Computing equipment and complete machine cabinet

Abstract

The utility model discloses a computing device and a whole machine cabinet, and relates to the technical field of computers, a switching unit and an image processor in the computing device are respectively designed on independent board cards, and decoupling of a switching board card and an image processor board card is realized. Therefore, the technical problem that the switching unit and the image processor in the computing device are coupled in the same board card in the prior art, so that the board card needs to be redesigned when the image processor has difference is solved. When the image processor board card or the exchange board card is changed, the design of the exchange board card or the image processor board card does not need to be modified, so that multiplexing of hardware design is realized, the research and development cost is reduced, each component in the computing equipment can be independently maintained and troubleshot due to the decoupling design, and the reliability of the computing equipment is improved. The technical effect of improving the reliability and maintainability of the computing device is achieved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to computing equipment and a complete cabinet. Background Art

[0002] In the existing technical solutions, such as Figure 1 As shown in the figure, the computing node architecture includes three major components: the central processing unit (CPU), the switching unit (Switch unit) and the image processing unit (OAM GPU). The CPU motherboard is designed using the existing motherboard, and the Switch unit and the OAM GPU are integrated on a board to implement the entire system architecture design. However, the existing technology cannot realize the decoupling design of the OAM GPU and the Switch unit. Therefore, when the actual product is adapted, if the OAM GPU is replaced, it means that the board needs to be redesigned, so there is a high cost problem. Utility Model Content

[0003] The present application provides a computing device and a whole cabinet to at least solve the problem in the related art that a switching unit and an image processor in a computing device are coupled in the same board, which results in the need to redesign the board when there are differences in the image processors.

[0004] The present application provides a computing device, including: a switching board and an image processor board; wherein the switching board and the image processor board are independent boards, the switching board may be installed with a switching unit and a first connector, the image processor board may be installed with multiple image processors and a second connector, and the switching board and the image processor board are interconnected through the first connector and the second connector.

[0005] Optionally, the image processor board may also be installed with: a high-density connector, and the high-density connector is used to interconnect with other computing devices.

[0006] Optionally, the computing device also includes: a central processing unit board, the central processing unit board, the switch board, and the image processor board are independent boards, the central processing unit board can be installed with a central processing unit and a third connector, and the interconnection between the central processing unit board and the switch board is achieved through the third connector.

[0007] Optionally, the computing device may also be installed with: multiple network cards, with the computing device viewed from a top-down angle as a reference, the multiple network cards and the switch board are deployed adjacent to the first long side of the computing device, the first ends of the multiple network cards are connected to the switch board, and the second ends of the multiple network cards are connected to the network cables.

[0008] Optionally, the computing device may also be installed with: a data processing unit. Taking the computing device viewed from a top-down angle as a reference, the data processing unit and the central processing unit board are deployed adjacent to the second long side of the computing device, the first end of the data processing unit is connected to the central processing unit board, and the second end of the data processing unit is connected to a network cable.

[0009] Optionally, the computing device may also be installed with: a hard disk, which is deployed between the multiple network cards and the data processing unit and is used to store data.

[0010] Optionally, the ratio of the length of the switch board to the length of the central processing unit board is between 1.2 times and 2 times, or the ratio of the length of the switch board to the length of the image processor board is between 1.2 times and 2 times.

[0011] Optionally, the computing device further includes: a power board, which is independent of the switch board, the image processor board and the central processing unit board. Taking the computing device viewed from a top-down angle as a reference, the power board is deployed adjacent to the first long side of the computing device.

[0012] Optionally, with reference to the computing device in a top-down view, the switch board is disposed adjacent to the image processor board, and the image processor board is disposed adjacent to a first short side of the computing device.

[0013] The present application also provides a whole cabinet, comprising: multiple computing devices, wherein the multiple computing devices are the above-mentioned computing devices; at least one switching device, the at least one switching device is connected to the multiple computing devices to achieve interconnection between the multiple computing devices.

[0014] Through the present application, since the switching unit and the image processor in the computing device are designed on independent boards respectively, the decoupling of the switching board and the image processor board is achieved, thereby solving the technical problem in the related art that the switching unit and the image processor in the computing device are coupled in the same board, resulting in the need to redesign the board when there are differences in the image processors. By designing the switching unit and the image processor on independent switching boards respectively, when the image processor board or the switching board changes, the design of the switching board or the image processor board does not need to be modified, thereby realizing the reuse of hardware design and reducing R&D costs. In addition, the decoupling design allows each component in the computing device to be independently maintained and troubleshooted, achieving the technical effect of improving the reliability and maintainability of the computing device. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0016] Figure 1 is a schematic diagram of a computing node in the prior art;

[0017] Figure 2 A schematic diagram of a computing device provided in an embodiment of the present application Figure 1 ;

[0018] Figure 3 A schematic diagram of a computing device provided in an embodiment of the present application Figure 2 ;

[0019] Figure 4 A schematic diagram of a whole cabinet provided in an embodiment of the present application. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0021] It should be noted that, in the description of this application, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. The terms "first", "second", etc. in this application are used to distinguish similar objects, and are not used to describe a specific order or sequence.

