Reusable server blade

Abstract

The invention provides a reusable server blade, belongs to the technical field of server hardware, and is applied to an array server. In order to solve the problems that a traditional server blade is high in upgrading cost, only supports a single framework and is complex in operation and maintenance, the blade comprises a basic carrier plate, a core calculation module (pluggable and supports X86 and ARM heterogeneous frameworks), a BMC management unit, a universal power supply system and the like, wherein the core calculation module, the BMC management unit, the universal power supply system and the like are integrated or connected through a standardized interface. The BMC serves as a control center, can automatically detect plugging and unplugging of a core module, read module identification, retrieve matched firmware and complete initialization, and also supports hot plugging, temperature over-limit regulation and control and redundant network communication. The platform can reuse the basic carrier plate to reduce hardware cost, flexibly adapt to multi-architecture and business requirements, reduce operation and maintenance operations and errors, guarantee reliable operation, and improve resource utilization rate and platform life cycle.

Description

Technical Field

[0001] This invention relates to the field of server hardware technology, specifically to a reusable server blade. Background Technology

[0002] Current traditional server blade technology has significant drawbacks, making it difficult to adapt to diverse computing needs: 1. High hardware iteration costs: The core components of traditional blades (such as CPU, SoC, GPU) are tightly integrated with the carrier board. If it is necessary to upgrade the core components (such as replacing the new generation CPU) or switch the computing architecture (such as switching from x86 to ARM), it is often necessary to replace the entire blade module or even the whole machine. This not only increases the cost of hardware procurement, but also leads to the disposal of a large number of blades with acceptable performance, generating electronic waste, which is not in line with the trend of green computing. 2. Poor flexibility in module upgrades: Even within the same architecture (such as different generations of x86 CPUs), the tightly integrated design makes it difficult to replace individual core modules, and in most cases, the entire module still needs to be replaced. 3. Severe isolation of architectural resources: Different architectures such as x86 and ARM have their own advantages (x86 is suitable for high-performance computing, and ARM is suitable for edge low-power scenarios), but traditional blade servers only support a single architecture. If users need to meet the needs of multiple architectures at the same time, they need to purchase and deploy multiple server systems separately, which not only occupies more data center space and increases management costs, but also cannot flexibly allocate computing resources of different architectures, resulting in low resource utilization.

[0003] To address this, a reusable server blade is proposed. Summary of the Invention

[0004] The present invention aims to solve the problems mentioned in the background art by providing a reusable server blade.

[0005] The specific technical solution is as follows: A reusable server blade includes a base carrier board, and a core computing module, a BMC management unit, a general power supply system, a standardized module interface, a general high-speed interconnect bus, an FPGA unit, a backplane MCU, a temperature sensor, a power button and power indicator, a module indicator, a UID button and UID indicator integrated on the base carrier board. The core computing module is detachably connected to the base carrier board through the standardized module interface. The universal power supply system is electrically connected to the standardized module interface to provide power to the core computing module. The first end of the universal high-speed interconnect bus is electrically connected to the standardized module interface, and the second end extends to the backplane connection end of the base carrier board to realize the connection of external I / O devices. The BMC management unit is electrically connected to the standardized module interface, the general power supply system, the general high-speed interconnect bus, the FPGA unit, the backplane MCU, the temperature sensor, the power button and power indicator, the module indicator, and the UID button and UID indicator. The BMC management unit communicates with the backplane MCU via a UART bus, the backplane MCU communicates with the FPGA unit via a UART, and the FPGA unit communicates with the core computing module via I2C. The BMC management unit is configured to: respond to the plugging / unplugging event of the core computing module through the standardized module interface, and, based on the identification information of the core computing module, call and load the corresponding firmware to control the core computing module to complete initialization and run normally.

[0006] In the aforementioned reusable server blade, the core computing module is at least one heterogeneous computing module, the architecture of which includes an x86 architecture or an ARM architecture, and the core computing module integrates a multi-core SoC.

[0007] In the aforementioned reusable server blade, the standardized module interface is a high-density connector. The pin configuration of the high-density connector meets the electrical connection requirements between the core computing module and the general power supply system, general high-speed interconnect bus, FPGA unit, and BMC management unit. Furthermore, the physical size of the high-density connector is adapted to the preset mounting position on the base carrier to achieve stable insertion and removal of the core computing module.

