of the principle of a host read split IO data migration mode according to an embodiment of the present application (I);
[0028] Fig.10 This is a schematic diagram of the principle of a host read split IO data migration mode according to an embodiment of the present application (II);
[0029] Fig.11 This is a schematic diagram of the principle of a host write data migration mode according to an embodiment of the present application (I);
[0030] Fig.12 This is a schematic diagram of the principle of a host write data migration mode according to an embodiment of the present application (II);
[0031] Fig.13 This is a schematic diagram of the principle of a host write-only data migration mode according to an embodiment of the present application (I);
[0032] Fig.14 This is a schematic diagram of the principle of a host write-only data migration mode according to an embodiment of the present application (II);
[0033] Fig.15 It is a structural diagram of a data migration system according to an embodiment of the present application. DETAILED DESCRIPTION
[0034] The embodiments of the present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.
[0035] It should be noted that the terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.
[0036] The method embodiments provided in the embodiments of the present application can be executed in a server device or a similar computing device. Taking running on a server device as an example, Figure 1 1 is a hardware structure block diagram of a server device of a data migration method according to an embodiment of the present application. Figure 1 As shown, the server device may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the server device may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above server device. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.
[0037] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the data migration method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, to implement the above method. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories may be connected to the server device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.
[0038] The transmission device 106 is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the server device. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.
[0039] Next, some technical terms in this application are explained:
[0040] AXI (Advanced eXtensible Interface) is an on-chip bus protocol oriented towards high performance, high bandwidth, and low latency.
[0041] Buffer is a kind of memory composed of RAM.
[0042] PCIE PCI-Express (peripheral component interconnect express) is a high-speed serial computer expansion bus standard. NVMe is a non-volatile memory host controller interface specification.
[0043] In this embodiment, a data migration method is provided, which is applied to a control chip, wherein the control chip belongs to a data migration system, and the data migration system at least includes an upper-level engine, the control chip, a lower-level engine and a host.
[0044] Figure 2is a flow chart of a data migration method according to an embodiment of the present application. Figure 2 As shown, the process includes the following steps:
[0045] Step S202, receiving a data migration instruction sent from the upper engine, wherein the data migration instruction carries a migration task;
[0046] Step S204, parsing the task information of the moving task to obtain a parsing result;
[0047] Step S206, generating a data migration request that complies with a preset protocol according to the analysis result, and sending the data migration request to the lower-level engine, so that the lower-level engine reads the target data from the host and stores the target data in a local storage area.
[0048] Through the above steps, since the data moving system including hardware such as the upper engine, the control chip, the lower engine and the host is pre-set, the data moving instruction sent from the upper engine can be received through the control chip of this system, wherein the data moving instruction carries the moving task; the task information of the moving task is parsed to obtain the parsing result; a data moving request that complies with the preset protocol is generated according to the parsing result, and the data moving request is sent to the lower engine, so that the lower engine reads the target data from the host and stores the target data in the local storage area. The present application can cope with the increasingly high integration of chip design, the increasing complexity of the chip, the increasing size of the chip, the increasing number of integrated modules, the increasing overall performance requirements, and the increasingly fast changing application scenarios. The circuit structure with configurable direction and multi-function multiplexing is provided, and the multi-function multiplexing structure greatly reduces the area of the chip, reduces the cost of the chip, and uses a smaller area to design a more functional data moving circuit, thereby improving the utilization rate of resources. Therefore, the technical problem of low data moving efficiency can be solved, and the effect of improving data moving efficiency can be achieved.
[0049] Step S202 to step S204 have a fixed execution order, that is, step S202, step S204, and step S206 are executed in sequence.
[0050] Optionally, in one embodiment, the method further includes: generating a data migration request that complies with a preset protocol according to the analysis result, and sending the data migration request to the lower-level engine, so that the lower-level engine reads the stored data from the local storage area and writes the stored data to the host. That is, the present application can realize reading data from the host and writing data to the host.
[0051] For example, Figure 3As shown, the above data migration system includes an upper engine, a control chip, a local information external cache unit, a host and a lower engine. Among them, the control chip also includes a task information parsing processing unit, a local query header pointer unit, a host query header pointer unit, a local address splitting and assembling control unit, a host address splitting and assembling control unit, an RR arbiter unit (i.e., the above arbiter unit), an AXI interface extraction unit, an address information PCIE interface extraction unit, an internal host address cache unit, an internal local address cache unit, a data splitting and assembling control unit, a direction conversion control unit, and a data information PCIE interface extraction unit.
[0052] It should be noted that the different data migration modes supported by this application require clear available environments first. The data migration method of this application is applicable to a variety of chip designs that implement data migration processing. The use of the directionally configurable and multi-functional multiplexed circuit structure in the above-mentioned data migration system can greatly reduce the area of the chip, reduce the cost of the chip, and thus improve resource utilization. Especially in the current situation of significant competition, lower-cost chip design has become increasingly important, and the data migration method of this application can gain the upper hand in seizing a larger market.
[0053] Further, combined with Figure 4 The process shown in the figure explains the process of data migration. The specific steps include:
[0054] Step 1, task information parsing processing unit: The main function is to receive the task information sent by the upper engine module for parsing, extract the task mode, host address list first address pointer (corresponding to the host address list first address pointer value), local address list first address pointer (corresponding to the local address list first address pointer value), host address list offset (corresponding to the host address list offset value), local list address offset (corresponding to the local list address offset value), data transfer length (corresponding to the data transfer length value), host offset (corresponding to the host offset value) and local offset (corresponding to the local offset value), split IO address offset and other information (corresponding to the split input and output address offset value). Send the parsed information to the lower module for processing.
[0055] Among them, the upper-level engine is responsible for issuing task commands and supplying relevant information to the downstream.
