A microkernel-based storage-computation-drawing integrated system and method thereof, and a storage medium

Abstract

The application discloses a storage-computation-drawing integrated system and method based on a microkernel and a storage medium, the method obtains task request data packets uploaded by a user terminal and analyzes the task request data packets to obtain a task type; a corresponding task workflow is matched from a configuration file based on the task type; a plurality of microkernels are cooperatively scheduled according to the task workflow matched currently to respectively execute corresponding tasks; a micro storage kernel calls metadata required in a current task execution process, a micro calculation kernel generates a to-be-rendered data set by logically operating on the metadata; and a micro rendering kernel adaptively executes a corresponding rendering mode based on the to-be-rendered data set, obtains a rendering result and outputs the rendering result to the user terminal. The application effectively solves the defects of high delay, large bandwidth occupation and complex architecture of the existing architecture, realizes low-delay efficient data processing, reduces enterprise operation and operation and maintenance costs, simplifies system integration difficulty, and improves system adaptability and practicality to adapt to real-time interaction requirements in multiple fields.

Description

Technical Field

[0001] This invention relates to the field of digital drawing technology, specifically to a microkernel-based integrated storage, computation, and drawing system and method, as well as a storage medium. Background Technology

[0002] With the deepening penetration of digital transformation technologies, fields such as industrial digital twins, smart cities, cloud gaming, online design and collaboration are placing higher demands on the efficiency, scale, and interactive experience of data processing. Current application scenarios generally build data processing workflows around three core links: data storage, business logic calculation, and graphics rendering. The mainstream technical solutions in the industry are all based on cloud computing infrastructure to build mutually separated cloud service stacks. The specific implementation paths are as follows: In the data storage stage, independent object storage services (such as AWS S3, Alibaba Cloud OSS, etc.) are commonly used to upload massive amounts of raw data, such as 3D models, satellite imagery, and design drawings, to designated storage buckets for unified management in the form of files or objects. In the business logic computing stage, it is mostly implemented by function computing services (such as AWS Lambda) or container services (such as Kubernetes). When a computing task is triggered, the raw data needs to be downloaded from the object storage to the computing environment, processed, and then the results are sent back to the storage bucket. In the graphics rendering stage, cloud servers or virtual desktops with powerful GPU computing power need to be deployed. The result data generated in the computing stage needs to be transmitted to the rendering node again, which completes the final image generation and is finally pushed to the user terminal in the form of images or video streams. However, the aforementioned storage, computation, and rendering separation architecture suffers from excessively high end-to-end latency due to its lengthy data processing links. For example, a complete task requires a complex data flow path of "upload → storage → download → computation → upload → download → rendering → push," with massive amounts of data frequently moved between different service nodes, generating numerous I / O operations. For real-time interactive applications, this multi-layered data transmission causes significant latency, failing to meet the performance requirements of low-latency scenarios such as industrial control and real-time gaming. Furthermore, the multiple round trips of gigabytes or even terabytes of data across the network not only cause a surge in network bandwidth consumption, significantly increasing enterprise cloud service bandwidth expenditure, but also make cross-service data transmission a system performance bottleneck, easily leading to data processing lag and response timeouts. In addition, developers need to integrate and manage three different types of cloud services—storage, computation, and rendering—simultaneously, coordinating authentication mechanisms, network connectivity, and data consistency between services. This results in extremely complex system architecture, a heavy integration and maintenance burden, and further increases long-term operating costs for enterprises.

[0003] In summary, the existing discrete architecture is no longer suitable for the core requirements of digital scenarios for efficient, low-cost, and highly flexible data processing. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a microkernel-based in-store computing and drawing integrated system and method, and storage medium, so as to solve the technical problems mentioned in the prior art.

[0005] A microkernel-based in-store computation and drawing method, the method comprising the following steps: S1. Obtain the task request data packet uploaded by the user terminal and parse it to obtain the task type; S2. Match the corresponding task workflow from the configuration file based on the task type; S3. Based on the currently matched task workflow, multiple microkernels are collaboratively scheduled to execute the corresponding tasks. The task execution method specifically includes: S310, the micro storage kernel retrieves the metadata required during the execution of the current task according to the task workflow; S320. The microcomputing kernel performs logical operations on the metadata output by the microstorage kernel according to the task workflow to generate a dataset to be rendered. S330: The micro-rendering kernel adaptively executes the corresponding rendering mode on the dataset to be rendered output by the micro-computing kernel according to the task workflow, obtains the rendering result, and outputs it to the user terminal.

