Computing node and server

Abstract

The invention discloses a computing node and a server, and relates to the technical field of servers, the computing node comprises a cold plate tray, a mainboard and a functional device, the mainboard is carried on the cold plate tray so as to dissipate heat through the cold plate tray, the functional device is fixed on the cold plate tray and is electrically connected with the mainboard, the cold plate tray is provided with a groove, and the groove is formed in the cold plate tray. The groove is located in the side, away from the main board, of the cold plate tray and used for being matched with a partition plate of the server. The cold plate tray integrates the functions of a tray and a cold plate, so that the thickness space of the computational node is saved. And the functional device is fixed on the cold plate tray and is electrically connected with the mainboard, so that the layout of the components is simplified, too tight arrangement among the components is avoided, and the blocking problem of a heat dissipation channel is reduced. The cold plate tray is provided with a groove, so that the influence of the thickness of a partition plate between the computing nodes on the height direction of the server is counteracted. Therefore, according to the computing node, the computing expansion capability and the heat dissipation efficiency are improved.

Description

Technical Field

[0001] The present application relates to the technical field of servers, and in particular to a computing node and a server. Background Art

[0002] With the economic development of society, especially with the development of artificial intelligence technology, higher demands are placed on the computing power of IT equipment such as servers.

[0003] In related technologies, limited by traditional chassis design and component layout, computing expansion capabilities are insufficient, making it difficult to achieve high-density deployment of CPUs and unable to meet the data processing density requirements of data-intensive applications. At the same time, in pursuit of high integration, internal computing, storage, network and other components are arranged too closely, resulting in narrow heat dissipation channels, making it difficult for the air cooling system to effectively remove heat, and limited space for laying liquid cooling pipes, which seriously restricts the stable operation of equipment in high-density environments. Summary of the invention

[0004] The present application provides a computing node and a server to at least solve the problems of insufficient computing expansion capability and low heat dissipation efficiency in the related art.

[0005] The present application provides a computing node, including a cold plate tray, a mainboard and a functional device. The mainboard is carried on the cold plate tray to dissipate heat through the cold plate tray. The functional device is fixed to the cold plate tray and is electrically connected to the mainboard. The cold plate tray is provided with a groove, which is located on a side of the cold plate tray away from the mainboard. The groove is used to cooperate with a partition of a server.

[0006] The present application also provides a server, including a chassis, a partition, and a plurality of the above-mentioned computing nodes, wherein the computing nodes are inserted into the chassis through the partition.

[0007] With the computing node and server of the present application, since the computing node adopts a cold plate tray design and the main board is directly carried on the cold plate tray, heat can be conducted and dissipated more efficiently through the cold plate tray. Compared with the traditional air-cooled system, the cold plate tray provides a larger heat dissipation surface area, significantly improving the heat dissipation efficiency, thus ensuring the stable operation of the device under high-density deployment. Compared with the traditional laying of liquid cooling pipelines, the cold plate tray integrates the functions of the tray and the cold plate, thus saving the thickness space of the computing node. The functional device is fixed to the cold plate tray and electrically connected to the main board. This design not only simplifies the layout of the components but also avoids overly close arrangements between the components, reducing the blockage problem of the heat dissipation channels. Through a reasonable layout design, the computing node can achieve a higher integration level without increasing the volume. The cold plate tray is provided with grooves, which cooperate with the partitions of the server, providing flexible installation adaptability. At the same time, the design of the grooves offsets the influence of the partition thickness between computing nodes on the height direction of the server. This design allows the computing node to be quickly deployed in different types of server environments, enhancing the compatibility and adaptability of the device. Therefore, the computing node of the present invention not only improves the computing expansion ability and heat dissipation efficiency but also enhances the stability and applicability of the device, and can better meet the needs of data-intensive applications for high-density data processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] To more clearly illustrate the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0009] Figure 1 Structural schematic diagram of a server provided by an embodiment of the present application;

[0010] Figure 2 For Figure 1 Partial exploded structural schematic diagram of the computing node of the shown server;

[0011] Figure 3 For Figure 1 Structural schematic diagram of another perspective of the shown computing node;

[0012] Figure 4 For Figure 1 Structural schematic diagram of the cooperation between the computing node and the partition in the shown server;

[0013] Figure 5 For Figure 2 Structural schematic diagram of the cooperation between the cold plate tray and the processor on the shown computing node;

[0014] Figure 6 ForFigure 5 One of the structural schematic diagrams of the shown shrapnel screw;

[0015] Figure 7 For Figure 5 Two of the structural schematic diagrams of the shown shrapnel screw;

[0016] Figure 8 For Figure 2 One of the structural schematic diagrams of the cold plate tray of the shown computing node;

[0017] Figure 9 For Figure 2 Two of the structural schematic diagrams of the cold plate tray of the shown computing node;

[0018] Figure 10 For Figure 2 The structural schematic diagram of the functional device of the shown computing node;

[0019] Figure 11 For Figure 2 The structural schematic diagram of the cooperation between the functional device of the shown computing node and the hard disk backplane;

[0020] Figure 12 For Figure 2 The structural schematic diagram of the cooperation between the shown computing node and the power backplane;

[0021] Figure 13 For Figure 2 The structural schematic diagram of the cooperation between the memory of the shown computing node and the cold plate tray;

[0022] Figure 14 For Figure 13 The structural schematic diagram of the cooperation between the second memory and the memory cold plate;

[0023] Figure 15 For Figure 2 The structural schematic diagram of the cooperation between the shown computing node and the water diverter.

[0024] Among them, the above-mentioned drawings include the following reference numerals:

[0025] 100 - Computing node; 10 - Cold plate tray; 11 - Main cold plate; 111 - Groove; 12 - Main heat exchange flow channel; 13 - Processor cold plate; 14 - Processor flow channel; 15 - Blind plug; 17 - Bottom plate; 18 - Side plate; 19 - Receiving groove; 16 - Fixed stud; 20 - Motherboard; 21 - Processor; 22 - Elastic screw; 221 - Stud; 222 - Elastic piece; 23 - Power connector; 24 - Memory; 241 - First memory; 242 - Second memory; 25 - Memory cold plate; 251 - Cold plate boss; 26 - Signal connector; 30 - Functional device; 31 - Functional component; 32 - Functional cold plate; 321 - Heat conduction block; 33 - Connector; 34 - Hard disk unit; 40 - Hard disk backplane; 41 - Hard disk connector; 200 - Server; 201 - Chassis; 202 - Partition; 203 - Power backplane; 204 - Water inlet distributor; 205 - Water outlet distributor; 206 - Water inlet; 207 - Water outlet. Detailed implementation manner

[0026] The following will clearly and completely describe the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the protection scope of the present application.