[0008] The aforementioned reusable server blade includes a general power supply system comprising a 12V main power module and multiple sub-power units. The 12V main power module provides input power to the multiple sub-power units. The multiple sub-power units supply power to the core computing module, the BMC management unit, and the FPGA unit, respectively, and the output voltage of each sub-power unit can be dynamically adjusted according to the corresponding load requirements.

[0009] In the aforementioned reusable server blade, the BMC management unit is also communicatively connected to a firmware library, which includes a pre-built firmware library and a remote firmware library. The BMC management unit is configured to retrieve the matching firmware from the pre-built firmware library or from the remote firmware library via the universal high-speed interconnect bus based on the identification information of the core computing module.

[0010] In the aforementioned reusable server blade, the universal high-speed interconnect bus includes a PCIe interface and an Ethernet interface; the PCIe interface is used for high-speed data interaction between the core computing module and external storage devices and accelerator cards; the Ethernet interface is connected to a switch unit on the base carrier board, the switch unit including switch A and switch B, and switch A and switch B communicate with the BMC management unit through the Ethernet bus to achieve redundant network connections.

[0011] In the aforementioned reusable server blade, there are two FPGA units. The FPGA units communicate bidirectionally with the core computing module via an I2C bus to achieve data forwarding, and communicate bidirectionally with the BMC management unit via a UART through a backplane MCU to enable the BMC management unit to read and write configuration parameters of the core computing module.

[0012] In the aforementioned reusable server blade, at least two temperature sensors are respectively arranged on the surface of the core computing module and the surface of the general power supply system. The temperature sensors collect temperature data in real time and transmit it to the BMC management unit. The BMC management unit is also configured to control the general power supply system to reduce the output power when the collected temperature data exceeds a preset threshold, and to provide an alarm prompt through the module indicator light.

[0013] The aforementioned reusable server blade, wherein the base carrier board also integrates a general auxiliary component interface, which includes a memory slot, an M.2 storage interface, a U.2 storage interface, and an expansion card slot; the memory slot is used to insert a DIMM memory module, the M.2 storage interface and the U.2 storage interface are used to connect external storage devices, and the expansion card slot is used to insert a discrete GPU or network card.

[0014] In the aforementioned reusable server blade, the core computing module supports plugging and unplugging operations under normal main power supply conditions of the server blade; the BMC management unit is further configured to: when a plugging and unplugging operation of the core computing module is detected, first control the general power supply system to cut off the power supply to the core computing module, and after completing firmware loading and configuration, control the general power supply system to restore the power supply to the core computing module to complete initialization.

[0015] The present invention also provides a module adaptive management method for reusable server blades, executed by the BMC management unit, the method comprising: Detect the presence status of the core computing module; When the insertion or replacement of the core computing module is detected, the module type identifier of the core computing module is read. Based on the module type identifier, retrieve the matching firmware; Verify the validity of the firmware; If the verification passes, the firmware is loaded into the core computing module; The core computing module is controlled to complete initialization based on the loaded firmware to operate normally; If the verification fails, the power button and power indicator light will be triggered to display an error message.

[0016] The present invention has the following beneficial effects: 1. Significantly reduce hardware costs and resource waste By adopting the design of "reusing the base carrier board + replacing the core module separately", users do not need to replace the entire blade or the whole machine when upgrading core components (such as CPU, SoC) or switching architectures. The base carrier board (including common components such as power supply, management, and interconnect bus) can be reused for a long time. Only the core computing module needs to be replaced, which directly reduces the cost of hardware iteration. At the same time, it reduces electronic waste generated by replacing the whole machine, which meets the requirements of green computing.

[0017] 2. Enhance the flexibility of computing architecture in adapting to business needs. It supports heterogeneous core computing modules such as x86 and ARM, and automatically adapts firmware through BMC. A single basic carrier board can be compatible with modules of different architectures. Users do not need to deploy servers separately for different architectures, saving data center space and procurement costs. Combined with hot-swappable design, core module replacement can be done without downtime, and blades can quickly respond to changes in business needs (such as using x86 modules to carry high-performance computing during the day and ARM modules to carry low-power edge processing at night), improving server resource utilization.