[0006] Optionally, in step S1, the task request data packet includes at least a task type identifier and input parameters.

[0007] Optionally, in step S2, the task workflow is configured with a corresponding microkernel call sequence, execution logic, and data dependencies between microkernels based on any task type and saved in a configuration file.

[0008] Optionally, in step S310, the micro-storage kernel dynamically responds to the received request information based on a built-in data access method; this data access method specifically includes: ① Data storage mode: The input raw data is divided into several data blocks according to a preset size; Configure a corresponding cryptographic hash value for each of the data blocks to obtain the block hash; Construct a Merkle tree using all the block hashes as leaf nodes, and calculate the root hash of the Merkle tree; The root hash and the file identification information of the original data are written into the blockchain network to achieve on-chain evidence storage. The data block is stored in IPFS to achieve off-chain storage, and a unique DataID is generated, which is mapped to the metadata index of the original data. ② Data retrieval mode: Based on the currently received data retrieval request instruction containing DataID, all data blocks are located through the DataID index, and the corresponding metadata is obtained by retrieving and reassembling them in parallel from the IPFS.

[0009] Optionally, the root hash of the Merkle tree is recalculated based on the currently received data retrieval request instruction containing the DataID, and compared and verified with the root hash stored on the blockchain network; when the two are verified to be consistent, the corresponding data block is transmitted through the internal bus.

[0010] Optionally, in S320, the method by which the microcomputing kernel performs logical operations on the metadata output by the microstorage kernel according to the task workflow includes: At least one of finite element analysis, image recognition, and data conversion.

[0011] Optionally, in step S330, the micro-rendering kernel performs rendering processing on the dataset to be rendered output by the micro-computing kernel based on a corresponding rendering mode matched by preset mode selection conditions. The preset mode selection conditions include at least one or more of network latency, model triangle count, and terminal type; the rendering mode includes at least: Cloud rendering streaming mode: For a dataset to be rendered where the number of triangles in the model is higher than a preset threshold, the micro-rendering kernel sends the dataset to be rendered output by the micro-computing kernel back to the cloud for rendering, obtains a video stream, and pushes it to the user terminal, where the user terminal decodes and displays the video stream. Standardized rendering mode: For a dataset to be rendered where the network latency is lower than a preset duration and / or the number of triangles in the model is lower than a preset threshold, the micro-rendering kernel generates a rendering instruction set decoupled from the specified graphics API from the dataset to be rendered output by the micro-computing kernel and sends it to the user terminal. The user terminal receives the rendering instruction set based on the instruction parser and completes local rendering.

[0012] A microkernel-based in-store, computational, and drawing system, wherein the system is applied in the above-described method, and the system comprises: The task orchestrator is used to obtain the task request data packets uploaded by the user terminal and parse them to obtain the task type. Based on the task type, it matches the corresponding task workflow from the configuration file and coordinates multiple microkernels to execute the corresponding tasks according to the currently matched task workflow. An internal communication bus module is used to establish communication links between the task orchestrator and multiple microkernels; the multiple microkernels include: The micro-storage kernel module is used to retrieve the metadata required during the execution of the current task; The microcomputing kernel module is used to perform logical operations on the metadata output by the microstorage kernel to generate a dataset to be rendered; The micro-rendering kernel module is used to adaptively execute the corresponding rendering mode on the dataset to be rendered output by the micro-computing kernel, obtain the rendering result, and output it to the user terminal.

[0013] Optionally, the user terminal is at least one of a web browser, a mobile app, and a desktop client.

[0014] A computer-readable storage medium storing a control program, which, when executed by a processor, enables the implementation of the steps included in the microkernel-based in-memory computing and drawing method described above.