[0027] It should be noted that the orientation or positional relationship indicated by terms such as "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation to the present application. The terms "mounted", "connected", and "coupled" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements. The terms "parallel", "perpendicular", and "equal" include the described situations and situations similar to the described situations, and the range of the similar situations is within an acceptable deviation range, where the acceptable deviation range is determined by those of ordinary skill in the art considering the measurements being discussed and the errors associated with the measurements of specific quantities (i.e., the limitations of the measurement system). For example, "parallel" includes absolute parallel and approximate parallel, where the acceptable deviation range for approximate parallel can be, for example, within 5° deviation; "perpendicular" includes absolute perpendicular and approximate perpendicular, where the acceptable deviation range for approximate perpendicular can also be, for example, within 5° deviation. "Equal" includes absolute equality and approximate equality, where the acceptable deviation range for approximate equality can be, for example, that the difference between the two equal ones is less than or equal to 5% of either one. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to specific situations.

[0028] In order to enable those skilled in the art of the present technology to better understand the solution of the present application, the present application will be further described in detail below with reference to the drawings and specific embodiments.

[0029] Figure 1 It is a schematic structural diagram of a server provided by an embodiment of the present application. Figure 2 For Figure 1 A partial exploded structural diagram of the computing node of the server shown.

[0030] As Figure 1 And Figure 2As shown, the computing node 100 provided by the embodiment of the present application is used in the server 200. The server 200 includes a chassis 201, a partition 202, and the computing node 100. The partition 202 is disposed in the chassis 201. The computing node 100 is inserted into the chassis 201 through the partition 202. Specifically, the partition 202 divides the internal space of the chassis 201 into multiple sub-spaces for accommodating the computing node 100. Each sub-space is used to accommodate one computing node 100. The computing node is inserted into the sub-space, and the partition 202 is located between adjacent computing nodes 100.

[0031] In some possible implementation manners, the computing node 100 includes a cold plate tray 10, a main board 20, and a functional device 30. The cold plate tray 10 is a frame part of the computing node 100 and is used to provide support and heat dissipation functions. The main board 20 is carried on the cold plate tray 10 to dissipate heat through the cold plate tray 10. The functional device 30 is fixed to the cold plate tray 10 and is electrically connected to the main board 20. The functional device 30 also dissipates heat through the cold plate tray 10.

[0032] By adopting the integrated cold plate method, the cold plate tray 10 also serves as a tray for the main board 20, solving the problem of cumbersome maintenance of liquid cooling components. At the same time, the main board 20 can be buckled and locked on the cold plate tray 10, enabling small-power devices to be in contact with the cold plate tray 10 for simultaneous heat dissipation of multiple heat sources, and also solving the heat dissipation problem of small-power devices, realizing a zero-fan all-liquid cooling design. This saves the thickness in the chassis 201.

[0033] In some possible implementation manners, in the computing node 100 of the present application, the thickness of the cold plate tray 10 is 6 mm, and the thickness of the main board 20 is 3 mm. In the conventionally configured computing node 100, the thickness of a single cold plate is 12 mm, and the thickness of the main board thereon is 2.4 mm.

[0034] The front window of each computing node 100 is a functional area where the functional device 30 can be arranged.

[0035] Since the computing node 100 adopts the cold plate tray 10 design, the main board 20 is directly carried on the cold plate tray 10, enabling heat to be conducted and dissipated more efficiently through the cold plate tray 10. Compared with the traditional air-cooled system, the cold plate tray 10 provides a larger heat dissipation surface area, significantly improving the heat dissipation efficiency, thus ensuring the stable operation of the device under high-density deployment. Compared with the traditional laying of liquid cooling pipelines, the cold plate tray 10 integrates the functions of the tray and the cold plate, thus saving the thickness space of the computing node 100. The functional device 30 is fixed to the cold plate tray 10 and electrically connected to the main board 20. This design not only simplifies the layout of the components but also avoids overly close arrangements between the components, reducing the blockage problem of the heat dissipation channels. Through reasonable layout design, the computing node 100 can achieve higher integration without increasing the volume.

[0036] Figure 3 For Figure 1 The structural schematic diagram of another perspective of the computing node shown. Figure 4 For Figure 1 The structural schematic diagram of the cooperation between the computing node and the partition in the server shown.

[0037] As Figure 3 And Figure 4 As shown, the cold plate tray 10 is provided with a groove 111. The groove 111 is located on the side of the cold plate tray 10 facing away from the main board 20. The groove 111 is used to cooperate with the partition 202 of the server 200.

[0038] The groove 111 is provided at two edges of the cold plate tray 10. When the computing node 100 is inserted into the server 200, it can accommodate the partition 202, thereby reducing the space occupied by the computing node 100 in the server 200.

[0039] The cold plate tray 10 is provided with a groove 111, which cooperates with the partition 202 of the server 200, providing flexible installation adaptability. At the same time, the design of the groove 111 offsets the influence of the thickness of the partition 202 between the computing nodes 100 on the height direction of the server 200. This design allows the computing node 100 to be quickly deployed in different types of server 200 environments, enhancing the compatibility and adaptability of the device. Therefore, the computing node 100 of the present invention not only improves the computing expansion ability and heat dissipation efficiency but also enhances the stability and applicability of the device, and can better meet the requirements of data-intensive applications for high-density data processing.

[0040] Figure 5 For Figure 2 The structural schematic diagram of the cooperation between the cold plate tray and the processor on the computing node shown.

[0041] As Figure 5As shown, in some possible implementations, the computing node 100 further includes a spring screw 22 and a processor 21. The processor 21 is disposed on a side of the mainboard 20 facing the cold plate tray 10. The spring screw 22 passes through the cold plate tray 10 and is connected to the processor 21, so that the processor 21 contacts the cold plate tray 10.

[0042] By optimizing the fixing design of the processor 21, the spring screw is changed into a spring screw 22. Under the premise of maintaining the snap-fitting force of the processor 21 slot, the space required for installing the processor 21 is compressed.

[0043] The processor 21 is directly arranged on the side of the mainboard 20 facing the cold plate tray 10, and is in close contact with the cold plate tray 10 through the spring screw 22. This design ensures that the heat generated by the processor 21 can be quickly transferred to the cold plate tray 10, thereby improving the heat dissipation efficiency. Compared with traditional heat dissipation methods, this direct contact method reduces thermal resistance and improves the speed and efficiency of heat transfer. The layout design of the processor 21 makes it close to the cold plate tray 10, saving internal space. This compact design helps to achieve higher component integration within the limited space of the computing node 100, and provides more flexibility for the layout of other functional devices.

[0044] By using the spring screw 22 to fix the processor 21 on the cold plate tray 10, additional mechanical stability is provided. The elastic characteristics of the spring screw 22 can absorb certain vibrations and shocks, protect the processor 21 from external physical stress, and thus improve the reliability of the computing node 100 during operation. At the same time, the spring screw 22 not only ensures a firm connection, but also facilitates the rapid replacement or maintenance of the processor 21 when necessary, making the installation and removal of the processor 21 easier, reducing equipment downtime and improving maintenance efficiency.

[0045] Figure 6 for Figure 5 One of the structural schematic diagrams of the spring screw shown. Figure 7 for Figure 5 The second structural schematic diagram of the spring screw shown.

[0046] like Figure 6 and Figure 7 As shown, in some possible implementations, the spring screw 22 includes a spring 222 and a stud 221. The stud 221 passes through the cold plate tray 10 and is connected to the processor 21. The spring 222 is sleeved on the stud 221 and is disposed on a side of the cold plate tray 10 away from the processor 21.