[0018] 3. Enhance operational reliability and system robustness The Ethernet interface of the general high-speed interconnect bus connects switch A and switch B to form a redundant network, avoiding network interruption caused by the failure of a single switch and ensuring the continuity of data transmission. Temperature sensors monitor the temperature of core components in real time. When the temperature exceeds the threshold, the BMC actively reduces the power supply and issues an alarm to prevent hardware damage or system crash due to overheating. The firmware verification process avoids loading invalid firmware, reducing startup failures or instability caused by firmware incompatibility. The sub-power supply units of the universal power supply system can dynamically adjust the output voltage and supply power to different components as needed, which avoids abnormal operation caused by insufficient power supply and prevents energy waste due to excessive power supply, thus balancing stability and energy efficiency.

[0019] 4. Reduce operational complexity and the risk of human error. The BMC management unit automates the entire process of core module plug-in / plug-out detection, firmware adaptation, and initialization, eliminating the need for manual configuration (such as jumpers or firmware flashing). This reduces the workload of maintenance personnel and avoids faults caused by human error (such as incorrect firmware flashing or incorrect configuration parameters). At the same time, error prompts (indicator lights + alarm information) can quickly locate problems (such as firmware incompatibility or poor module contact), shortening fault handling time.

[0020] 5. Extend the platform lifecycle and improve scalability The firmware library includes both pre-built and remote types. Even if a new model of core computing module appears in the future, the adapted firmware can be obtained through the remote firmware library without replacing the base carrier board, thus extending the life cycle of the entire blade platform and protecting the user's hardware investment. The integrated memory slots, M.2 / U.2 storage interfaces, and expansion card slots on the base carrier board support flexible expansion of memory, storage, and computing capabilities, enabling the blades to adapt to diverse business scenarios, from general data processing to AI inference and high-performance computing, without the need to purchase dedicated blades for specific scenarios. Attached Figure Description

[0021] Figure 1 The system architecture and component connection diagram provided for embodiments of the present invention; Figure 2 The BMC adaptive management flowchart is provided for embodiments of the present invention. Detailed Implementation

[0022] The technical solution of the present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0023] The accompanying drawings are for illustrative purposes only and are schematic diagrams, not actual images. They should not be construed as limiting the scope of this application. To better illustrate the embodiments of the present invention, some parts in the drawings may be omitted, enlarged, or reduced, and do not represent the actual dimensions of the product. It is understandable to those skilled in the art that some well-known structures and their descriptions may be omitted in the drawings.

[0024] In the accompanying drawings of the embodiments of the present invention, the same or similar reference numerals correspond to the same or similar components. In the description of the present invention, it should be understood that if terms such as "upper," "lower," "left," "right," "inner," and "outer" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, the terms used to describe positional relationships in the drawings are only for illustrative purposes and should not be construed as limiting the present application. For those skilled in the art, the specific meaning of the above terms can be understood according to the specific circumstances.

[0025] In the description of this invention, unless otherwise explicitly specified and limited, the term "connection" or similar designation indicating a connection between components should be interpreted broadly. For example, it can refer to a fixed connection, a detachable connection, or an integral part; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can refer to the internal communication between two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0026] Example 1

[0027] The reusable server blade provided in this embodiment, such as Figure 1 As shown, it includes a base carrier board, and core computing modules, BMC management unit, general power supply system, standardized module interface, general high-speed interconnect bus, FPGA unit, backplane MCU, temperature sensor, power button and power indicator, module indicator, UID button and UID indicator integrated on the base carrier board. The core computing module is detachably connected to the base carrier board via a standardized module interface. The general power supply system is electrically connected to the standardized module interface to provide power to the core computing module. The first end of the general high-speed interconnect bus is electrically connected to the standardized module interface, and the second end extends to the backplane connection end of the base carrier board to realize the connection of external I / O devices. The BMC management unit is electrically connected to the standardized module interface, the general power supply system, the general high-speed interconnect bus, the FPGA unit, the backplane MCU, the temperature sensor, the power button and power indicator, the module indicator, and the UID button and UID indicator. The BMC management unit communicates with the backplane MCU via the UART bus, the backplane MCU communicates with the FPGA unit via UART, and the FPGA unit communicates with the core computing module via I2C. The BMC management unit is configured to: respond to the plugging / unplugging events of the core computing module through the standardized module interface, and, based on the identification information of the core computing module, call and load the corresponding firmware to control the core computing module to complete initialization and normal operation.