[0056] The task information parsing processing unit can parse the source address first pointer, destination address first pointer, amount of data moved, control mode, and report content according to the system scenario. After parsing the information, it checks which of the lower-level subtask processing engines is idle and sends the information to the idle subtask processing engine.
[0057] Step 2, host query header pointer unit and local query header pointer unit: receive the local address list first address pointer and the host address list first address pointer from the task information parsing processing unit respectively, recursively find out the local address list pointer and host address list pointer information actually used according to the local address list offset, and send the information to the lower-level module (host address splitting and assembly control unit / local address splitting and assembly control unit) for processing.
[0058] Step 3, the host address splitting and assembly control unit composes the host address single move address task information (the maximum single move address is 4KB) according to the host list pointer, host data offset information and the number of addresses required for the moved information (corresponding to the first address number) and sends it to the lower-level module (RR arbitrator unit) for processing.
[0059] At the same time, the local address splitting and assembly control unit composes the host address single move address task information (the maximum single move address is 4KB) according to the local list pointer, local data offset information and the number of addresses required for the moved information (corresponding to the second address number) and sends it to the lower-level module (RR arbitrator unit) for processing.
[0060] Step 4, RR arbitrator unit: polls and arbitrates the task requests of the local address splitting and assembly control unit and the host address splitting and assembly control unit, and sends the arbitrated task requests to the lower-level module (AXI interface extraction unit) for processing.
[0061] Step 5, address information PCIE interface extraction unit: receive the task request of the host address splitting and assembling control unit, generate a bus request that complies with the PCIE protocol specification to read the address information from the host, and store the information in the internal host address cache unit.
[0062] Step 6, AXI interface extraction unit: Generate a bus request that complies with the AXI protocol specification to read address information from the local information external cache unit and store it in the internal local address cache unit.
[0063] Step 7, data splitting and assembly control unit: read the host address and local address from the internal local address cache unit and the internal local address cache unit, and then generate a data movement task based on the size of the moved data (data movement length value) and the host offset value and local offset value and other information and send it to the lower-level module (direction conversion control unit) for processing.
[0064] The information such as the size of the moved data, the host offset value, and the local offset value represents the data movement information in the above analysis result.
[0065] The local information external cache unit is a memory for storing the local address list.
[0066] The internal local address cache unit refers to a cache storing local addresses inside the control chip.
[0067] The internal host address cache unit refers to a cache storing the host address inside the control chip.
[0068] Step 8, the direction conversion control unit can determine the corresponding task type according to the host mode. If it is determined to be a host read mode, it is a task type of moving data from the host to the local, and the address direction information is not adjusted, and the data is moved from the host to the local; if it is determined to be a host write mode, it is a task type of moving data from the local to the host, and the address direction information is adjusted, and the data is moved from the local to the host. And the corresponding reorganized data moving task information and the task request corresponding to the reorganized data moving task information are sent to the lower-level module (data information PCIE interface extraction unit) for processing.
[0069] Step 9, data information PCIE interface extraction unit: In the host read mode, the receiving direction conversion control unit generates a bus request that complies with the PCIE protocol specification based on the task request sent by the reorganized data migration task information, and sends it to the lower-level engine to control it to read data from the host to the local. In the host write mode, the data information PCIE interface extraction unit uses the PCIE interface of the lower-level engine to write local data into the host.
[0070] The host refers to a component that stores host address information and data.
[0071] The lower-level engine represents a component that executes the PCIE command. The lower-level processing engine corresponding to the lower-level engine is responsible for executing the command and performing data movement processing.
[0072] Through the above content, a circuit structure with configurable direction and multi-function multiplexing is provided, which unloads data calculation from the host CPU to the data processing acceleration unit close to the storage unit, reduces the corresponding system data movement operation, greatly reduces the chip area, greatly reduces the circuit area and chip cost, and improves the resource utilization. Thus, the system performance is released as much as possible. And by providing a multi-function multiplexing solution for the host read direction, a computing engine can simultaneously realize the same-direction data movement mode of the host read data movement mode, the host read only move list mode and the host read split IO data movement mode. And by providing a multi-function multiplexing solution for the host write direction, a computing engine can simultaneously realize the same-direction data movement mode of the host write data movement mode and the host write only data movement mode. These five data movement modes have certain similarities in hardware structure, that is, they all need to use functions such as data parsing and data splitting. Therefore, a solution with any configurable direction can be provided, so that a computing engine can realize any five working modes, and can also realize data movement in two directions of host read and host write. By multiplexing these identical circuits and adding control modules to switch other functions, the goal of implementing five different modes with one hardware can be achieved.
[0073] In an exemplary embodiment, the control chip at least includes a task information parsing processing unit; the process of parsing the task information of the migration task and obtaining the parsing result may specifically include: parsing the task information through the task information parsing processing unit to obtain the task mode, the host address list first address pointer value, the local address list first address pointer value, the host address list offset value, the local list address offset value, the data migration length value, the host offset value, the local offset value and the split input and output address offset value; determining the parsing result according to the task mode, the host address list first address pointer value, the local address list first address pointer value, the host address list offset value, the local list address offset value, the data migration length value, the host offset value, the local offset value and the split input and output address offset value. This embodiment can parse the task information and obtain the different contents required for data migration, so as to facilitate the realization of multiple data transfer modes.
[0074] In an exemplary embodiment, the control chip also includes a host query head pointer unit and a host address splitting and assembly control unit; after parsing the task information of the relocation task and obtaining the parsing result, the host address list first address pointer obtained by the host query head pointer unit from the parsing result can also be obtained; by recursively searching the host address list, the current host address list pointer information corresponding to the host address list first address pointer is obtained; the current host address list pointer information is sent to the host address splitting and assembly control unit; the host address splitting and assembly control unit is used to obtain the host address list first address pointer, host data offset information and the first address number required to execute the relocation task; and first relocation information is generated according to the host address list first address pointer, the host data offset information and the first address number required to execute the relocation task, wherein the first relocation information represents the host address single relocation address task information; a data relocation request that complies with a preset protocol is generated according to the parsing result, including: generating a data relocation request that complies with a preset protocol based on the first relocation information. This embodiment generates first migration information based on the current host address list pointer information in the parsing result, and then uses the first migration information to generate a data migration request that complies with a preset protocol to achieve migration of host data.