[0047] The combined design of the stud 221 and the spring 222 allows for quick installation and removal of the processor 21, reducing maintenance time and complexity while ensuring that components are not damaged during maintenance, making the installation and maintenance process easier.

[0048] The stud 221 passes through the cold plate tray 10 and is directly connected to the processor 21, ensuring close contact between the processor 21 and the cold plate tray 10. This design minimizes the thermal resistance and improves the efficiency of heat conduction from the processor 21 to the cold plate tray 10, thereby enhancing the overall heat dissipation performance.

[0049] The shrapnel 222 is sleeved on the stud 221 and is located on the side of the cold plate tray 10 away from the processor 21. This configuration can provide a uniform pressure distribution when pressure is applied by the stud 221. The elastic characteristics of the shrapnel 222 allow it to deform within a certain range, thus maintaining a stable contact pressure between the processor 21 and the cold plate tray 10 and avoiding poor contact caused by thermal expansion or mechanical stress. The use of the shrapnel 222 not only provides stable pressure but also has good shock absorption and anti-vibration capabilities. This is particularly important for devices operating in high-performance computing environments because it can reduce the impact of mechanical stress caused by vibration on the processor 21 and other components, improving the reliability and durability of the system. At the same time, by setting the shrapnel 222 on the side of the cold plate tray 10 away from the processor 21, the design effectively utilizes the space, avoids occupying the space around the processor 21, and provides a more flexible component layout solution.

[0050] To maintain the clamping force F of the processor 21, the cross-sectional dimensions of the shrapnel 222 are designed through the calculation formula F = (ESh 3 )(L - x) / (4L 3 ), where E is the elastic modulus of the material of the shrapnel 222, L is the height of the shrapnel 222, x is the compression amount, S is the effective perimeter of the shrapnel 222, generally the perimeter at the midpoint of the inner and outer diameters of the shrapnel 222 can be taken for estimation, and h is the thickness of the cross-section of the shrapnel 222.

[0051] Among some possible implementation manners, L is between 2 mm and 4 mm, and x is between 0.5 mm and 1 mm. The remaining design parameters can be comprehensively varied as long as the clamping force F meets the clamping requirements of the processor 21.

[0052] In some possible implementation manners, the clamping force F requirement of the processor 21 is between 175 kgf and 205 kgf.

[0053] Figure 8 For Figure 2 One of the structural schematic diagrams of the cold plate tray of the computing node shown. Figure 9 For Figure 2 Another structural schematic diagram of the cold plate tray of the computing node shown.

[0054] Such as Figure 8 And Figure 9As shown, in some possible implementation manners, the cold plate tray 10 is provided with a main cold plate 11, a main heat exchange flow channel 12, a processor cold plate 13 and a processor flow channel 14. The main heat exchange flow channel 12 is arranged on the main cold plate 11, and the main board 20 is carried on the main cold plate 11 to dissipate heat through the main heat exchange flow channel 12 on the main cold plate 11. The processor cold plate 13 is arranged on the main cold plate 11 and is opposite to the processor 21. The processor cold plate 13 is used to dissipate heat from the processor 21. The processor flow channel 14 is connected to the processor cold plate 13.

[0055] The integrated design of the cold plate tray 10 reduces the number of cold plate components and the number of fluid connectors, thereby reducing the material cost and improving the maintainability.

[0056] The independent design of the main heat exchange flow channel 12 and the processor flow channel 14 allows for separate management and optimization of different heat sources. In this way, the fluid flow rate and flow volume can be adjusted according to the heat load requirements of different components to achieve more precise and efficient heat management. Through this split design of the cold plate and the flow channel, the system can maintain a lower temperature during high-performance operation, thereby improving the stability and reliability of the system, extending the service life of the components, and ensuring the performance of the device under high-load conditions. At the same time, the modular characteristics of this design make it convenient to perform maintenance and component upgrades when needed. The independent flow channel design also facilitates local cooling system adjustment or maintenance without affecting other components.

[0057] The main heat exchange flow channel 12 is arranged on the main cold plate 11, and the main board 20 is directly carried on the main cold plate 11 for heat dissipation. This design allows the heat generated by the main board 12 to be quickly conducted to the main heat exchange flow channel 12 through the main cold plate 11, and the heat is carried away through the flow of the fluid, thereby improving the overall heat dissipation efficiency.

[0058] The processor cold plate 13 is specifically arranged on the main cold plate 11 and is opposite to the processor 21, and the processor flow channel 14 is connected to the processor cold plate 13. This dedicated heat dissipation path ensures that the heat generated by the processor 21 can be quickly conducted to the processor cold plate 13 and effective heat exchange is carried out through the processor flow channel 14, which is particularly suitable for the high heat density heat dissipation requirements of the processor 21.

[0059] In some possible implementation manners, the main cold plate 11 is made of aluminum, and the processor cold plate 13 is made of copper. The processor cold plate 13 is locked to the main cold plate 11, thereby integrating the cold plate tray 10.

[0060] Figure 10 For Figure 2 The structural schematic diagram of the functional device of the computing node shown.

[0061] As Figure 10As shown, in some possible implementations, the cold plate tray 10 is provided with a blind plug 15. The blind plug 15 is provided on a side of the cold plate tray 10 away from the functional device 30. The main heat exchange flow channel 12 and the processor flow channel 14 are respectively connected to the blind plug 15.

[0062] Since the blind plug 15 is located on the side of the cold plate tray 10 away from the functional device 30, automatic docking can be achieved without precise alignment during installation, which greatly simplifies the installation process and reduces the complexity and time of manual operation. The blind plug 15 is located on the side of the cold plate tray 10 away from the functional device 30. This layout optimizes the use of internal space, avoids interference of cooling connections on the functional device 30, and provides more flexibility for the arrangement of other components.

[0063] The Blind Plug 15 design reduces the risk of misoperation that may occur during the connection process, such as damage caused by incorrect connection or excessive force. Its automatic docking feature ensures the reliability and consistency of each connection, reducing the possibility of equipment failure. The use of Blind Plug 15 makes connecting and disconnecting the cooling system simpler and faster. This modular design facilitates system maintenance and upgrades, allowing the cooling system to be adjusted or replaced without affecting other components, improving the maintainability of the system.

[0064] The design of the blind plug 15 allows the main heat exchange flow channel 12 and the processor flow channel 14 to be quickly and easily connected to the external cooling system. Through the design of the blind plug 15, the system can be more flexibly adapted to different cooling requirements and configuration changes. Whether adding a new cooling circuit or adjusting an existing cooling solution, the blind plug 15 can provide convenient interface support.

[0065] In some possible implementations, at one blind plug 15, the main liquid inlet channel connected to the blind plug 15 is divided into two to form the liquid inlet channel of the main heat exchange channel 12 and the processor channel 14, and at another blind plug 15, the liquid outlet channels of the main heat exchange channel 12 and the processor channel 14 are combined into one to form the main liquid outlet channel connected to the blind plug 15.