[0015] The beneficial effects that this invention can produce include: 1. This invention provides a microkernel-based integrated storage-computation-drawing system and method, as well as a storage medium. The task orchestrator matches task workflows based on parsed task types and collaboratively schedules micro-storage kernels, micro-computation kernels, and micro-rendering kernels to execute related tasks according to predefined task workflows. It rapidly acquires metadata input directly into the micro-computation kernel via a built-in parallel retrieval method within the micro-storage kernel, eliminating the need for additional downloads. Simultaneously, the micro-rendering kernel adaptively selects rendering modes to render the dataset obtained from the micro-computation kernel's computation, reducing transmission steps. Furthermore, it utilizes an internal communication bus to replace external network transmission and Merkle tree verification to avoid repeated processing of data errors. Through the coordinated control of various technical features, redundant links in the repeated data transmission across nodes are eliminated, constructing an integrated path encompassing task scheduling, storage-computation-drawing collaboration, efficient transmission, and data verification. Ultimately, this significantly reduces end-to-end latency, meeting the needs of low-latency scenarios such as industrial control and real-time gaming, and avoiding issues such as lag and response timeouts.

[0016] 2. This invention reduces storage costs by employing IPFS off-chain storage and data partitioning technology. It also reduces bandwidth consumption by combining micro-rendering kernels with adaptive rendering modes such as cloud rendering streaming to reduce data volume and local rendering to avoid cloud bandwidth consumption. Simultaneously, the task orchestrator uniformly schedules multiple micro-kernels to avoid the redundant costs of independent service deployments, and the computer-readable storage medium is fixed into an integrated method to avoid repetitive development costs. Thus, through the synergistic effect of these technical features, a cost optimization path is constructed that achieves low-cost storage, efficient bandwidth utilization, simplified deployment, and reusable scalability, ultimately reducing enterprise cloud service bandwidth, storage, operation and maintenance, and development costs, thereby improving enterprise economic efficiency.

[0017] 3. This invention pre-defines task workflows through a task orchestrator, clearly defining microkernel call sequences and data dependencies, thus overcoming the drawbacks of manually configuring collaborative logic between multiple cloud stacks. The micro-storage, micro-computing, and micro-rendering kernels are integrated, with built-in core functions such as micro-storage kernel access methods and micro-computing kernel various computation methods, eliminating the need for external service integration. An internal communication bus constructs a unified communication link, simplifying the linkage logic between modules. The system is compatible with multiple terminals such as web browsers, mobile apps, and desktop clients, without requiring independent deployment of adaptation services. Therefore, based on the synergistic effects of the above-mentioned technical features, a unified scheduling, built-in functions, simplified linkage, and multi-terminal adaptation architecture are constructed to optimize data transmission paths, reducing system integration complexity and operational burden, and improving system adaptability and usability. Attached Figure Description

[0018] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system architecture block diagram of the present invention; Figure 3 This is a flowchart illustrating the micro-storage kernel execution request information in this invention; Figure 4 This is a flowchart of the adaptive rendering process performed by the micro-rendering kernel in this invention; Figure 5 This is a flowchart illustrating the application of the method and system of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 As shown, this invention provides a microkernel-based in-store computation and drawing integrated method, which includes the following steps: Step S1: Obtain the task request data packet uploaded by the user terminal. The task orchestrator is responsible for receiving and parsing the data packet to obtain the task type. The task request data packet contains at least the task type identifier (such as stress analysis and rendering) and input parameters (such as model DataID, analysis parameters, etc.), which can provide an accurate initial basis for subsequent task workflow matching and microkernel collaborative scheduling, avoiding resource waste and process redundancy caused by indiscriminate scheduling.

[0021] Step S2: Match the corresponding task workflow from the configuration file based on the task type; the task workflow is configured with the corresponding microkernel call sequence, execution logic, and data dependencies between microkernels for any task type and saved in the configuration file. For example, in the "Stress Analysis and Rendering" task type, the task workflow is configured as follows: The process involves calling the micro-storage kernel to acquire data, calling the micro-computing kernel to perform analysis, and calling the micro-rendering kernel for visualization. This overcomes the drawbacks of independent operation and lack of unified scheduling in the storage, computation, and rendering separation architecture, enabling orderly collaboration among core components such as storage, computation, and rendering. It clarifies the data flow path, avoiding disordered data transfer between different service nodes, simplifying the data processing chain, shortening end-to-end latency, and meeting the low-latency requirements of real-time interactive scenarios. Furthermore, by predefining specific parameters of the task workflow through configuration files, it reduces the need for manual configuration of the collaborative logic of each component, lowering system integration complexity and operational burden.