[0028] By designing the core computing module as pluggable and decoupling it from common systems on the base carrier board (such as power supply, management, and interconnect buses), a single base carrier board model can be adapted to core computing modules with various architectures. This fundamentally changes the traditional situation where the core computing unit of a server blade is fixed and difficult to upgrade. The BMC management unit, as the control center, can automatically respond to module plugging and unplugging events and proactively complete firmware loading and initialization. This solution frees system administrators from tedious manual configuration work specific to hardware, significantly reducing the complexity of operation and maintenance and the risk of human error. The entire initialization process is uniformly scheduled and controlled by the BMC, ensuring that the core computing module always starts with the correct firmware and configuration, avoiding system startup failures or instability caused by firmware incompatibility or configuration errors, and improving the reliability of the single board and even the entire server system.

[0029] Specifically, in this embodiment, the core computing module is at least one heterogeneous computing module. The architecture of the heterogeneous computing module includes x86 architecture or ARM architecture, and the core computing module integrates a multi-core SoC. By explicitly supporting heterogeneous architectures such as x86 and ARM, this solution allows computing units with different instruction sets to be deployed in a mixed manner within the same server chassis. Users can flexibly select the most suitable core module according to the computing characteristics and cost requirements of specific applications, thereby achieving optimal computing power allocation and energy efficiency.

[0030] Specifically, in this embodiment, the standardized module interface is a high-density connector. The pin configuration of the high-density connector meets the electrical connection requirements between the core computing module and the general power supply system, general high-speed interconnect bus, FPGA unit, and BMC management unit. Furthermore, the physical dimensions of the high-density connector are adapted to the preset mounting positions on the base carrier board to achieve stable insertion and removal of the core computing module. The high-density connector solution provides a stable, high-quality physical channel for all possible electrical signals and power transmission. Its standardized physical dimensions and pin definitions ensure the accuracy of mechanical alignment and electrical connection during insertion and removal of different core computing modules, achieving seamless replacement and reliable connection between modules.

[0031] Specifically, in this embodiment, the general power supply system includes a 12V main power module and multiple sub-power units. The 12V main power module provides input power to the multiple sub-power units. The multiple sub-power units power the core computing module, the BMC management unit, and the FPGA unit, respectively, and the output voltage of each sub-power unit can be dynamically adjusted according to the corresponding load requirements. This discrete, dynamically adjustable sub-power unit scheme allows the power supply system to provide precise voltages to different components as needed. This not only meets the specific voltage requirements that may exist for core modules with different architectures but also avoids energy waste, improves the overall energy efficiency of the blade, and helps reduce heat generation.

[0032] Specifically, in this embodiment, the BMC management unit is also communicatively connected to a firmware library, which includes a pre-built firmware library and a remote firmware library. The BMC management unit is configured to retrieve the matching firmware from the pre-built firmware library or from the remote firmware library via a universal high-speed interconnect bus based on the identification information of the core computing module. This combination of local and remote firmware library solutions ensures plug-and-play functionality for basic model modules in the absence of a network, while also providing firmware support for potentially new core module models in the future. This gives the server blade platform long-term viability and versatility in embracing new technologies, effectively protecting the user's hardware investment.

[0033] Specifically, in this embodiment, the general-purpose high-speed interconnect bus includes a PCIe interface and an Ethernet interface. The PCIe interface is used for high-speed data interaction between the core computing module and external storage devices and accelerator cards. The Ethernet interface connects to a switch unit on the base carrier board. The switch unit includes switch A and switch B, which communicate with the BMC management unit via the Ethernet bus to achieve redundant network connections. This dual high-speed interface scheme of PCIe and Ethernet ensures that the core computing module obtains sufficient bandwidth in both data read / write and network communication, two key dimensions. The redundant design of the switch unit further eliminates single points of failure in the network, providing continuous, reliable, and high-performance data exchange capabilities for upper-layer applications.

[0034] Specifically, in this embodiment, there are two FPGA units. The FPGA units communicate bidirectionally with the core computing module via an I2C bus to forward data, and communicate bidirectionally with the BMC management unit via a UART through the backplane MCU to enable the BMC management unit to read and write configuration parameters of the core computing module. This two-FPGA-per-function approach separates system management configuration (via I2C) from high-speed data forwarding (via UART) at the hardware level. This division of labor avoids interference from management traffic to business data, improves data processing efficiency and real-time performance, and makes the monitoring and configuration of the computing module more flexible and precise.