[0066] In some possible implementations, the functional device 30 includes a functional component 31 and a functional cold plate 32. The functional component 31 is disposed on the functional cold plate 32. The functional cold plate 32 is provided with a heat conducting block 321. The functional cold plate 32 is fixed to the cold plate tray 10, and the heat conducting block 321 is in contact with the cold plate tray 10.

[0067] For the functional component 31 that needs to be hot-swapped, the functional component 31 and the functional cold plate 32 can be fixed into a module first, and then slid into the cold plate tray 10 , so that the heat conduction block 321 on the functional cold plate 32 contacts the cold plate tray 10 .

[0068] The heat conduction block 321 on the functional cold plate 32 is in direct contact with the cold plate tray 10, ensuring that the heat generated by the functional component 31 can be quickly conducted to the cold plate tray 10. This design effectively utilizes the heat dissipation capacity of the cold plate tray 10, improves the heat dissipation efficiency of the functional component 31, and ensures the stable operation of the device under high-load conditions.

[0069] By arranging the functional component 31 on the functional cold plate 32 and using the heat conduction block 321 for heat conduction, the system can better manage the heat of different components. This design allows for independent management of the heat load of the functional component 31, prevents local overheating, and improves the overall thermal balance of the system. The contact between the heat conduction block 321 and the cold plate tray 10 not only provides a good heat conduction path but also enhances the mechanical stability of the functional component 31. Through this tight structural connection, the risk of component loosening or damage caused by vibration or impact is reduced.

[0070] The functional cold plate 32 is fixed to the cold plate tray 10. This modular design makes the installation and disassembly of the functional component 31 more convenient. Users can easily replace or upgrade the functional component 31 without having to perform large-scale disassembly of the entire system, reducing maintenance time and complexity. This design allows for flexible adjustment or replacement of the functional component 31 without affecting other components, enabling the system to quickly adapt to different application requirements and environmental changes and improving the adaptability of the device.

[0071] In some possible implementation manners, the functional device 30 further includes a connector 33. The connector 33 is disposed on one side of the functional device 30 facing the main board 20. The functional component 31 is electrically connected to the main board 20 through the connector 33.

[0072] Through the layout of the connector 33, the functional device 30 can be plugged into the main board 20 and fixed in the cold plate tray 10 by screws or the like, realizing a cable-free design with good maintainability and convenient assembly. A series of impacts caused by cable wiring are reduced.

[0073] The connector 33 is disposed on one side of the functional device 30 facing the main board 20, enabling the functional component 31 to be directly electrically connected to the main board 20 through the connector 33. This design simplifies the electrical connection process, reduces the complexity of wiring and possible wiring errors, and improves the installation efficiency. By using the connector 33 for electrical connection, the installation and disassembly of the functional component 31 become simpler and faster. This modular design facilitates the maintenance and upgrade of the system, allowing for replacement or upgrade of the functional component 31 without affecting other components and improving the maintainability of the system. At the same time, the design of the connector 33 makes the electrical connection more compact, reducing the occupation of internal space. This layout optimization provides more flexibility for the arrangement of other components and enhances the overall integration of the system.

[0074] The connector 33 provides a stable electrical connection interface, reducing interference and loss during signal transmission. With this design, the signal transmission between the functional component 31 and the main board 20 is more reliable, ensuring the stable operation of the system. The use of the connector 33 allows the functional component 31 to be quickly configured and reorganized in different application scenarios. This flexibility enables the system to adapt to different functional requirements and environmental changes, improving the adaptability and expandability of the device.

[0075] In some possible implementation manners, the heat conducting block 321 is disposed on one side of the functional cold plate 32 facing the main board 20 and is arranged side by side with the connector 33.

[0076] After the functional cold plate 32 near the tail connector 33 avoids the connector 33, it contacts the cold plate tray 10 through the heat conducting block 321 at the tail. The cross-sectional area of the heat conducting block 321 is made as large as possible, and this surface is used to transfer the heat on the functional component 31 to the cold plate tray 10 for heat dissipation.

[0077] The heat conducting block 321 is directly disposed on one side of the functional cold plate 32 facing the main board 20, enabling the heat to be conducted more directly to the cold plate tray 10. This close heat conduction path reduces the thermal resistance, improves the heat dissipation efficiency of the functional component 31, and ensures the stable operation of the device under high load conditions.

[0078] Arranging the heat conducting block 321 side by side with the connector 33 effectively utilizes the limited space. This layout design enables the electrical connection and heat management functions to be achieved on the same plane, avoiding mutual interference between components and optimizing the use of the internal space. This design allows for a higher integration degree within the limited space, supporting the implementation of more functions. This high integration degree is particularly important for application scenarios that require high performance and a compact design.

[0079] By arranging the heat conducting block 321 and the connector 33 side by side, the installation and maintenance processes become more convenient. The compact layout of the components reduces the installation steps and maintenance time, improving the maintainability of the system. The heat conducting block 321 and the connector 33 arranged side by side provide better mechanical stability. The tight structural layout reduces the risk of component loosening or damage caused by vibration or impact, improving the reliability of the system.

[0080] In some possible implementation manners, the cross-section of the heat conducting block 321 is 18mm * 15mm.

[0081] Figure 11 For Figure 2 Schematic structural diagram of the cooperation between the functional device of the computing node and the hard disk backplane shown.

[0082] As Figure 11As shown, in some possible implementation manners, the computing node 100 further includes a hard disk backplane 40. The hard disk backplane 40 is provided with a hard disk connector 41. The functional component 31 includes a hard disk unit 34. The hard disk unit 34 is electrically connected to the hard disk backplane 40 through a connector 33. The hard disk backplane 40 is electrically connected to the main board 20 through the hard disk connector 41.

[0083] Through the layout of the hard disk connector 41, a cable-free design is realized, which has good maintainability and is convenient for assembly. A series of impacts caused by cable wiring are reduced.

[0084] The hard disk unit 34 is electrically connected to the hard disk backplane 40 through a connector 33, and the hard disk backplane 40 is electrically connected to the main board 20 through the hard disk connector 41. This design simplifies the connection path between the hard disk unit 34 and the main board 20, reduces the wiring complexity, and improves the installation and configuration efficiency of the system.

[0085] The hard disk backplane 40 and the hard disk connector 41 provide a stable electrical connection interface, reducing interference and loss during signal transmission. Through this design, the signal transmission between the hard disk unit 34 and the main board 20 is more reliable, ensuring data integrity and stable operation of the system. The design of the hard disk backplane 40 makes the layout of the hard disk unit 34 more compact, reducing the occupation of internal space. This layout optimization provides more flexibility for the arrangement of other components, improving the overall integration of the system.

[0086] The use of the hard disk backplane 40 makes the installation and removal of the hard disk unit 34 more convenient. Users can easily replace or upgrade the hard disk without having to disassemble the entire system on a large scale, reducing maintenance time and complexity. Through the hard disk backplane 40, the system can more flexibly support different numbers and types of hard disk units 34. This design allows for quick adjustment of the hard disk configuration according to changes in storage requirements, improving the scalability and adaptability of the system.