[0022] In order to enable those skilled in the art to better understand the present application, the present application is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0023] An embodiment of the present application provides a computing device, such as Figure 2As shown, the computing device includes: a switch board 10 and an image processor board 20. The switch board 10 and the image processor board 20 are independent boards. The switch board 10 may be installed with a switch unit 100 and a first connector 101, and the image processor board 20 may be installed with a plurality of image processors 200 and a second connector 201. The first connector 101 and the second connector 201 are used to realize the interconnection between the switch board 10 and the image processor board 20.

[0024] Alternatively, if Figure 2 As shown, the computing device provided in the embodiment of the present application includes a switch board 10 and an image processor board 20. It should be noted that the switch board 10 and the image processor board 20 are independent boards. The independently designed switch board 10 and the image processor board 20 make the computing device more flexible when maintaining, upgrading or replacing components. For example, if the image processor board 20 needs to be upgraded to meet higher computing requirements, the image processor board 20 can be replaced alone without affecting the function of the switch board 10, and vice versa.

[0025] The switch board 10 is used for data exchange. A switch unit 100 and a first connector 101 may be installed in the switch board 10. A plurality of image processors 200 and a second connector 201 may be installed in the image processor board 20. Through the interconnection between the first connector 101 and the second connector 201, the computing device realizes high-speed data transmission between the switch board 10 and the image processor board 20.

[0026] It should be noted that the multiple image processors 200 may be OAM GPUs (Open Accelerator Module Graphics Processing Unit). OAM GPUs are graphics processor (GPU) modules designed based on the OAM (Open Accelerator Module) standard. OAM is an open accelerator module standard that aims to provide a standard, modular accelerator design for high-performance computing, artificial intelligence, and data center applications.

[0027] To summarize, since the switching unit and the image processor in the computing device are designed on independent boards respectively, the decoupling of the switching board and the image processor board is achieved, thereby solving the technical problem in the related art that the switching unit and the image processor in the computing device are coupled in the same board, resulting in the need to redesign the board when there are differences in the image processors. By designing the switching unit and the image processor on independent switching boards respectively, when the image processor board or the switching board changes, the design of the switching board or the image processor board does not need to be modified, thereby realizing the reuse of hardware design and reducing R&D costs. In addition, the decoupling design allows each component in the computing device to be independently maintained and troubleshooted, thereby achieving the technical effect of improving the reliability and maintainability of the computing device.

[0028] Optionally, in the computing device provided in the embodiment of the present application, the image processor board 20 may also be installed with: a high-density connector 202, and the high-density connector 202 is used to interconnect with other computing devices.

[0029] In an optional embodiment, if Figure 2 As shown, the image processor board 20 may also be installed with: a high-density connector 202. For example, the high-density connector 202 is deployed on the rear window of the image processor board 20 to facilitate the subsequent interconnection of the computing device with other computing devices, thereby realizing data interaction between multiple computing nodes.

[0030] By adopting high-density connectors, computing devices can better adapt to the high-density deployment needs of data centers while ensuring that the communication performance between nodes meets the stringent requirements of modern AI workloads.

[0031] Optionally, in the computing device provided in the embodiment of the present application, the computing device also includes: a central processing unit board 30, the central processing unit board 30, the switch board 10, and the image processor board 20 are independent boards, the central processing unit board 30 can be installed with a central processing unit 300 and a third connector 301, and the interconnection between the central processing unit board 30 and the switch board 10 is realized through the third connector 301.

[0032] In an optional embodiment, if Figure 2 As shown, the computing device further includes: a central processing unit board 30. It should be noted that the central processing unit board 30, the switching board 10, and the image processor board 20 are independent boards. The central processing unit board 30, the switching board 10, and the image processor board 20 maintain a highly decoupled design between the three components, making the computing device more flexible when maintaining, upgrading, or replacing components.