[0022] Step S3: Based on the currently matched task workflow, coordinate and schedule multiple microkernels to execute the corresponding tasks. The task execution method specifically includes: Step S310: The micro-storage kernel retrieves the metadata required during the execution of the current task according to the task workflow; Step S320: The microcomputing kernel performs logical operations on the metadata output by the microstorage kernel according to the task workflow to generate a dataset to be rendered. Step S330: The micro-rendering kernel adaptively executes the corresponding rendering mode on the dataset to be rendered output by the micro-computing kernel according to the task workflow, obtains the rendering result, and outputs it to the user terminal.

[0023] In the above, by integrating multiple microkernels into a unified process, the barriers to independent operation of each link can be eliminated, the data flow path can be shortened, and the number of data transmissions across nodes can be reduced, thus reducing I / O transmission operations. Compared with the previous independent operation of multiple nodes, it significantly reduces end-to-end latency, meeting the needs of low-latency scenarios such as industrial control and real-time gaming. At the same time, it reduces the bandwidth consumption caused by multiple round trips of massive data transmissions, reduces enterprise cloud service bandwidth expenditures, and avoids lag and response timeout issues caused by data transmission.

[0024] Further, in step S310, as Figure 3 As shown, the micro-storage kernel dynamically responds to received request information based on its built-in data access methods; these data access methods specifically include: ① Data Storage Mode: The input raw data is divided into several data blocks according to a preset size, such as 256MB. Raw data smaller than 256MB is stored in a separate data block to avoid data loss. This not only facilitates parallel transmission but also facilitates deduplication. If multiple files contain the same parts (such as a standard parts library), the system only needs to store one copy. Each data block is configured with a corresponding cryptographic hash value, such as SHA-256, to obtain a block hash; this hash is the unique digital fingerprint of the data block. All block hashes are used as leaf nodes to construct a Merkle tree, and the root hash of the Merkle tree is calculated. The root hash and the file identification information of the raw data are written to the blockchain network to achieve on-chain notarization. The file identification information of the raw data includes the file name, size, timestamp, and creator identity. The data blocks are stored in IPFS (Distributed Storage Module) to achieve off-chain storage, and a unique DataID is generated. This DataID maps to the metadata index of the raw data. ② Data retrieval mode: Based on the currently received data retrieval request instruction containing DataID, all data blocks are located through the DataID index, and the corresponding metadata is obtained by retrieving and reassembling them in parallel from IPFS.

[0025] Furthermore, based on the currently received data retrieval request instruction containing the DataID, the root hash of the Merkle tree is recalculated and compared with the root hash stored on the blockchain network. When the two match, it proves that the data is 100% complete and has not been tampered with, and the corresponding data block is transmitted through the internal bus. By replacing the previous single storage / retrieval mode of independent object storage services with the above data access method, it is possible to achieve integrated efficient data access and secure evidence preservation. At the same time, parallel retrieval improves data acquisition efficiency and further reduces data latency. Moreover, the use of Merkle tree hash verification and blockchain evidence preservation can ensure data integrity, avoid repeated processing caused by errors in data transmission or storage, and reduce bandwidth and computing power waste. Furthermore, using IPFS to implement off-chain storage can improve the elasticity and security of massive data storage and reduce enterprise storage costs.

[0026] Furthermore, in step S320, the method for the microcomputing kernel to perform logical operations on the metadata output by the microstorage kernel according to the task workflow includes at least one of finite element analysis, image recognition, and data conversion, in order to adapt to the task requirements of multiple fields such as industrial digital twins and smart cities, without the need to deploy multiple additional computing services, thereby achieving integrated computing capabilities.