[0075] In an exemplary embodiment, the control chip also includes a local query head pointer unit and a local address splitting and assembly control unit; further proposed is a specific implementation process for generating a data move request that complies with a preset protocol based on the first move information, including the following steps: obtaining the local address list first address pointer obtained by the local query head pointer unit from the parsing result; obtaining the current local address list pointer information corresponding to the local address list first address pointer by recursively searching the local address list; sending the current local address list pointer information to the local address splitting and assembly control unit; using the local address splitting and assembly control unit to obtain the local address list first address pointer, local data offset information and the number of second addresses required to execute the move task; and generating a second move information based on the local address list first address pointer, the local data offset information and the number of second addresses required to execute the move task, wherein the second move information represents the local address single move address task information; generating a data move request that complies with the preset protocol based on the first move information and the second move information. Through this embodiment, a data migration request that complies with the preset protocol can be generated according to the first migration information and the second migration information, thereby realizing data migration from the host to the local, or realizing data migration from the local to the host.
[0076] In an exemplary embodiment, the control chip also includes an arbitrator unit; for the technical solution of generating a data transfer request that complies with a preset protocol based on the first transfer information and the second transfer information, the following implementation steps are proposed: the arbitrator unit performs polling arbitration on the first task request corresponding to the first transfer information and the second task request corresponding to the second transfer information to obtain an arbitration result; determines a target task request from the arbitration result; and generates a data transfer request of a preset protocol based on the target task request.
[0077] In an exemplary embodiment, the control chip also includes an address information PCIE interface extraction unit, an AXI interface extraction unit, a data splitting and assembly control unit, an internal local address cache unit, and an internal host address cache unit; the data migration system also includes a local information external cache unit, wherein the AXI interface extraction unit and the local information external cache unit are connected; the implementation scheme of generating a data migration request of a preset protocol based on the target task request is described through the following content: obtaining a bus request that complies with the AXI protocol specification and is generated by the AXI interface extraction unit based on the target task request; reading local address information from the local information external cache unit according to the bus request that complies with the AXI protocol specification; storing the local address information in the internal local address cache unit; receiving a third task request from the host address splitting and assembly control unit; storing the The third task request is sent to the address information PCIE interface extraction unit; a bus request that complies with the PCIE protocol specification and is generated by the address information PCIE interface extraction unit based on the third task request is obtained; host address information is read from the host according to the bus request that complies with the PCIE protocol specification; the host address information is stored in the internal host address cache unit; the local address information is obtained from the internal local address cache unit through the data splitting and assembling control unit, and the host address information is obtained from the internal host address cache unit through the data splitting and assembling control unit; and a data moving task is generated by the data splitting and assembling control unit based on the local address information, the host address information, and the data moving information in the parsing result; a data moving request that complies with the preset protocol is generated according to the task type corresponding to the data moving task.
[0078] In an exemplary embodiment, the control chip also includes a direction conversion control unit and a data information PCIE interface extraction unit; further proposed is a process for implementing a data transfer request that complies with a preset protocol according to a transfer task type corresponding to the data transfer task, including: determining a fourth task request corresponding to the transfer task type through the direction conversion control unit, wherein the transfer task type includes one of the following: a task type of moving data from the host to the local storage area, and a task type of moving data from the local storage area to the host; sending the fourth task request to the data information PCIE interface extraction unit; obtaining a bus request that complies with the PCIE protocol specification generated by the data information PCIE interface extraction unit based on the fourth task request, wherein the data transfer request of the preset protocol at least includes the bus request that complies with the PCIE protocol specification; sending the data transfer request to the lower-level engine, including: when determining that the transfer task type is a task type of moving data from the host to the local storage area, sending the bus request that complies with the PCIE protocol specification to the lower-level engine.
[0079] Among them, this embodiment can be based on the corresponding task type determined according to the host mode. If it is a host read mode, it is a task type of moving data from the host to the local. If it is a host write mode, it is a task type of moving data from the local storage area to the host.
[0080] Optionally, the method also includes: when determining that the task type is a task type of moving data from the host to the local storage area, sending the bus request that complies with the PCIE protocol specification to the lower-level engine, so that the lower-level engine reads the stored data from the local storage area and writes the stored data to the host.
[0081] Next, combine the attached Figure 5 To Attachment Fig.14 The architecture shown provides a detailed description of the data migration process of this application.
[0082] For the above-mentioned multi-function multiplexing solution in the host read direction, the host read data migration mode, the host read only migration list mode and the host read split IO data migration mode can be implemented by one engine, or different engines can be used to implement each mode separately.
[0083] For solutions that implement different modes through one engine, you can combine Figure 5 , Figure 7 , Fig. 9 To explain.
[0084] like Figure 5As shown, based on the host read data migration mode under the multiplexed data migration hardware structure, the following implementation process is proposed: the task information parsing processing unit receives the task from the upper engine and parses out the host and local first list addresses, data migration length and offset and other information, and sends it to the lower engine for processing. The local query header pointer unit reads the actual local address list pointer according to the local address list first address pointer. At the same time, the host query header pointer unit reads the actual host address list pointer according to the host address list first address pointer. The local address splitting and assembly control unit composes the host address single migration address task information according to the local list pointer, local data offset information and the number of addresses required for the migrated information, and sends it to the lower module for processing (the maximum single migration address is 4KB). (The address table cascades the address list in units of 4KB, and the list pointers can be discretely distributed in the storage space. This solution will read the address at a maximum of 4KB each time until the address required for data movement is completed.) At the same time, the host address splitting and assembly control unit composes the host address single move address task information according to the host list pointer, host data offset information and the number of addresses required for the moved information, and sends it to the lower-level module (the single maximum move address is 4KB) for processing.