[0035] Specifically, in this embodiment, at least two temperature sensors are used, respectively positioned on the surface of the core computing module and the surface of the general power supply system. The temperature sensors collect temperature data in real time and transmit it to the BMC management unit. The BMC management unit is also configured to control the general power supply system to reduce its output power and issue an alarm via module indicator lights when the collected temperature data exceeds a preset threshold. By deploying sensors at key heat sources and combining this with BMC monitoring and strategy control, this solution can perceive the system temperature status in real time. In the event of overheating risk, the system can proactively intervene by reducing frequency or power and issuing an alarm via indicator lights, thereby effectively preventing hardware damage or system crashes caused by overheating and improving system robustness.

[0036] Specifically, in this embodiment, the base carrier also integrates general-purpose auxiliary component interfaces, including memory slots, M.2 storage interfaces, U.2 storage interfaces, and expansion card slots. The memory slots are used to connect DIMM memory modules, the M.2 and U.2 storage interfaces are used to connect external storage devices, and the expansion card slots are used to connect discrete GPUs or network cards. Integrating multiple general-purpose auxiliary component interfaces on the base carrier enables this blade application to have rich peripheral device connectivity and capability expansion possibilities. Users can flexibly add memory, storage, or accelerator cards as needed, allowing this standardized platform to adapt to diverse business scenarios ranging from general computing to specific acceleration requirements.

[0037] Specifically, in this embodiment, the core computing module supports plugging and unplugging operations under normal server blade main power supply conditions. The BMC management unit is further configured to: upon detecting a plugging or unplugging operation of the core computing module, first control the general power supply system to cut off the power supply to the core computing module; after firmware loading and configuration are completed, control the general power supply system to restore the power supply to the core computing module to complete initialization. The hot-swappable design allows replacement of the core computing module without shutting down the server, reducing server downtime and improving server availability; the operation of disconnecting power before plugging or unplugging avoids module damage caused by hot-plugging, reducing the risk of hardware loss and improving the safety and reliability of core computing module replacement.

[0038] Example 2

[0039] This embodiment provides a module adaptive management method for the reusable server blade in Embodiment 1, executed by the BMC management unit, such as... Figure 2 As shown, the method includes: Detect the presence status of the core computing module; When the insertion or replacement of a core computing module is detected, the module type identifier of the core computing module is read. Retrieve matching firmware based on module type identifier; Verify the validity of the firmware; If the verification passes, the firmware is loaded into the core computing module; The core computing module of the control system completes initialization based on the loaded firmware to operate normally; If the verification fails, the power button and power indicator light will be activated to display an error message.

[0040] The entire process is fully automated, eliminating the need for manual configuration and reducing manual operation costs and the risk of configuration errors. The firmware verification step can avoid initialization failures caused by loading invalid firmware, improving the initialization success rate. Error messages can help maintenance personnel quickly locate problems such as firmware incompatibility, shortening troubleshooting time and improving the overall initialization efficiency and reliability after the core computing module is replaced.

[0041] In summary, the working principle of the reusable server blade provided in this embodiment is as follows: The working principle of reusable server blades revolves around "universal carrier board + pluggable core module + BMC automated management and control", and is divided into two parts: hardware architecture logic and control flow logic. (a) Hardware Architecture Logic The base carrier board is the core component, and all functional components are integrated into or connected to the carrier board through standardized interfaces. The connections and functions of each component are clearly defined. 1. Core component connection relationship: The core computing module is detachably connected to the base carrier board through a standardized module interface (high-density connector) - this interface meets the electrical connection requirements (transmission of power and data signals) and ensures physical stability; the general power supply system (including a 12V main power supply module and multiple sub-power supply units) supplies power to the core computing module through a standardized interface, and also supplies power to other components such as the BMC management unit and FPGA unit separately; 2. Data and Control Paths: The general-purpose high-speed interconnect bus is divided into PCIe and Ethernet interfaces. The PCIe interface is responsible for high-speed data interaction between the core computing module and external storage devices and accelerator cards. The Ethernet interface connects switch A and switch B (forming a redundant network) to ensure uninterrupted network communication. The two FPGA units have clearly defined functions: the first FPGA transmits service data to the core computing module via the SPI bus, and the second FPGA interacts with the BMC management unit to configure parameters via the I2C bus, avoiding management traffic from interfering with service data. The backplane MCU communicates with the BMC via the UART bus to assist in module status monitoring. 3. Auxiliary and monitoring components: Temperature sensors are placed on the surface of the core computing module and general power supply system to collect temperature data in real time and transmit it to the BMC; the base board also integrates memory slots (for DIMM memory), M.2 / U.2 storage interfaces, and expansion card slots (for GPU / network cards) to meet peripheral expansion needs; the power button and indicator light, and the UID button and indicator light are connected to the BMC for status prompts and error reports.