[0027] Further, in step S330, as Figure 4As shown, the micro-rendering kernel performs rendering processing on the dataset to be rendered output by the micro-computing kernel based on the corresponding rendering mode matched by preset mode selection conditions. The preset mode selection conditions include at least one or more of network latency, model triangle count, and terminal type; the rendering modes include at least: Cloud rendering streaming mode: For a dataset to be rendered where the number of triangles in the model exceeds a preset threshold, the micro-rendering kernel sends the dataset to be rendered output by the micro-computing kernel back to the cloud for rendering, obtains a video stream, and pushes it to the user terminal, where the user terminal decodes and displays the video stream. Standardized rendering mode: For datasets to be rendered where the network latency is lower than a preset duration and / or the number of model triangles is lower than a preset threshold, the micro-rendering kernel generates a rendering instruction set decoupled from the specified graphics API from the dataset output by the micro-computation kernel and sends it to the user terminal. The user terminal receives the rendering instruction set based on the instruction parser and completes local rendering.

[0028] The above methods can dynamically match the rendering mode according to the needs of the scene, avoiding resource waste or excessive latency caused by a single rendering mode, while adapting the rendering capabilities to the terminal and network conditions. For example, cloud rendering is used for high-complexity models to avoid lag caused by insufficient terminal computing power, while streaming transmission reduces data transmission volume and bandwidth consumption. Local rendering is used for low-complexity models or scenes with good network conditions to further reduce latency and improve the user interaction experience. As a result, there is no need to deploy multiple additional rendering services, simplifying the system architecture, reducing operation and maintenance costs, and improving system adaptability.

[0029] like Figure 2As shown, this invention also provides a microkernel-based integrated storage, computation, and rendering system. This system is applied to the aforementioned method and includes a task orchestrator and multiple microkernels, as well as an internal communication bus module for constructing communication links between the task orchestrator and the multiple microkernels. The internal communication bus module can employ high-performance RPC (Remote Procedure Call) or message queues such as gRPC or Apache Kafka. The task orchestrator acquires task request data packets uploaded by the user terminal and parses them to obtain the task type. Based on the task type, it matches the corresponding task workflow from a configuration file and collaboratively schedules multiple microkernels to execute the corresponding tasks according to the currently matched task workflow. The multiple microkernels include a micro-storage kernel module, a micro-computation kernel module, and a micro-rendering kernel module. The micro-storage kernel module retrieves the metadata required during the current task execution process. The micro-computation kernel module performs logical operations on the metadata output by the micro-storage kernel based on built-in Serverless functions to generate a dataset to be rendered. The micro-rendering kernel module adaptively executes the corresponding rendering mode on the dataset to be rendered output by the micro-computation kernel, obtains the rendering result, and outputs it to the user terminal. The user terminal is at least any one of a web browser, a mobile app, or a desktop client. By constructing an integrated system architecture encompassing task orchestration, microkernel collaboration, and internal communication, the previously fragmented cloud service stack is replaced, enabling close collaboration among modules and adapting to task request data packets input from various user terminals. The internal communication bus replaces external network transmission, significantly reducing data transmission latency and bandwidth consumption. The task orchestrator achieves unified scheduling of each microkernel, preventing disordered operation of various stages, simplifying the system architecture, reducing the integration and maintenance burden on developers, and lowering long-term operating costs for enterprises. Adaptation to multiple user terminals enhances system usability, while eliminating the need to deploy independent services for different terminals, further reducing maintenance costs.

[0030] As a computer-readable storage medium of the present invention, the computer-readable storage medium stores a control program. When the control program is executed by a processor, it can realize the steps included in the above-described microkernel-based storage-computation-drawing integrated method, which facilitates system deployment and promotion.

[0031] Example Take, for example, the emergency response and decision-making following a major natural disaster such as a landslide. In traditional emergency control plans, suppose a massive landslide occurs in a mountainous area due to heavy rainfall, blocking a river channel, forming a barrier lake, and seriously threatening the safety of downstream residents. In this situation, the emergency command center needs to immediately execute the following tasks: 1) Obtain the latest high-resolution satellite remote sensing images and lidar scanning data of the disaster area; 2) Quickly calculate the earthwork volume of the landslide, the water storage capacity of the barrier lake, and the threatened population and infrastructure; 3) Visually render a 3D model of the disaster body, and perform annotations and simulation analyses such as dam failure analysis on the model to support expert group consultations and rescue decisions.