[0085] Next, the RR arbiter unit polls the task requests of the arbitration local address splitting and assembly control unit and the host address splitting and assembly control unit, and sends the arbitrated task requests to the lower-level modules for processing (in this mode, the host address reading does not enter this path, so this arbiter is always open to the local address splitting pointer unit). The AXI interface extraction unit reads the local address information and caches it to the internal local address cache unit, while the address information PCIE interface extraction unit reads the host address information and caches it to the internal host address cache unit.
[0086] Subsequently, the data splitting and assembly control unit reads the host address and local address from the internal host address cache unit and the internal local address cache unit, and generates a data movement control task to the lower-level direction conversion control unit according to the size of the moved data. The direction conversion control unit determines that it is the host read mode, does not adjust the address direction information, and moves the data from the host to the local. The data information PCIE interface extraction unit uses the PCIE interface to read the host data into the local memory.
[0087] like Figure 7As shown, based on the host read-only list mode under the multiplexed data migration scheme, the following implementation scheme is proposed: the task information parsing processing unit receives the task from the upper engine and parses the host and local first list address, data migration length and offset and other information, and sends it to the lower engine for processing. The local query head pointer unit reads the actual local address list pointer according to the local address list first address pointer. At the same time, the host query head pointer unit reads the actual host address list pointer according to the host address list first address pointer. The local address splitting and assembly control unit composes the host address single migration address task information according to the local list pointer, local data offset information and the number of addresses required for the migration information and sends it to the lower module for processing (the maximum single migration address is 4KB). (The address table cascades the address list in units of 4KB, and the list pointer can be discretely distributed in the storage space. This scheme will read the address at a maximum of 4KB each time until the address migration required for data migration is completed).
[0088] Next, the RR arbitrator unit polls the task requests of the arbitration local address splitting and assembly control unit and the host address splitting and assembly control unit, and sends the arbitrated task requests to the lower-level modules for processing (in this mode, the host needs to read the host list pointer through AXI and use this path, so this arbitrator needs to poll and arbitrate two upper-level modules).
[0089] The AXI interface extraction unit reads the local address information and caches it into the internal local address cache unit.
[0090] The data splitting and assembly control unit, for the first list move, reads the host address and local address from the host address splitting and assembly control unit and the internal local address cache unit, and generates a data move control task to the lower-level direction conversion control unit according to the size of the moved data.
[0091] Then the direction conversion control unit determines that it is the host read mode, does not adjust the address direction information, and moves data from the host to the local. The data information PCIE interface extraction unit uses the PCIE interface to read the host list into the host memory. When a 4KB list page is moved, the host address splitting and assembly control unit is started, and according to the list moved to the local, the next list pointer is read into the internal host address cache unit through the AXI bus.
[0092] The data splitting and assembly control unit reads the host address and the local address from the internal host address cache unit and the internal local address cache unit for the second and subsequent list migrations, and generates a data migration control task to the lower-level direction conversion control unit according to the size of the migrated data.
[0093] The direction conversion control unit determines that it is in host read mode, does not adjust the address direction information, and moves data from the host to the local.
[0094] The data information PCIE interface extraction unit uses the PCIE interface to read the host list into the local memory.
[0095] like Fig. 9 As shown, based on the host read split IO data migration mode under the multiplexed data migration scheme, the following implementation process is proposed: the task information parsing processing unit receives the task from the upper engine and parses the host and local first list addresses, split IO address offsets, data migration length and offset and other information, and sends it to the lower engine for processing. The local query head pointer unit reads the actual local address list pointer according to the local address list first address pointer. At the same time, the host query head pointer unit reads the actual host address list pointer according to the host address list first address pointer.
[0096] The local address splitting and assembly control unit composes the host address single move address task information according to the local list pointer, local data offset information and the number of addresses required for the moved information, and sends it to the lower-level module for processing (the single maximum move address is 4KB). (The address table is cascaded in units of 4KB, and the list pointer can be discretely distributed in the storage space. This solution will read the address at a maximum of 4KB each time until the address required for data movement is completed).
[0097] At the same time, the host address splitting and assembly control unit first recursively queries the host first list pointer after splitting IO through the PCIE interface according to the host list pointer and split IO address offset information, and then forms the task of moving the host list to the lower-level unit. (The address table is cascaded in units of 4KB, and the list pointer can be discretely distributed in the storage space. This solution will read the address at a maximum of 4KB each time until the address required for data migration is completed).
[0098] The RR arbitrator unit polls the task requests of the local address splitting and assembly control unit and the host address splitting and assembly control unit, and sends the arbitrated task requests to the lower-level modules for processing (in this mode, the host's address read does not enter this path, so this arbitrator is always open to the local address splitting pointer unit).
[0099] The AXI interface extraction unit reads the local address information and caches it in the internal local address cache unit, while the address information PCIE interface extraction unit reads the host address information and caches it in the internal host address cache unit.
[0100] The data splitting and assembling control unit reads the host address and the local address from the internal host address cache unit and the internal local address cache unit, and generates a data moving control task to the lower-level direction conversion control unit according to the size of the moved data.
[0101] The direction conversion control unit determines that it is in host read mode, does not adjust the address direction information, and moves data from the host to the local.
[0102] The data information PCIE interface extraction unit uses the PCIE interface to read the host data into the local memory.
[0103] For each mode, different engines are used to implement each mode separately. Figure 6 , Figure 8 , Fig.10 To explain.