[0087] In some possible implementation manners, the functional component 31 further includes an input / output module, a network interface module, and a security control module.

[0088] By integrating the input / output module, the network interface module, and the security control module, the computing node 100 can provide rich functional support. This diverse design of the functional device 30 enables the system to meet the requirements of different application scenarios, thereby improving the applicability of the device. This modular design allows for flexible configuration of different functional devices 30 according to specific requirements. Users can select appropriate module combinations according to application needs, thereby improving the flexibility and scalability of the system.

[0089] The integration of the network interface module provides an efficient data transmission path, supporting higher network bandwidth and lower latency. This is particularly important for application scenarios that require fast data exchange and high network performance. The integration of the security control module provides an additional security layer for the system, capable of effectively managing and protecting data access, authentication, and other security functions. This design improves the overall security of the system and protects sensitive data from unauthorized access.

[0090] By integrating these functional modules, the overall design of the system is more compact, reducing the need for external devices and connections. This integrated design simplifies the management and maintenance of the system and improves operational efficiency. By integrating multiple functional modules into the functional component 31, the system can make more effective use of internal space. This compact layout design provides more flexibility for the arrangement of other components and enhances the overall integration of the system.

[0091] Among them, an input / output module, a network interface module, a security control module, and a hard disk unit can be respectively set on each computing node 100.

[0092] Figure 12 For Figure 2 the structural schematic diagram of the cooperation between the shown computing node and the power backplane.

[0093] As Figure 12 shown, in some possible implementation manners, a power connector 23 is provided on a side of the main board 20 facing away from the functional device 30. The power connector 23 is used to be electrically connected to the power backplane 203 of the server 200 when the computing node 100 is inserted into the server 200.

[0094] Through the layout of the power connector 23, a cable-free design is achieved, with good maintainability and convenient assembly. A series of impacts caused by cable wiring are reduced.

[0095] The power connector 23 is directly provided on the main board 20, enabling the computing node 100 to quickly dock with the power backplane 203 of the server 200 when inserted into the server 200. This design simplifies the power connection process, reduces the installation time and complexity, and improves the deployment efficiency of the system. Through the directly docked power connector 23 design, the risk of misoperation during the connection process is reduced, such as connection errors or damage caused by excessive force. Its automatic docking feature ensures the reliability and consistency of each connection. The power connector 23 provides a stable electrical connection interface, ensuring reliable power transmission. Through this design, the computing node 100 can obtain a stable power supply, reducing the risk of failures caused by poor power connection.

[0096] Setting the power connector 23 on the side of the main board 20 away from the functional device 30 helps to optimize the use of the internal space. It avoids the entanglement of power cords between the functional device 30 and other components, providing more flexibility for the layout of other components. Through the design of the power connector 23, the computing node 100 can be easily inserted into or removed from the server 200 without complex power cord connections. This modular design facilitates the maintenance and upgrade of the system, allowing for the quick replacement or maintenance of the computing node 100, improving the maintainability of the system. At the same time, this design allows the computing node 100 to be quickly deployed and reconfigured in different server 200 environments. Users can easily adjust the configuration of the computing node 100 according to their needs, improving the flexibility and scalability of the system.

[0097] Figure 13 For Figure 2 Schematic diagram of the structure where the memory of the shown computing node cooperates with the cold plate tray.

[0098] As Figure 13 As shown, in some possible implementation manners, the computing node 100 further includes a memory 24. The memory 24 is disposed on the side of the main board 20 facing the cold plate tray 10 so that the memory 24 contacts the cold plate tray 10.

[0099] The memory 24 directly contacts the cold plate tray 10, which can effectively conduct the heat generated by the memory 24 to the cold plate tray 10 for heat dissipation. This design improves the heat dissipation efficiency of the memory 24, ensures the stable operation of the memory module under high load conditions, and extends the service life of the memory 24.

[0100] At the same time, setting the memory 24 on the side of the main board 20 facing the cold plate tray 10 helps to optimize the use of the internal space of the system. This layout design reduces space waste and provides more flexibility for the arrangement of other components. By directly contacting the memory 24 with the cold plate tray 10, the need for additional heat dissipation devices is reduced. This design simplifies the heat dissipation solution of the system, reduces the complexity and cost of the system. This design allows for the integration of more memory 24 in a limited space, supports higher memory 24 capacity and density, and meets the requirements of high-performance computing and data-intensive applications.

[0101] Through effective heat dissipation management, the operating temperature of the memory 24 is controlled within a safe range, reducing the risk of failures caused by overheating and improving the overall reliability of the system. Through effective heat dissipation management, the memory 24 can operate at a lower temperature, which helps to improve the performance and response speed of the memory 24. The lower operating temperature can reduce heat-induced latency and errors, thus enhancing the overall system performance.

[0102] Figure 14 For Figure 13Schematic diagram of the structure of the second memory cooperating with the memory cold plate shown.

[0103] As Figure 14 shown, in some possible implementation manners, the computing node 100 further includes a memory cold plate 25. The memory 24 includes a first memory 241 and a second memory 242. The first memory 241 and the second memory 242 are stacked on the main board 20. The first memory 241 is in direct contact with the cold plate tray 10. The memory cold plate 25 is disposed between the first memory 241 and the second memory 242 and is fixedly connected to the second memory 242. The memory cold plate 25 is provided with a cold plate boss 251, and the memory cold plate 25 contacts the cold plate tray 10 through the cold plate boss 251.

[0104] In order to ensure that the capacity of the memory 24 can meet the system requirements, for the memory 24 adopting the horizontal double-layer stacking structure scheme, since the memory 24 is a double-layer structure, the second memory 242 cannot directly dissipate heat through the cold plate tray 10. Therefore, an indirect heat dissipation method is provided. The second memory 242 is first fixed to the memory cold plate 25, and the memory cold plate 25 is designed with a cold plate boss 251 that can contact the cold plate tray 10. The height of this cold plate boss 251 is the same as the height of the particles on the first memory 241, thereby solving the heat dissipation problem of the following second memory 242. The upper first memory 241 will directly dissipate heat through the cold plate tray 10, realizing the optimization of the node height.

[0105] The first memory 241 is in direct contact with the cold plate tray 10, while the second memory 242 contacts the cold plate tray 10 through the memory cold plate 25 and the cold plate boss 251. This multi-layer heat dissipation path design ensures that each memory 24 can effectively dissipate heat. Especially for the stacked memory 24, it can more evenly disperse heat. Through the multi-layer heat dissipation path, the working temperature of the memory 24 is controlled within a safe range, reducing the risk of failures caused by overheating and improving the overall reliability of the system. Through effective thermal management, the memory 24 can operate at a lower temperature, reducing the performance degradation and failure risk caused by heat. This design improves the performance of the memory 24 and the overall stability of the system.

[0106] The design of the stacked memory 24 and the memory cold plate 25 effectively utilizes the vertical space, enabling the system to increase the capacity of the memory 24 without increasing the occupied area. This layout optimization provides more flexibility for the arrangement of other components. By stacking the memory 24, the system can integrate more memory 24 capacity within a limited space. This high-density design is particularly important for high-performance computing applications that require a large memory 24 capacity.