[0033] like Figure 2As shown, a central processor 300 and a third connector 301 may be installed in the central processor board 30. The central processor 300 may be a general-purpose processor responsible for executing the operating system, coordinating I / O operations, and managing other hardware resources in the device. The central processor 300 may also perform task scheduling to ensure that the GPU on the image processor board 20 is effectively utilized. The third connector 301 is used to establish a stable electrical connection with the switch board 10. That is, the interconnection between the central processor board 30 and the switch board 10 is achieved through the third connector 301. It should be noted that a specific connector may be installed in the switch board 10 to connect to the third connector 301.

[0034] In the computing device, the central processing unit board 30, the switch board 10 and the image processor board 20 are interconnected through respective connectors (the third connector 301, the first connector 101 and the second connector 201), forming a powerful and flexible computing platform.

[0035] Optionally, in the computing device provided in the embodiment of the present application, taking the computing device in a top-down view as a reference, the switch board 10 is deployed adjacent to the image processor board 20, and the image processor board 20 is deployed adjacent to the first short side of the computing device.

[0036] In an optional embodiment, taking the computing device in a top view as a reference, the top view of the board corresponding to the computing device is a rectangle, the length of the rectangle is divided into a first long side and a second long side, and the width of the rectangle is divided into a first short side and a second short side, such as Figure 2 The switch board 10 is disposed adjacent to the image processor board 20, and the image processor board 20 is disposed adjacent to the first short side of the computing device.

[0037] The switch board 10 and the image processor board 20 are deployed adjacent to each other, and the physical distance between them is minimized. This helps to reduce the delay of signal transmission and improve the efficiency of data exchange. In addition, the adjacent deployment is conducive to simplifying the internal wiring and reducing the crossing and complexity of cables.

[0038] Optionally, in the computing device provided in the embodiment of the present application, the computing device may also be installed with: multiple network cards 40, taking the computing device from a top-down angle as a reference, the multiple network cards 40 are deployed adjacent to the first long side of the computing device of the switch board 10, the first ends of the multiple network cards 40 are connected to the switch board 10, and the second ends of the multiple network cards 40 are connected to the network cable.

[0039] In an optional embodiment, the top view of the board corresponding to the computing device is a rectangle, the length of the rectangle is divided into a first long side and a second long side, and the width of the rectangle is divided into a first short side and a second short side. Figure 3As shown. The computing device may also be installed with multiple network cards 40, and the multiple network cards 40 may be HHHL network cards, HHHL network cards, Half Height Half Length network cards, which are a common type of network interface card in servers and high-performance computing devices. Figure 3 As shown, taking the computing device in a top view as a reference, multiple network cards 40 and the switch board 10 are deployed adjacent to the first long side of the computing device, that is, multiple network cards 40 and the switch board 10 are deployed on the same side. In addition, the first ends of the multiple network cards 40 are connected to the switch board 10, and the second ends of the multiple network cards 40 are connected to the network cable.

[0040] Since the uplink interfaces of the multiple network cards 40 are the switching units 100, the multiple network cards 40 are deployed adjacent to the switch board 10 on the same side. This layout design can effectively utilize the physical space of the computing device, avoid signal interference between network cards, and facilitate the wiring and management of network cables.

[0041] Optionally, in the computing device provided in the embodiment of the present application, the computing device may also be installed with: a data processing unit 50, taking the computing device from a top-down angle as a reference, the data processing unit 50 and the central processing unit board 30 are deployed adjacent to the second long side of the computing device, the first end of the data processing unit 50 is connected to the central processing unit board 30, and the second end of the data processing unit 50 is connected to the network cable.

[0042] In an optional embodiment, if Figure 3 As shown, the computing device may also be installed with a data processing unit 50, which is a DPU (Data Processing Unit, a hardware unit specifically used to accelerate network data processing, storage access and security tasks). In which, taking the computing device from a top view as a reference, the data processing unit 50 and the central processing unit board 30 are deployed adjacent to the second long side of the computing device, that is, the data processing unit 50 and the central processing unit board 30 are deployed on the same side, and the first end of the data processing unit 50 is connected to the central processing unit board 30, and the second end is connected to the network cable.

[0043] Since the uplink interface of the data processing unit 50 is the CPU, the data processing unit 50 and the central processing unit board 30 are deployed adjacent to each other on the same side. By deploying the data processing unit 50 on one side of the central processing unit board 30, the physical proximity of the DPU and the CPU can minimize signal transmission delay and improve data processing efficiency. Secondly, deployment on the side of the CPU board helps to optimize the layout and heat dissipation design inside the computing node, ensure that the system can still run stably under high load, reduce potential signal interference and attenuation, and ensure the integrity and reliability of data transmission.