[0042] (ii) Control Flow Logic The BMC management unit is the "control center" of the entire blade, automating module adaptation and operational management throughout the entire process. The core process is as follows: 1. Status Detection: BMC monitors the presence status of the core computing modules in real time and senses module insertion or replacement actions; 2. Identification Reading: When module insertion or removal is detected, the BMC reads the identification information of the core computing module to confirm the module architecture (such as X86 / ARM) and model. 3. Firmware Retrieval and Verification: The BMC prioritizes searching the local pre-built firmware library based on the identification information. If no matching firmware is found, it connects to the remote server via the universal high-speed interconnect bus to obtain it from the remote firmware library. After obtaining the firmware, the validity of the firmware is verified to avoid loading invalid firmware. 4. Firmware Loading and Initialization: After successful verification, the BMC loads the firmware into the core computing module, and the control module completes initialization based on the firmware, adapts to the target architecture, and runs normally; if verification fails, the BMC triggers the power indicator light to report an error and alerts the maintenance personnel. 5. Special Scenario Management: Supports hot-swapping of core computing modules. When the BMC detects hot-swapping, it first controls the general power supply system to cut off the power supply to the module (to avoid damaging the hardware by hot-swapping). After the firmware is loaded and configured, the power supply is restored to complete the initialization. If the temperature collected by the temperature sensor exceeds the preset threshold, the BMC controls the general power supply system to reduce the output power (to prevent overheating) and triggers the UID indicator light alarm.

[0043] How to use The use of reusable server blades is designed around the principles of "low intervention and high flexibility," covering three core scenarios: daily operation, module replacement, and functional expansion. (I) Daily Operation 1. Deploy the blade server into a standard server chassis and connect the backplane power supply to the external network; 2. Based on the initial business requirements, plug DIMM memory modules, external storage devices (such as M.2 solid-state drives) or expansion cards (such as GPUs) into the general auxiliary interfaces of the base carrier board, and insert the core computing modules adapted to the architecture through standardized module interfaces; 3. After power-on, the BMC automatically completes the identification reading, firmware matching and initialization of the core computing module without manual configuration; after initialization, the blade can be connected to the server cluster to carry the corresponding business (such as high-performance computing, edge data processing). 4. During routine maintenance, the blade status can be observed through the UID indicator and power indicator. If the indicator is abnormal, locate the problem by combining it with BMC alarm information (such as firmware incompatibility or excessive temperature).

[0044] (ii) Replacement of core computing module When business requirements change (such as switching from x86 architecture to ARM architecture) or core modules need to be upgraded: 1. No need to shut down the entire server power supply (supports hot-swapping), the old core computing module can be directly removed from the standardized module interface; 2. When a new core computing module is inserted, the BMC automatically detects the module replacement and automatically processes it according to the process of "reading identifier → retrieving firmware → verifying → loading → initializing". During this process, the power supply to the new module will be briefly cut off (to protect the hardware). Power supply will be restored after the process is completed. 3. After initialization, the blade automatically switches to the architecture corresponding to the new module and runs without manual configuration.

[0045] (III) Functional Expansion When it is necessary to increase the memory capacity, storage capacity, or computing performance of the blade: 1. To expand memory, simply insert the DIMM memory module into the memory slot of the base carrier board. The BMC will automatically recognize and enable it. 2. To expand storage, connect an external storage device (such as a U.2 hard drive) to the corresponding M.2 / U.2 interface. The BMC will automatically detect the storage device and complete the adaptation. 3. To enhance computing power (such as AI inference), insert a dedicated GPU or accelerator card into the expansion card slot. Establish a high-speed data path with the core computing module through the PCIe interface. The BMC will automatically recognize the expansion card and load the appropriate driver (based on firmware library support).

[0046] The above are merely preferred embodiments of the present invention and are not intended to limit the implementation methods and protection scope of the present invention. Those skilled in the art should recognize that any equivalent substitutions and obvious changes made based on the description and illustrations of the present invention should be included within the protection scope of the present invention.