[0107] Moreover, the hierarchical design of the memory cold plate 25 and the memory 24 makes maintenance and upgrade easier. Users can easily replace or upgrade the memory 24 without having to disassemble the entire system on a large scale.

[0108] All the heat-generating surfaces of the devices within the computing node 100 are in one plane, and the memory cold plate 25 is also parallel to the first memory 241 and other devices on the main board 20. Therefore, the integrated cold plate tray 10 can solve the heat dissipation problem of all heat sources within the computing node 100.

[0109] In some possible implementation manners, the stacked thickness of the memory 24 is 9.7 mm, and the thickness of the memory cold plate 25 is 4.4 mm.

[0110] In some possible implementation manners, a signal connector 26 is provided on a side of the main board 20 facing away from the functional device 30. The signal connector 26 is used to be electrically connected to the power backplane 203 of the server 200 when the computing node 100 is inserted into the server 200.

[0111] Through the layout of the signal connector 26, a cable-free design is achieved, which has good maintainability and is convenient for assembly. A series of influences caused by cable wiring are reduced.

[0112] The signal connector 26 is directly provided on the main board 20, enabling the computing node 100 to quickly dock with the power backplane 203 of the server 200 when inserted into the server 200. This design simplifies the signal connection process, reduces the installation time and complexity, and improves the system deployment efficiency. Through the design of the directly docking signal connector 26, the risk of misoperation during the connection process, such as connection errors or damage caused by excessive force, is reduced. Its automatic docking feature ensures the reliability and consistency of each connection. The signal connector 26 provides a stable electrical connection interface, ensuring reliable signal transmission. Through this design, the computing node 100 can obtain stable signal transmission, reducing the risk of failures caused by poor connections.

[0113] Setting the signal connector 26 on a side of the main board 20 facing away from the functional device 30 helps to optimize the use of the internal space. It avoids the entanglement of signal lines between the functional device 30 and other components, providing more flexibility for the layout of other components. Through this design, the computing node 100 can be conveniently inserted into or removed from the server 200 without complex signal line connections. This modular design facilitates system maintenance and upgrade, allowing for quick replacement or maintenance of the computing node 100, improving the system maintainability. At the same time, this design allows the computing node 100 to be quickly deployed and reconfigured in different server 200 environments. Users can easily adjust the configuration of the computing node 100 according to needs, improving the system flexibility and scalability.

[0114] In some possible implementation manners, the cold plate tray 10 is provided with a receiving groove 19, and the main board 20 is disposed in the receiving groove 19.

[0115] Specifically, the cold plate tray 10 includes a bottom plate 17 and side plates 18 connected to both sides of the bottom plate 17, and the side plates 18 on both sides and the bottom plate 17 form a receiving groove 19. The main plate 20 is disposed in the receiving groove 19 and is attached to the bottom plate 17. The cooling channel of the cold plate tray 10 is disposed in the bottom plate 17.

[0116] In some possible implementations, the cold plate tray 10 is provided with fixing studs 16 , and the main board 20 is connected to the cold plate tray 10 via the fixing studs 16 .

[0117] Specifically, the fixing stud 16 is disposed in the receiving groove 19 . More specifically, the fixing stud 16 is disposed on a side of the bottom plate 17 facing the main plate 20 .

[0118] The fixing studs 16 provide a strong mechanical connection, allowing the motherboard 20 to be firmly fixed to the cold plate tray 10. This design improves the structural stability of the entire computing node 100 and reduces the risk of loosening or damage of components due to vibration or impact. By fixing the motherboard 20 directly to the cold plate tray 10, heat can be more effectively transferred from the motherboard 20 to the cold plate tray 10 for heat dissipation. This design helps to improve the heat dissipation efficiency of the system and ensure stable operation of components under high load conditions.

[0119] The use of the fixing studs 16 can simplify the installation and removal process of the motherboard 20. The fixing studs 16 provide clear positioning and fixing points, making the installation process more intuitive and quick, while also simplifying subsequent maintenance and upgrades. The fixing studs 16 provide precise alignment points to ensure that the connection between the motherboard 20 and the cold plate tray 10 and other components is accurate. This precise alignment helps to optimize the reliability of electrical connections and signal transmission.

[0120] The fixing studs 16 are evenly distributed on the mainboard 20, which can effectively disperse mechanical stress and reduce deformation of the mainboard 20 caused by thermal expansion or other physical factors. This design helps to extend the service life of the mainboard 20 and other connected components. The use of the fixing studs 16 supports modular design, so that the mainboard 20 can be replaced or upgraded as an independent module. This modular design improves the flexibility and maintainability of the system and facilitates rapid response to technology updates or fault handling.

[0121] The computing node 100 provided in the embodiment of the present application includes a cold plate tray 10, a mainboard 20 and a functional device 30. The mainboard 20 is carried on the cold plate tray 10 to dissipate heat through the cold plate tray 10. The functional device 30 is fixed to the cold plate tray 10 and is electrically connected to the mainboard 20. The cold plate tray 10 is provided with a groove 111, and the groove 111 is located on the side of the cold plate tray 10 away from the mainboard 20. The groove 111 is used to cooperate with the partition 202 of the server 200.

[0122] Since the computing node 100 adopts the cold plate tray 10 design, the main board 20 is directly carried on the cold plate tray 10, enabling heat to be conducted and dissipated more efficiently through the cold plate tray 10. Compared with the traditional air-cooled system, the cold plate tray 10 provides a larger heat dissipation surface area, significantly improving the heat dissipation efficiency, thus ensuring the stable operation of the device under high-density deployment. Compared with the traditional laying of liquid cooling pipelines, the cold plate tray 10 integrates the functions of the tray and the cold plate, thus saving the thickness space of the computing node 100. The functional device 30 is fixed to the cold plate tray 10 and electrically connected to the main board 20. This design not only simplifies the layout of the components but also avoids overly close arrangements between the components, reducing the blockage problem of the heat dissipation channels. Through reasonable layout design, the computing node 100 can achieve higher integration without increasing the volume. The cold plate tray 10 is provided with a groove 111, which cooperates with the partition 202 of the server 200, providing flexible installation adaptability. At the same time, the design of the groove 111 offsets the influence of the thickness of the partition 202 between the computing nodes 100 on the height direction of the server 200. This design allows the computing node 100 to be quickly deployed in different types of server 200 environments, enhancing the compatibility and adaptability of the device. Therefore, the computing node 100 of the present invention not only improves the computing expansion ability and heat dissipation efficiency but also enhances the stability and applicability of the device, and can better meet the requirements of data-intensive applications for high-density data processing.

[0123] In addition, an embodiment of the present application further provides a server 200, including a chassis 201, a partition 202, and a plurality of computing nodes 100. The computing nodes 100 are inserted into the chassis 201 through the partition 202.

[0124] In view of the fact that the server 200 in this embodiment includes the liquid computing node 100 described in any of the above embodiments, therefore, the server 200 includes the structure and beneficial effects of the computing node 100, and will not be elaborated herein.