[0104] like Figure 6 As shown, for mode 1 host read data migration mode under the data migration hardware structure, the following implementation process is provided: the task information parsing processing unit receives the task from the upper engine and parses out the host and local first list addresses, data migration length and offset and other information, and sends it to the lower engine for processing.
[0105] Mode 1: The local address control unit reads the local address according to the first list address of the local. Mode 1: The host address control unit reads the host address according to the first list address of the host.
[0106] The AXI interface extraction unit reads the local address information and caches it to the internal host address cache unit.
[0107] The address information PCIE interface extraction unit reads the address information of the host and caches it in the internal host address cache unit.
[0108] Mode 1 data movement assembly control unit reads the host address and local address from the internal host address cache unit and the internal local address cache unit, and generates a data movement control task to the lower-level module (ie, the data information PCIE interface extraction unit) according to the size of the moved data.
[0109] The data information PCIE interface extraction unit uses the PCIE interface to read the host data into the local memory.
[0110] like Figure 8 As shown, the workflow for mode 2 host read-only list mode under the data migration hardware structure is as follows: the task information parsing processing unit receives the task from the upper-level engine and parses out the host and local first list addresses, data migration length and offset and other information, and sends it to the lower-level engine for processing.
[0111] Mode 2: The local address control unit reads the local address according to the local first list address.
[0112] Mode 2: The data transfer assembly control unit generates a list transfer control task to the lower-level module by using the host list head address and the transfer data size.
[0113] Mode 2 Host Address Control Unit When the Mode 2 Data Splitting and Assembly Control Unit consumes the first set of lists, this module will start AXI to read the first address of the next list for use by the Data Splitting and Assembly Control Unit.
[0114] The AXI interface extraction unit reads the local address information and caches it in the internal local address cache unit and caches the first address of the host list in the host address cache unit.
[0115] Mode 2 data movement assembly control unit reads the host address and local address from the internal host address cache unit and the internal local address cache unit, and generates a list of movement control tasks to the lower-level module according to the size of the moved data.
[0116] The data information PCIE interface extraction unit uses the PCIE interface to read the host list into the local memory.
[0117] like Fig.10 As shown, for the mode 3 host read split IO data migration mode under the data migration hardware structure, the specific implementation process is as follows:
[0118] The task information parsing processing unit receives tasks from the upper engine and parses out the host and local first list addresses, data transfer length and offset information, and sends it to the lower engine for processing.
[0119] Mode 3: The local address control unit reads the local address according to the local first list address. Mode 3: The host address control unit recursively searches for the first real valid list address pointer according to the host first list address, IO offset length, data transfer length and other information.
[0120] The AXI interface extraction unit reads the local address information and caches it in the internal local address cache unit and caches the first address of the host list in the host address cache unit.
[0121] The address information PCIE interface extraction unit reads the host address into the internal host address cache unit through the PCIE interface.
[0122] Mode three data movement assembly control unit reads the host address and local address from the internal host address cache unit and the internal local address cache unit, and generates split IO movement control tasks to the lower-level modules according to the size of the moved data.
[0123] The data information PCIE interface extraction unit uses the PCIE interface to read the host data into the local memory.
[0124] Similarly, for the above-mentioned host write direction multi-function multiplexing solution, the host write data migration mode and the host write only data migration mode can be implemented by one engine, or different engines can be used to implement each mode separately.
[0125] For solutions that implement different modes through one engine, you can combine Fig.11 , Fig.13 To explain.
[0126] like Fig.11 As shown, based on the host write data migration mode under the multiplexed data migration scheme, the following implementation scheme is proposed: the task information parsing processing unit receives the task from the upper-level engine and parses out the host and local first list addresses, data migration length and offset and other information, and sends it to the lower-level engine for processing.
[0127] The local query head pointer unit reads the actual local address list pointer according to the first address pointer of the local address list. Meanwhile, the host query head pointer unit reads the actual host address list pointer according to the first address pointer of the host address list.
[0128] The local address splitting and assembly control unit composes the host address single move address task information according to the local list pointer, local data offset information, and the number of addresses required for the moved information, and sends it to the lower-level module (the maximum single move address is 4KB) for processing. (The address table cascades the address list in units of 4KB, and the list pointer can be discretely distributed in the storage space. This solution will read the address with a maximum of 4KB each time until the address required for data movement is completed). At the same time, the host address splitting and assembly control unit composes the host address single move address task information according to the host list pointer, host data offset information, and the number of addresses required for the moved information, and sends it to the lower-level module (the maximum single move address is 4KB) for processing.
[0129] The RR arbitrator unit polls the task requests of the local address splitting and assembly control unit and the host address splitting and assembly control unit, and sends the arbitrated task requests to the lower-level modules for processing (in this mode, the host's address read does not enter this path, so this arbitrator is always open to the local address splitting pointer unit).
[0130] The AXI interface extraction unit reads the local address information and caches it in the internal local address cache unit, while the address information PCIE interface extraction unit reads the host address information and caches it in the internal host address cache unit.
[0131] The data splitting and assembling control unit reads the host address and the local address from the internal host address cache unit and the internal local address cache unit, and generates a data moving control task to the lower-level direction conversion control unit according to the size of the moved data.
[0132] The direction conversion control unit determines that it is the host write mode, adjusts the address direction information, and moves the local data to the host.
[0133] The data information PCIE interface extraction unit uses the PCIE interface to write local data into the host.
[0134] like Fig.13 As shown, based on the host write-only data migration mode under the multiplexed data migration solution, the following implementation steps can be proposed:
[0135] The task information parsing processing unit receives tasks from the upper engine and parses out the host and local first list addresses, data transfer length and offset information, and sends it to the lower engine for processing.
[0136] The local query head pointer unit reads the actual local address list pointer according to the first address pointer of the local address list. Meanwhile, the host query head pointer unit reads the actual host address list pointer according to the first address pointer of the host address list.