[0044] Optionally, in the computing device provided in the embodiment of the present application, the computing device may also be installed with: a hard disk 60, which is deployed between the multiple network cards 40 and the data processing unit 50 for storing data.

[0045] In an optional embodiment, if Figure 3 As shown, the computing device may also be installed with a hard disk 60, and the hard disk 60 is used to store data of multiple graphics processors.

[0046] In the design of the front window of the whole cabinet in the prior art, the DPU card and the network card are placed on the left side of the chassis. During the wiring process of the front window of the whole cabinet, in order to avoid the wiring affecting the plugging and unplugging of the hard disk, the DPU card and the network card cables need to be placed on the left side of the cabinet for routing. However, a large number of wiring is placed on the left side of the cabinet, resulting in a large space at the front end of the cabinet, and the utilization rate of the right side is low, resulting in an increase in the depth of the whole cabinet, which is not conducive to product design. In the computing device provided in this application, if Figure 3 As shown in the figure, the uplink interface of the network card is the switching unit, so it is placed on the side of the switch board; the uplink interface of the DPU card is the CPU, so it is placed on the side of the motherboard; the hard disk that does not require routing is placed in the middle.

[0047] Through this layout optimization, the cables are dispersed on both sides of the cabinet, and the space on the left and right sides is used more evenly, which helps to reduce the depth of the entire cabinet and avoids heat dissipation problems and signal interference caused by cable concentration, helping to improve the stability of computing equipment and the reliability of data transmission.

[0048] Optionally, in the computing device provided in the embodiment of the present application, the computing device also includes: a power board 70, which is independent of the switch board 10, the image processor board 20 and the central processing unit board 30. Taking the computing device viewed from a top-down angle as a reference, the power board 70 is deployed adjacent to the first long side of the computing device.

[0049] In an optional embodiment, if Figure 3 As shown, the computing device further includes: a power board 70. It should be noted that the power board 70, the switching board 10, the image processor board 20 and the central processing unit board 30 are independent boards. Taking the computing device in a top view as a reference, the power board 70 is disposed adjacent to the first long side of the computing device. Figure 3 As shown, the power board 70 and the switch board 10 are on the same side, between the network card 40 and the switch board 10 .

[0050] Space utilization and layout are critical design considerations within computing devices. Deploying the power board 70 on one side of the switch board 10 helps to build a more compact device structure. This helps to control the overall size of the computing device and reduce the floor space requirements of the data center.

[0051] In an optional embodiment, a cooling component manifold and a fan may also be installed in the computing device. The cooling component may be deployed between the central processing unit board 30 and the image processor board 20 to quickly absorb and export the heat generated by the CPU and GPU, thereby reducing the accumulation of heat inside the device, helping to maintain the temperature control balance of the entire computing node device and reducing the thermal impact on other components. The fan may be deployed between multiple network cards 40 and the image processor board 20, and the fan ensures that the computing node device can maintain a good operating temperature under high load, thereby ensuring its stability and performance.

[0052] Optionally, in the computing device provided in the embodiment of the present application, the length of the switch board 10 and the length of the central processing unit board 30 satisfy a multiple relationship ranging from 1.2 times to 2 times, or the length of the switch board 10 and the length of the image processor board 20 satisfy a multiple relationship ranging from 1.2 times to 2 times.

[0053] In an optional embodiment, in the computing device provided in the embodiment of the present application, the ratio of the length of the switch board 10 to the length of the central processing unit board 30 is between 1.2 times and 2 times, or the ratio of the length of the switch board 10 to the length of the image processor board 20 is between 1.2 times and 2 times. For example, the length of the switch board 10 can be 220 mm, the length of the central processing unit board 30 is 330 mm, and the length of the image processor board 20 is 330 mm.

[0054] By setting the ratio between the length of the switch board 10 and the length of the central processing unit board 30, and the length of the switch board 10 and the length of the image processor board 20, it is beneficial to the efficient use of the internal space of the computing device, and improves the simplicity of wiring between the switch board 10, the image processor board 20 and the central processing unit board 30.

[0055] The embodiment of the present application provides a whole cabinet, such as Figure 4 As shown, the entire cabinet includes: a plurality of computing devices 80 and at least one switching device 90 .

[0056] A plurality of computing devices 80, wherein the plurality of computing devices 80 are any of the computing devices 80 described above;

[0057] At least one switching device 90 , wherein the at least one switching device 90 is connected to the plurality of computing devices 80 to achieve interconnection between the plurality of computing devices 80 .