[0032] The existing architecture that separates storage, computation, and rendering generally faces the following challenges when performing the above tasks: Data bloat: The latest remote sensing data is often in the terabyte range, and it takes several hours to download it to the local command center.

[0033] Computational bottleneck: Earthwork volume calculation is an extremely CPU / GPU-intensive task, which the command center's computer may not be able to complete quickly.

[0034] Rendering difficulties: High-precision 3D models of disaster areas cannot be loaded and rendered smoothly on ordinary office computers or laptops, let alone interactively analyzed.

[0035] Inefficient collaboration: Experts scattered across different locations need to wait for the center to process the data before distributing the huge model files, resulting in extremely low efficiency in collaborative consultations.

[0036] Reference Figure 5As shown, this invention possesses highly efficient collaboration and powerful computing rendering capabilities when performing the aforementioned tasks, enabling data storage and preparation to be completed within seconds. Specifically, commanders can upload terabyte-level satellite imagery and laser point cloud data of the disaster area to this system. At this time, the micro-storage kernel starts, divides the data into blocks, calculates the "data fingerprint" (i.e., the root hash of the Merkle tree), and instantly writes the hash into the blockchain, completing data storage and ensuring that all subsequent decisions are based on the tamper-proof original data. The data blocks themselves are stored in a high-performance distributed storage module. After completing the storage task, the system returns a lightweight DataID, which represents the complete disaster area data that has been "trustedly sealed." Then, a one-click intelligent analysis command is initiated, enabling efficient cloud-based computation. The specific process is as follows: The commander selects the DataID on the web interface and clicks the "Intelligent Analysis" button, sending the task request data packet to the task orchestrator. The task orchestrator schedules the microcomputing kernel to execute the corresponding logical operation task. At this time, the Serverless function in the microcomputing kernel is triggered. It has built-in AI recognition models and geometric algorithms. The Serverless function directly pulls the required data blocks from the micro storage kernel through the internal bus (without downloading the entire data), quickly identifying the landslide boundary and water body boundary, and accurately calculating the landslide earthwork volume and the water storage volume of the barrier lake. Then, the calculation results, such as the dangerous area polygon, are instantly transmitted to the micro rendering kernel. The micro rendering kernel integrates the high-precision terrain model, remote sensing imagery, and the calculated dangerous area, performing fusion and 3D rendering in the cloud. Considering the extreme complexity of the model and the high demands for real-time performance, it automatically adopts a cloud rendering streaming mode to push a low-latency, high-definition 3D scene video stream in real time to the large screen of the emergency command center, as well as to the laptops or tablets of expert members scattered across the country. This allows any expert to click on the video stream, query the coordinates and elevation of a point, or add annotations. To further improve the real-time interactive performance of the data, any identifiers added to the video stream, such as annotation features with different color symbols, are sent back to the cloud-based microcomputing kernel for processing. Simultaneously, the processed data stream is input into the micro-rendering kernel to dynamically update the rendering results. The updated rendering results are then synchronized to the video streams of all user terminals via the cloud, achieving collaborative decision-making as if everyone were in the same room, even when separated by thousands of miles.

Claims

1. A microkernel-based in-memory computing and drawing integrated method, characterized in that, The method includes the following steps: S1. Obtain the task request data packet uploaded by the user terminal and parse it to obtain the task type; S2. Match the corresponding task workflow from the configuration file based on the task type; S3. Based on the currently matched task workflow, multiple microkernels are collaboratively scheduled to execute the corresponding tasks. The task execution method specifically includes: S310, the micro storage kernel retrieves the metadata required during the execution of the current task according to the task workflow; S320. The microcomputing kernel performs logical operations on the metadata output by the microstorage kernel according to the task workflow to generate a dataset to be rendered. S330: The micro-rendering kernel adaptively executes the corresponding rendering mode on the dataset to be rendered output by the micro-computing kernel according to the task workflow, obtains the rendering result, and outputs it to the user terminal.