[0137] The local address splitting and assembly control unit composes the local address single move address task information according to the local list pointer, local data offset information, and the number of addresses required for the moved information, and sends it to the lower-level module for processing (the single maximum move address is 4KB). (The address table cascades the address list in units of 4KB, and the list pointer can be discretely distributed in the storage space. This solution will read the address at a maximum of 4KB each time until the address required for data movement is completed). At the same time, the host address splitting and assembly control unit generates a list move request that complies with the AXI protocol based on the host list pointer, host data offset information, and the number of addresses required for the moved information, and sends it to the lower-level module for processing.
[0138] The RR arbitrator unit polls and arbitrates the task requests of the local address splitting and assembly control unit and the host address splitting and assembly control unit, and sends the arbitrated task requests to the lower-level modules for processing (in this mode, the host needs to read the host address through AXI and needs to use this path, so this arbitrator needs to poll and arbitrate the two upper-level modules).
[0139] The AXI interface extraction unit reads the local address information cached into the internal local address cache unit and reads the host address information cached into the host address cache unit.
[0140] The data splitting and assembling control unit reads the host address and the local address from the internal host address cache unit and the internal local address cache unit, and generates a data moving control task to the lower-level direction conversion control unit according to the size of the moved data.
[0141] The direction conversion control unit determines that it is in host write mode, adjusts the address direction information, and moves data from the local to the host.
[0142] The data information PCIE interface extraction unit uses the PCIE interface to write local data into the host.
[0143] Similarly, different engines can be used to implement each mode separately. Fig.12 , Fig.14 To explain.
[0144] like Fig.12 As shown, for the mode 4 host write data migration mode under the data migration hardware structure, the specific implementation is as follows:
[0145] The task information parsing processing unit receives tasks from the upper engine and parses out the host and local first list addresses, data transfer length and offset information, and sends it to the lower engine for processing.
[0146] Mode 4: The local address control unit reads the local address according to the local first list address. Mode 4: The host address control unit reads the host address according to the host first list address.
[0147] The AXI interface extraction unit reads the local address information and caches it into the internal local address cache unit.
[0148] The address information PCIE interface extraction unit reads the address information of the host and caches it in the internal local address cache unit.
[0149] Mode 4 data movement assembly control unit reads the host address and local address from the internal host address cache unit and the internal local address cache unit, and generates a data movement control task to the lower-level module according to the size of the moved data.
[0150] The data information PCIE interface extraction unit uses the PCIE interface to write local data into the host memory.
[0151] like Fig.14 As shown, for mode 5 under the data migration hardware structure, the following implementation process is proposed: the task information parsing processing unit receives the task from the upper engine and parses out the host and local first list addresses, data migration length and offset and other information, and sends it to the lower engine for processing.
[0152] Mode 5: The local address control unit reads the local address according to the local first list address. Mode 5: The host address control unit reads the host address according to the host first list address (reads the host address through the AXI interface).
[0153] The AXI interface extraction unit reads the local address information cached into the internal local address cache unit and reads the host address information into the host address cache unit.
[0154] Mode five data movement assembly control unit reads the host address and local address from the internal host address cache unit and the internal local address cache unit, and generates a data movement control task to the lower-level module according to the size of the moved data.
[0155] The data information PCIE interface extraction unit uses the PCIE interface to write local data into the host memory.
[0156] Based on the above embodiments, it can be seen that compared with Figure 6 , Figure 8 , Fig.10 , Fig.12 and Fig.14 The embodiment shown requires five sets of hardware circuits to implement the five mode functions, based on Figure 5 , Figure 7 , Fig. 9 , Fig.11 and Fig.13 In the embodiment shown, only one set of circuit system is needed to realize the five modes. It can be seen that the parallel mechanism based on host address, local address and host data migration in this application improves the parallel capability of the computing engine. In addition, a hierarchical control mechanism is provided for the host first address list, local first address list, host address migration and local address migration to improve the pipeline capability of the computing engine. At the same time, it occupies one fifth of the area of the old solution, greatly reducing the chip area.
[0157] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods described in each embodiment of the present application.
[0158] In this embodiment, a data migration system is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.
[0159] Fig.15 is a structural diagram of a data migration system according to an embodiment of the present application, such as Fig.15 As shown, the system includes: an upper-level engine 1502 , a control chip 1504 , a lower-level engine 1506 and a host 1508 .
[0160] Among them, the control chip is used to receive data movement instructions sent from the upper-level engine, wherein the data movement instructions carry a movement task; parse the task information of the movement task to obtain a parsing result; generate a data movement request that complies with a preset protocol based on the parsing result, and send the data movement request to the lower-level engine, so that the lower-level engine reads the target data from the host and stores the target data in a local storage area.
[0161] Through the above system, since the data moving system including hardware such as the upper engine, the control chip, the lower engine and the host is pre-set, the data moving instruction sent from the upper engine can be received through the control chip of this system, wherein the data moving instruction carries the moving task; the task information of the moving task is parsed to obtain the parsing result; a data moving request that complies with the preset protocol is generated according to the parsing result, and the data moving request is sent to the lower engine, so that the lower engine reads the target data from the host and stores the target data in the local storage area. The present application can cope with the increasingly high integration of chip design, the increasing complexity of the chip, the increasing size of the chip, the increasing number of integrated modules, the increasing overall performance requirements, and the increasingly fast changing application scenarios. The circuit structure with configurable direction and multi-function reuse is provided, and the multi-function reuse structure greatly reduces the area of the chip, reduces the cost of the chip, and uses a smaller area to design a more functional data moving circuit, thereby improving the utilization rate of resources. Therefore, the technical problem of low data moving efficiency can be solved, and the effect of improving data moving efficiency can be achieved.