[0125] By using the partition 202 to insert a plurality of computing nodes 100 into the chassis 201, the server 200 can achieve a highly modular design. Each computing node 100 can be maintained independently without mutual influence. This design allows adding or removing computing nodes 100 as needed, thus easily expanding or reducing the computing power of the system. The computing nodes 100 are inserted into the chassis 201 through the partition 202, making the installation and disassembly process more convenient. The partition 202 provides additional physical support and isolation, reducing the risk of component loosening or damage caused by vibration or impact. At the same time, the partition 202 can help reduce electromagnetic interference and improve the reliability of signal transmission. Users can quickly replace or upgrade the computing node 100 without having to perform a large-scale disassembly of the entire server 200, reducing the maintenance difficulty and time.

[0126] Since the computing node 100 can be maintained independently and is pluggable as a whole with the cold plate tray 10, it can automatically cut off water, electricity, and signals without affecting the operation of other computing nodes 100, thus minimizing the reliability explosion radius.

[0127] The use of the partition 202 enables the computing nodes 100 to be arranged in the chassis 201 in a compact and orderly manner, maximizing the utilization of the internal space of the chassis 201. This layout optimization provides more flexibility for the arrangement of other components, enhancing the overall integration of the system. This design allows the server 200 to flexibly configure the computing nodes 100 according to different application requirements. Users can select appropriate node combinations according to specific computing tasks or load requirements, improving the adaptability and management efficiency of the system.

[0128] Through the design of the partition 202, effective air flow channels can also be formed between the computing nodes 100 to promote air flow and heat dissipation. This design helps to maintain the system temperature within a safe range, improving the heat dissipation efficiency and stability of the server 200.

[0129] In some possible implementation manners, the depth of the groove 111 is less than or equal to the thickness of the partition 202.

[0130] An appropriate depth of the groove 111 can ensure that the computing node 100 can be firmly embedded during installation without loosening or instability caused by an overly deep groove 111. This design helps to ensure the safety and stability of the components during transportation or operation. The depth of the groove 111 matches the thickness of the partition 202, ensuring the reasonable utilization of the internal space of the chassis 201, thus maximizing the available space inside the chassis 201 and providing more flexibility for the arrangement of other components.

[0131] By restricting the depth of the groove 111, the impact of possible errors and deviations in the manufacturing and assembly processes on the system performance can be reduced, improving the reliability and consistency of the system.

[0132] By limiting the depth of the groove 111 within the thickness of the partition 202, the overall structural strength of the partition 202 can be ensured not to be weakened. This design avoids the possible decrease in the strength of the partition 202 caused by an overly deep groove 111, thus improving the stability and durability of the entire system. The reasonable depth design of the groove 111 also helps to maintain good air flow and avoid hindering the air flow due to an overly deep groove 111. This design helps to improve the heat dissipation efficiency of the system and ensure that the components operate at an appropriate temperature.

[0133] In some possible implementation manners, the thickness of the partition 202 is 1 mm, and the depth of the groove 111 is less than or equal to 1 mm.

[0134] In some possible implementations, the cold plate tray 10 is provided with a blind plug 15, which is arranged on the side of the cold plate tray 10 away from the functional device 30. The blind plug 15 includes two blind plugs, one of which is connected to the liquid inlet channel of the cold plate tray 10, and the other blind plug 15 is connected to the liquid outlet channel of the cold plate tray 10.

[0135] The blind plug 15 design allows the cold plate tray 10 to be connected to the liquid cooling system without precise alignment during installation. This design simplifies the installation and removal process of the liquid cooling system and improves the convenience and efficiency of operation. The blind plug 15 provides an automatic docking function, reducing the risk of poor connection or leakage due to human operating errors. This design improves the reliability and safety of the liquid cooling system and ensures the stable operation of the system. Through the design of the liquid inlet and outlet flow channels, the cold plate tray 10 can efficiently circulate liquid, thereby effectively removing heat. This design helps to improve the heat dissipation efficiency of the system and ensure that the components maintain a suitable temperature under high load conditions.

[0136] At the same time, the design of the blind plug 15 reduces the need for external pipes and connectors, optimizes the space utilization inside the chassis 201, reduces the complexity of the liquid cooling system, and reduces the overall cost of the system. Through the design of the blind plug 15, the cold plate tray 10 can be quickly deployed and reconfigured in different system configurations, and the user can easily adjust the configuration of the liquid cooling system as needed, which improves the flexibility and scalability of the system. The use of the blind plug 15 supports modular design, so that the cold plate tray 10 can be replaced or upgraded as an independent module. This modular design improves the flexibility and maintainability of the system and facilitates rapid response to technology updates or fault handling.

[0137] Figure 15 for Figure 2 The schematic diagram of the structure of the computing node and the water distributor is shown.

[0138] like Figure 15 As shown, in some possible implementations, the server 200 further includes an inlet water splitter 204 and an outlet water splitter 205. One end of the inlet water splitter 204 is respectively connected to the blind plugs 15 that communicate with the inlet flow channel of the plurality of computing nodes 100, and the other end is provided with an inlet 206. One end of the outlet water splitter 205 is respectively connected to the blind plugs 15 that communicate with the outlet flow channel of the plurality of computing nodes 100, and the other end is provided with an outlet 207.

[0139] The use of the inlet water diverter 204 and the outlet water diverter 205 can effectively organize and manage the pipeline connections of the liquid cooling system. By centrally managing the inlet and outlet water flow channels, the complex pipeline layout is reduced, making the system cleaner and easier to maintain. The inlet water diverter 204 and the outlet water diverter 205 centrally manage the liquid flow, simplifying the installation and maintenance process of the liquid cooling system. Users can more easily conduct inspections, cleaning, and maintenance of the system, improving the operation efficiency. Through the design of the inlet water diverter 204 and the outlet water diverter 205, the crossing and entanglement of pipelines in the liquid cooling system can be reduced, optimizing the space utilization inside the chassis 201. This layout optimization provides more flexibility for the arrangement of other components. The inlet water diverter 204 and the outlet water diverter 205 can evenly distribute the coolant to each computing node 100, ensuring that each node can obtain sufficient coolant for heat dissipation. This uniform fluid distribution helps to improve the overall heat dissipation efficiency of the system and maintain the temperature stability of each node.

[0140] At the same time, through the centralized management of the inlet water diverter 204 and the outlet water diverter 205, the possible leakage points and the risk of poor connections in the liquid cooling system are reduced, improving the reliability and safety of the liquid cooling system and ensuring the stable operation of the system. The use of the inlet water diverter 204 and the outlet water diverter 205 also supports modular design, enabling the liquid cooling system to be easily expanded or reduced according to needs. Users can adjust the distribution of the coolant according to the heat dissipation requirements of the system, improving the flexibility and scalability of the system.

[0141] In some possible implementation manners, there are eight computing nodes 100, and four computing nodes 100 form a group. Along the height direction of the chassis 201, the four computing nodes 100 in each group are stacked.