2. The microkernel-based in-memory computing and drawing method according to claim 1, characterized in that, In step S1, the task request data packet includes at least a task type identifier and input parameters.

3. The microkernel-based in-memory computing and drawing method according to claim 1, characterized in that, In S2, the task workflow is configured with a corresponding microkernel call sequence, execution logic, and data dependencies between microkernels based on any task type and saved in a configuration file.

4. The microkernel-based in-memory computing and drawing method according to claim 1, characterized in that, In step S310, the micro-storage kernel dynamically responds to the received request information based on a built-in data access method; this data access method specifically includes: ① Data storage mode: The input raw data is divided into several data blocks according to a preset size; Configure a corresponding cryptographic hash value for each of the data blocks to obtain the block hash; Construct a Merkle tree using all the block hashes as leaf nodes, and calculate the root hash of the Merkle tree; The root hash and the file identification information of the original data are written into the blockchain network to achieve on-chain evidence storage. The data block is stored in IPFS to achieve off-chain storage, and a unique DataID is generated, which is mapped to the metadata index of the original data. ② Data retrieval mode: Based on the currently received data retrieval request instruction containing DataID, all data blocks are located through the DataID index, and the corresponding metadata is obtained by retrieving and reassembling them in parallel from the IPFS.

5. The microkernel-based in-memory computing and drawing method according to claim 4, characterized in that, Based on the currently received data retrieval request instruction containing DataID, the root hash of the Merkle tree is recalculated and compared with the root hash stored on the blockchain network; when the two are verified to be consistent, the corresponding data block is transmitted through the internal bus.

6. The microkernel-based in-memory computing and drawing method according to claim 1, characterized in that, In S320, the method by which the microcomputing kernel performs logical operations on the metadata output by the microstorage kernel according to the task workflow includes: At least one of finite element analysis, image recognition, and data conversion.

7. The microkernel-based in-memory computing and drawing integrated method according to claim 1, characterized in that, In S330, the micro-rendering kernel performs rendering processing on the dataset to be rendered output by the micro-computing kernel by matching the corresponding rendering mode based on preset mode selection conditions. The preset mode selection conditions include at least one or more of network latency, number of model triangles, and terminal type. The rendering modes include at least: Cloud rendering streaming mode: For a dataset to be rendered where the number of triangles in the model is higher than a preset threshold, the micro-rendering kernel sends the dataset to be rendered output by the micro-computing kernel back to the cloud for rendering, obtains a video stream, and pushes it to the user terminal, where the user terminal decodes and displays the video stream. Standardized rendering mode: For a dataset to be rendered where the network latency is lower than a preset duration and / or the number of triangles in the model is lower than a preset threshold, the micro-rendering kernel generates a rendering instruction set decoupled from the specified graphics API from the dataset to be rendered output by the micro-computing kernel and sends it to the user terminal. The user terminal receives the rendering instruction set based on the instruction parser and completes local rendering.

8. A microkernel-based in-memory computing and drawing integrated system, characterized in that, The system is applied in the method according to any one of claims 1-7, the system comprising: The task orchestrator is used to obtain the task request data packets uploaded by the user terminal and parse them to obtain the task type. Based on the task type, it matches the corresponding task workflow from the configuration file and coordinates multiple microkernels to execute the corresponding tasks according to the currently matched task workflow. An internal communication bus module is used to establish communication links between the task orchestrator and multiple microkernels; the multiple microkernels include: The micro-storage kernel module is used to retrieve the metadata required during the execution of the current task; The microcomputing kernel module is used to perform logical operations on the metadata output by the microstorage kernel to generate a dataset to be rendered; The micro-rendering kernel module is used to adaptively execute the corresponding rendering mode on the dataset to be rendered output by the micro-computing kernel, obtain the rendering result, and output it to the user terminal.

9. A microkernel-based in-memory computing and drawing system according to claim 8, characterized in that, The user terminal is at least one of a web browser, a mobile app, and a desktop client.

10. A computer-readable storage medium storing a control program, characterized in that, When the control program is executed by the processor, it can implement the steps included in the microkernel-based in-memory computing and drawing method as described in any one of claims 1-7.

Images