[0162] In an exemplary embodiment, the control chip includes at least a task information parsing processing unit; the control chip is also used to: parse the task information through the task information parsing processing unit to obtain the task mode, the host address list first address pointer value, the local address list first address pointer value, the host address list offset value, the local list address offset value, the data movement length value, the host offset value, the local offset value and the split input and output address offset value; determine the parsing result according to the task mode, the host address list first address pointer value, the local address list first address pointer value, the host address list offset value, the local list address offset value, the data movement length value, the host offset value, the local offset value and the split input and output address offset value.
[0163] In an exemplary embodiment, the control chip also includes a host query head pointer unit and a host address splitting and assembly control unit; the control chip is also used to parse the task information of the relocation task and obtain the parsing result, then obtain the host address list first address pointer obtained by the host query head pointer unit from the parsing result; obtain the current host address list pointer information corresponding to the host address list first address pointer by recursively searching the host address list; send the current host address list pointer information to the host address splitting and assembly control unit; use the host address splitting and assembly control unit to obtain the host address list first address pointer, host data offset information and the first address number required to execute the relocation task; and generate first relocation information based on the host address list first address pointer, the host data offset information and the first address number required to execute the relocation task, wherein the first relocation information represents the host address single relocation address task information; generate a data relocation request that complies with a preset protocol based on the parsing result, including: generating a data relocation request that complies with a preset protocol based on the first relocation information.
[0164] In an exemplary embodiment, the control chip also includes a local query head pointer unit and a local address splitting and assembly control unit; the control chip is also used to: obtain the local address list first address pointer obtained by the local query head pointer unit from the parsing result; obtain the current local address list pointer information corresponding to the local address list first address pointer by recursively searching the local address list; send the current local address list pointer information to the local address splitting and assembly control unit; use the local address splitting and assembly control unit to obtain the local address list first address pointer, local data offset information and the number of second addresses required to perform the relocation task; and generate second relocation information based on the local address list first address pointer, the local data offset information and the number of second addresses required to perform the relocation task, wherein the second relocation information represents the local address single relocation address task information; generate a data relocation request that complies with a preset protocol based on the first relocation information and the second relocation information.
[0165] In an exemplary embodiment, the control chip also includes an arbitrator unit; the control chip is also used to: perform polling arbitration on a first task request corresponding to the first transfer information and a second task request corresponding to the second transfer information through the arbitrator unit to obtain an arbitration result; determine a target task request from the arbitration result; and generate a data transfer request of a preset protocol based on the target task request.
[0166] In an exemplary embodiment, the control chip also includes an address information PCIE interface extraction unit, an AXI interface extraction unit, a data splitting and assembly control unit, an internal local address cache unit, and an internal host address cache unit; the data migration system also includes a local information external cache unit, wherein the AXI interface extraction unit and the local information external cache unit are connected; the control chip is also used to: obtain a bus request that complies with the AXI protocol specification and is generated by the AXI interface extraction unit based on the target task request; read local address information from the local information external cache unit according to the bus request that complies with the AXI protocol specification; store the local address information in the internal local address cache unit; receive a third task request from the host address splitting and assembly control unit; send the third task request to the address information a PCIE interface extraction unit for obtaining the address information; a bus request that complies with the PCIE protocol specification and is generated by the PCIE interface extraction unit based on the third task request; host address information is read from the host according to the bus request that complies with the PCIE protocol specification; the host address information is stored in the internal host address cache unit; the local address information is obtained from the internal local address cache unit through the data splitting and assembling control unit, and the host address information is obtained from the internal host address cache unit through the data splitting and assembling control unit; and a data moving task is obtained that is generated by the data splitting and assembling control unit based on the local address information, the host address information, and the data moving information in the parsing result; a data moving request that complies with the preset protocol is generated according to the task type corresponding to the data moving task.
[0167] In an exemplary embodiment, the control chip also includes a direction conversion control unit and a data information PCIE interface extraction unit; the control chip is also used to: determine a fourth task request corresponding to the moving task type through the direction conversion control unit, wherein the moving task type includes one of the following: a task type of moving data from the host to the local storage area, a task type of moving data from the local storage area to the host; send the fourth task request to the data information PCIE interface extraction unit; obtain a bus request that complies with the PCIE protocol specification generated by the data information PCIE interface extraction unit based on the fourth task request, wherein the data moving request of the preset protocol at least includes the bus request that complies with the PCIE protocol specification; send the data moving request to the lower-level engine, including: when it is determined that the moving task type is a task type of moving data from the host to the local storage area, send the bus request that complies with the PCIE protocol specification to the lower-level engine.
[0168] It should be noted that the above modules can be implemented by software or hardware. For the latter, it can be implemented in the following ways, but not limited to: the above modules are all located in the same processor; or the above modules are located in different processors in any combination.
[0169] An embodiment of the present application further provides a computer-readable storage medium, in which a computer program is stored, wherein the computer program is configured to execute the steps of any of the above method embodiments when running.
[0170] In an exemplary embodiment, the computer-readable storage medium may include, but is not limited to, various media that can store computer programs, such as a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk.
[0171] An embodiment of the present application further provides an electronic device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.
[0172] In an exemplary embodiment, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.
[0173] An embodiment of the present application further provides a computer program product, which includes a computer program. When the computer program is executed by a processor, the steps in any one of the above method embodiments are implemented.
[0174] An embodiment of the present application further provides another computer program product, including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps in any of the above method embodiments are implemented.
[0175] An embodiment of the present application also provides a computer program, which includes computer instructions, which are stored in a computer-readable storage medium; a processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device performs the steps in any one of the above method embodiments.
[0176] For specific examples in this embodiment, reference may be made to the examples described in the above embodiments and exemplary implementation modes, and this embodiment will not be described in detail herein.
[0177] Obviously, those skilled in the art should understand that the above modules or steps of the present application can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that herein, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.
[0178] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.