Claims

1. A reusable server blade, characterized in that, Includes a base carrier board, and integrated on the base carrier board the core computing module, BMC management unit, general power supply system, standardized module interface, general high-speed interconnect bus, FPGA unit, backplane MCU, temperature sensor, power button and power indicator, module indicator, UID button and UID indicator; The core computing module is detachably connected to the base carrier board through the standardized module interface. The universal power supply system is electrically connected to the standardized module interface to provide power to the core computing module. The first end of the universal high-speed interconnect bus is electrically connected to the standardized module interface, and the second end extends to the backplane connection end of the base carrier board to realize the connection of external I / O devices. The BMC management unit is electrically connected to the standardized module interface, the general power supply system, the general high-speed interconnect bus, the FPGA unit, the backplane MCU, the temperature sensor, the power button and power indicator, the module indicator, and the UID button and UID indicator. The BMC management unit communicates with the backplane MCU via a UART bus, the backplane MCU communicates with the FPGA unit via a UART, and the FPGA unit communicates with the core computing module via I2C. The BMC management unit is configured to: respond to the plugging / unplugging event of the core computing module through the standardized module interface, and, based on the identification information of the core computing module, call and load the corresponding firmware to control the core computing module to complete initialization and run normally.

2. The reusable server blade according to claim 1, characterized in that, The core computing module is at least one heterogeneous computing module, and the architecture of the heterogeneous computing module includes x86 architecture or ARM architecture, and the core computing module integrates a multi-core SoC.

3. The reusable server blade according to claim 1, characterized in that, The standardized module interface is a high-density connector. The pin configuration of the high-density connector meets the electrical connection requirements between the core computing module and the general power supply system, general high-speed interconnect bus, FPGA unit, and BMC management unit. Furthermore, the physical size of the high-density connector is adapted to the preset mounting position on the base carrier to achieve stable insertion and removal of the core computing module.

4. The reusable server blade according to claim 1, characterized in that, The general power supply system includes a 12V main power module and multiple sub-power units. The 12V main power module provides input power to the multiple sub-power units. The multiple sub-power units supply power to the core computing module, the BMC management unit, and the FPGA unit, respectively, and the output voltage of each sub-power unit can be dynamically adjusted according to the corresponding load requirements.

5. The reusable server blade according to claim 1, characterized in that, The BMC management unit is also communicatively connected to a firmware library, which includes a pre-built firmware library and a remote firmware library. The BMC management unit is configured to retrieve a matching firmware from the pre-built firmware library or from the remote firmware library via the universal high-speed interconnect bus based on the identification information of the core computing module.

6. The reusable server blade according to claim 1, characterized in that, The general-purpose high-speed interconnect bus includes a PCIe interface and an Ethernet interface; the PCIe interface is used for high-speed data interaction between the core computing module and external storage devices and accelerator cards; the Ethernet interface is connected to a switch unit on the base carrier board, the switch unit includes switch A and switch B, and switch A and switch B communicate with the BMC management unit through the Ethernet bus to achieve redundant network connections.

7. The reusable server blade according to claim 1, characterized in that, The FPGA unit consists of two units. The FPGA unit communicates bidirectionally with the core computing module via an I2C bus to forward data, and communicates bidirectionally with the BMC management unit via a UART through a backplane MCU to enable the BMC management unit to read and write configuration parameters of the core computing module.

8. The reusable server blade according to claim 1, characterized in that, At least two temperature sensors are arranged on the surface of the core computing module and the surface of the general power supply system, respectively. The temperature sensors collect temperature data in real time and transmit it to the BMC management unit. The BMC management unit is also configured to control the general power supply system to reduce the output power when the collected temperature data exceeds a preset threshold, and to issue an alarm through the module indicator light.

9. The reusable server blade according to claim 1, characterized in that, The base carrier board also integrates a general auxiliary component interface, which includes a memory slot, an M.2 storage interface, a U.2 storage interface, and an expansion card slot. The memory slot is used to insert a DIMM memory module, the M.2 storage interface and the U.2 storage interface are used to connect external storage devices, and the expansion card slot is used to insert a discrete GPU or network card.

10. The reusable server blade according to any one of claims 1-9, characterized in that, The core computing module supports plugging and unplugging operations under normal main power supply conditions of the server blade; the BMC management unit is further configured to: when the plugging and unplugging operation of the core computing module is detected, first control the general power supply system to cut off the power supply to the core computing module, and after the firmware loading and configuration are completed, control the general power supply system to restore the power supply to the core computing module to complete the initialization.

Images