[0058] Alternatively, if Figure 4As shown, an embodiment of the present application also provides a whole cabinet integrating AI computing and high-speed data exchange functions, and the whole cabinet includes: multiple computing devices 80 and at least one switching device 90. In the whole cabinet of the present application, multiple computing devices 80 are connected to at least one switching device 90. This interconnection design allows data to flow freely between various computing nodes, forming a highly interconnected computing network. Through the high-speed data exchange capability of at least one switching device 90, multiple computing devices 80 can quickly share model parameters, training data and other computing resources, which greatly enhances the overall computing efficiency and parallel processing capabilities of the system.

[0059] In summary, the whole cabinet of this application builds an infrastructure that can efficiently handle large-scale AI computing tasks through the layout and interconnection mechanism of computing nodes and switching nodes. In the application of actual data centers, it not only provides powerful computing power, but also ensures the high speed and stability of network communication.

[0060] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0061] For ease of description, spatially relative terms such as "on", "above", "on the upper surface of", "upper", etc. may be used herein to describe the spatial positional relationship between a device or feature and other devices or features as shown in the figures. It should be understood that the spatially relative terms are intended to include different orientations of the device in use or operation in addition to the orientation described in the figures.

[0062] For example, if a device in a drawing is inverted, a device described as "above" or "over" other devices or structures would then be oriented "below" or "beneath" the other devices or structures. Thus, the exemplary term "above" can encompass both "above" and "below." The device may be otherwise oriented (rotated 90 degrees or at other orientations) and the spatially relative descriptors used herein interpreted accordingly.

[0063] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method.

[0064] The above is a detailed introduction to a computing device and a whole cabinet provided by the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method and core idea of ​​the present application. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present application, several improvements and modifications can be made to the present application, and these improvements and modifications also fall within the scope of protection of the claims of the present application.

Claims

1. A computing device, characterized in that include: A switch board (10) and an image processor board (20); The switch board (10) and the image processor board (20) are independent boards; the switch board (10) may be installed with a switch unit (100) and a first connector (101); the image processor board (20) may be installed with a plurality of image processors (200) and a second connector (201); the switch board (10) and the image processor board (20) are interconnected via the first connector (101) and the second connector (201).

2. The computing device according to claim 1, wherein: The image processor board (20) may also be installed with: A high-density connector (202) is used to interconnect with other computing devices.

3. The computing device according to claim 1, wherein: The computing device further comprises: A central processing unit board (30), wherein the central processing unit board (30), the switching board (10), and the image processing unit board (20) are independent boards; the central processing unit board (30) may be installed with a central processing unit (300) and a third connector (301); and the central processing unit board (30) and the switching board (10) are interconnected via the third connector (301).

4. The computing device according to claim 3, characterized in that The computing device may also be installed with: A plurality of network cards (40), with the computing device in a top-down view as a reference, the plurality of network cards (40) and the switch board (10) are deployed adjacent to a first long side of the computing device, the first ends of the plurality of network cards (40) are connected to the switch board (10), and the second ends of the plurality of network cards (40) are connected to a network cable.

5. The computing device according to claim 4, characterized in that The computing device may also be installed with: A data processing unit (50), with the computing device in a top view as a reference, the data processing unit (50) and the central processing unit board (30) are arranged adjacent to the second long side of the computing device, a first end of the data processing unit (50) is connected to the central processing unit board (30), and a second end of the data processing unit (50) is connected to a network cable.

6. The computing device according to claim 5, characterized in that The computing device may also be installed with: A hard disk (60), the hard disk (60) being deployed between the plurality of network cards (40) and the data processing unit (50) and being used for storing data.

7. The computing device according to claim 3, characterized in that The ratio between the length of the switch board (10) and the length of the central processing unit board (30) is between 1.2 times and 2 times, or the ratio between the length of the switch board (10) and the length of the image processing unit board (20) is between 1.2 times and 2 times.

8. The computing device according to claim 3, characterized in that The computing device further comprises: A power board (70), wherein the power board (70) and the switch board (10), the image processor board (20), and the central processing unit board (30) are independent boards from each other, and taking the computing device in a top-down view as a reference, the power board (70) is disposed adjacent to a first long side of the computing device.

9. The computing device according to claim 1, wherein: Taking the computing device from a top-down perspective as a reference, the switch board (10) is deployed adjacent to the image processor board (20), and the image processor board (20) is deployed adjacent to a first short side of the computing device.

10. A complete cabinet, characterized in that: include: A plurality of computing devices (80), wherein the plurality of computing devices (80) are computing devices according to any one of claims 1 to 9; At least one switching device (90), the at least one switching device (90) being connected to the plurality of computing devices (80) to achieve interconnection between the plurality of computing devices (80).

Images