[0142] By stacking the computing nodes 100 in the height direction of the chassis 201, the vertical space can be maximally utilized. This design allows more computing nodes 100 to be accommodated within the limited volume of the chassis 201, improving the computing density of the system. The stacking design can support a higher computing density and is suitable for application scenarios that require high-performance computing capabilities, such as AI servers, edge servers, data centers, cloud computing, and scientific computing, etc. This stacking design supports modular expansion, and users can increase or decrease the number of computing nodes 100 according to needs to adapt to different computing requirements. This flexibility enables the system to adapt to changing workloads and application requirements.

[0143] The layout of the stacked computing nodes 100 can simplify the routing of power and signal connections. Through reasonable layout design, the length and complexity of the cables are reduced, improving the cleanliness and maintenance convenience of the system. With a modular stacking design, each computing node 100 of the system can be independently maintained and replaced, reducing the impact on other nodes. This design improves the reliability and stability of the system, facilitating fault troubleshooting and maintenance. The vertical stacking design can utilize a combination of natural convection and forced air cooling to enhance the heat dissipation efficiency.

[0144] The water distribution system in the chassis 201 is also designed with four water distributors, namely two inlet water distributors 204 and two outlet water distributors 205. There are two inlets for the total inlet water and two outlets for the total outlet water, realizing eight inlets and eight outlets of the coolant on the corresponding computing nodes 100.

[0145] In some possible implementation manners, the chassis 201 is a 2U chassis. Each computing node 100 is provided with two processors 21 and 32 memories 24, thus realizing a layout of 8 computing nodes 100, 16 processors 21, and 256 memories 24 in a 2U space.

[0146] Among them, the height of each computing node 100 is only 0.5U (22 mm), increasing the number of configurations per unit volume of the computing node 100.

[0147] The above has introduced in detail a computing node and a server provided by the present application. Specific examples are used herein to elaborate on the principle and implementation manner of the present application. The description of the above embodiments is only used to help understand the method and its core idea of the present application. It should be noted that for those of ordinary skill in the art of this technology, without departing from the principle of the present application, several improvements and modifications can still be made to the present application, and these improvements and modifications also fall within the protection scope of the claims of the present application.

Claims

1. A computing node, characterized in that, It includes a cold plate tray, a main board and a functional device. The main board is carried on the cold plate tray to dissipate heat through the cold plate tray. The functional device is fixed to the cold plate tray and electrically connected to the main board. The cold plate tray is provided with a groove, and the groove is located on the side of the cold plate tray facing away from the main board. The groove is used to cooperate with the partition of the server.

2. The computing node according to claim 1, characterized in that The computing node further includes a shrapnel screw and a processor. The processor is arranged on the side of the main board facing the cold plate tray. The shrapnel screw passes through the cold plate tray and is connected to the processor so that the processor contacts the cold plate tray.

3. The computing node according to claim 2, wherein The shrapnel screw includes a shrapnel and a stud. The stud passes through the cold plate tray and is connected to the processor. The shrapnel is sleeved on the stud and is arranged on the side of the cold plate tray facing away from the processor.

4. The computing node according to claim 1, wherein The computing node further includes a processor. The processor is arranged on the side of the main board facing the cold plate tray. The cold plate tray is provided with a main cold plate, a main heat exchange channel, a processor cold plate and a processor channel. The main heat exchange channel is arranged on the main cold plate. The main board is carried on the main cold plate to dissipate heat through the main cold plate. The processor cold plate is arranged on the main cold plate and is opposite to the processor. The processor channel connects the processor cold plate.

5. The computing node according to claim 4, wherein The cold plate tray is provided with a blind plug. The blind plug is arranged on the side of the cold plate tray facing away from the functional device. The main heat exchange channel and the processor channel are respectively communicated with the blind plug.

6. The computing node according to claim 1, characterized in that, The functional device includes a functional component and a functional cold plate. The functional component is arranged on the functional cold plate. The functional cold plate is provided with a heat conduction block. The functional cold plate is fixed to the cold plate tray, and the heat conduction block contacts the cold plate tray.

7. The computing node according to claim 6, characterized in that, The functional device further includes a connector. The connector is arranged on the side of the functional device facing the main board. The functional component is electrically connected to the main board through the connector.

8. The computing node according to claim 7, wherein The heat conduction block is arranged on the side of the functional cold plate facing the main board and is arranged side by side with the connector.

9. The computing node according to claim 6, characterized in that The computing node further includes a hard disk backplane. The hard disk backplane is provided with a hard disk connector. The functional component includes a hard disk unit. The hard disk unit is electrically connected to the hard disk backplane through the connector. The hard disk backplane is electrically connected to the main board through the hard disk connector.

10. The computing node according to claim 6, wherein The functional component further includes an input / output module, a network interface module, and a security control module.

11. The computing node according to claim 1, wherein A power connector is arranged on the side of the main board facing away from the functional device. The power connector is used to be electrically connected to the power backplane of the server when the computing node is inserted into the server.

12. The computing node according to claim 1, wherein The computing node further includes a memory. The memory is arranged on the side of the main board facing the cold plate tray so that the memory contacts the cold plate tray.

13. The computing node according to claim 12, characterized in that, The computing node further includes a memory cold plate. The memory includes a first memory and a second memory. The first memory and the second memory are stacked on the motherboard. The first memory is in direct contact with the cold plate tray. The memory cold plate is disposed between the first memory and the second memory and is fixedly connected to the second memory. The memory cold plate is provided with a cold plate boss, and the memory cold plate contacts the cold plate tray through the cold plate boss.

14. The computing node according to claim 1, wherein A signal connector is provided on a side of the motherboard facing away from the functional device. The signal connector is used for electrically connecting to a power backplane of the server when the computing node is inserted into the server.

15. The computing node according to claim 1, wherein The cold plate tray is provided with a receiving groove, and the motherboard is disposed in the receiving groove.

16. The computing node according to claim 15, wherein, Fixing studs are provided in the receiving groove, and the motherboard is connected to the cold plate tray through the fixing studs.

17. A server, characterized in that, It includes a chassis, a partition, and a plurality of computing nodes as described in any one of claims 1-15. The computing nodes are inserted into the chassis through the partition.

18. The server according to claim 17, wherein The depth of the groove is less than or equal to the thickness of the partition.

19. The server according to claim 17, wherein The cold plate tray is provided with blind plugs. The blind plugs are disposed on a side of the cold plate tray facing away from the functional device. There are two blind plugs. One of the blind plugs communicates with the liquid inlet channel of the cold plate tray, and the other blind plug communicates with the liquid outlet channel of the cold plate tray. The server further includes a water inlet distributor and a water outlet distributor. One end of the water inlet distributor is respectively connected to the blind plugs of the plurality of computing nodes communicating with the liquid inlet channel, and the other end is provided with a water inlet. One end of the water outlet distributor is respectively connected to the blind plugs of the plurality of computing nodes communicating with the liquid outlet channel, and the other end is provided with a water outlet.

20. The server according to claim 17, characterized in that, There are eight computing nodes. Four computing nodes are in a group, and along the height direction of the chassis, the four computing nodes in each group are stacked.

Images