Claims
1. A data migration method, characterized in that: Applied to a control chip, wherein the control chip belongs to a data migration system, and the data migration system at least includes an upper-level engine, the control chip, a lower-level engine and a host, including: Receiving a data moving instruction sent from the upper engine, wherein the data moving instruction carries a moving task; Parsing the task information of the moving task to obtain a parsing result; A data migration request that complies with a preset protocol is generated according to the analysis result, and the data migration request is sent to the lower-level engine, so that the lower-level engine reads the target data from the host and stores the target data in a local storage area.
2. The method according to claim 1, characterized in that: The control chip at least includes a task information analysis and processing unit; The task information of the moving task is parsed to obtain a parsing result, including: The task information is parsed by the task information parsing processing unit to obtain the task mode, the host address list first address pointer value, the local address list first address pointer value, the host address list offset value, the local list address offset value, the data transfer length value, the host offset value, the local offset value and the split input and output address offset value; The parsing result is determined according to the task mode, the host address list first address pointer value, the local address list first address pointer value, the host address list offset value, the local list address offset value, the data movement length value, the host offset value, the local offset value and the split input and output address offset value.
3. The method according to claim 1, characterized in that: The control chip also includes a host query head pointer unit and a host address splitting and assembling control unit; After parsing the task information of the moving task to obtain the parsing result, the method further includes: Obtaining the host address list head pointer obtained by the host query head pointer unit from the parsing result; By recursively searching the host address list, the current host address list pointer information corresponding to the first address pointer of the host address list is obtained; Sending the current host address list pointer information to the host address disassembly and assembly control unit; The host address disassembly and assembly control unit is used to obtain the first address pointer of the host address list, the host data offset information and the number of first addresses required to perform the migration task; and generating first migration information according to the first address pointer of the host address list, the host data offset information and the number of first addresses required to perform the migration task, wherein the first migration information represents the single migration address task information of the host address; Generating a data migration request that complies with a preset protocol according to the analysis result, including: A data migration request that complies with a preset protocol is generated based on the first migration information.
4. The method according to claim 3, characterized in that The control chip also includes a local query head pointer unit and a local address splitting and assembling control unit; Generating a data migration request that complies with a preset protocol based on the first migration information includes: Obtaining a first address pointer of a local address list obtained by the local query head pointer unit from the parsing result; By recursively searching the local address list, current local address list pointer information corresponding to the first address pointer of the local address list is obtained; Sending the current local address list pointer information to the local address splitting and assembling control unit; The local address disassembly and assembly control unit is used to obtain the first address pointer of the local address list, the local data offset information and the number of second addresses required to perform the migration task; and generating second migration information according to the first address pointer of the local address list, the local data offset information and the number of second addresses required to perform the migration task, wherein the second migration information represents the local address single migration address task information; A data migration request that complies with a preset protocol is generated based on the first migration information and the second migration information.
5. The method according to claim 4, characterized in that The control chip also includes an arbitrator unit; Generating a data migration request that complies with a preset protocol based on the first migration information and the second migration information includes: Performing round-robin arbitration on the first task request corresponding to the first migration information and the second task request corresponding to the second migration information by the arbitrator unit to obtain an arbitration result; Determining a target task request from the arbitration result; A data migration request with a preset protocol is generated based on the target task request.
6. The method according to claim 5, characterized in that The control chip also includes an address information PCIE interface extraction unit, an AXI interface extraction unit, a data splitting and assembly control unit, an internal local address cache unit, and an internal host address cache unit; The data migration system further comprises a local information external cache unit, wherein the AXI interface extraction unit and the local information external cache unit are connected; Generating a data migration request of a preset protocol based on the target task request includes: Acquire a bus request that complies with the AXI protocol specification and is generated by the AXI interface extraction unit based on the target task request; Reading local address information from a local information external cache unit according to the bus request conforming to the AXI protocol specification; Storing the local address information in the internal local address cache unit; receiving a third task request from the host address disassembly and assembly control unit; Sending the third task request to the address information PCIE interface extraction unit; Obtaining a bus request that complies with the PCIE protocol specification and is generated by the address information PCIE interface extraction unit based on the third task request; Reading host address information from the host according to the bus request that complies with the PCIE protocol specification; Storing the host address information in the internal host address cache unit; Acquire the local address information from the internal local address cache unit through the data splitting and assembling control unit, and acquire the host address information from the internal host address cache unit through the data splitting and assembling control unit; And, obtaining a data migration task generated by the data splitting and assembling control unit based on the local address information, the host address information, and the data migration information in the parsing result; A data migration request that complies with a preset protocol is generated according to a task type corresponding to the data migration task.
7. The method according to claim 6, characterized in that The control chip also includes a direction conversion control unit and a data information PCIE interface extraction unit; Generating a data migration request that complies with a preset protocol according to a migration task type corresponding to the data migration task, including: Determining, by the direction conversion control unit, a fourth task request corresponding to the migration task type, wherein the migration task type includes one of the following: a task type for migrating data from the host to the local storage area, and a task type for migrating data from the local storage area to the host; Sending the fourth task request to the data information PCIE interface extraction unit; Acquire the bus request that complies with the PCIE protocol specification and is generated by the data information PCIE interface extraction unit based on the fourth task request, wherein the data movement request of the preset protocol at least includes the bus request that complies with the PCIE protocol specification; Sending the data migration request to the lower-level engine includes: When it is determined that the migration task type is a task type of migrating data from the host to the local storage area, the bus request that complies with the PCIE protocol specification is sent to the lower-level engine.
8. A data migration system, characterized in that: include: Upper engine, control chip, lower engine and host; Wherein, the control chip is used to implement the steps of the method described in any one of claims 1 to 7.
9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein the computer program implements the steps of the method described in any one of claims 1 to 7 when executed by a processor.
10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method described in any one of claims 1 to 7 are implemented.
