Server cabinet and server system

Abstract

The present application discloses a server cabinet and a server system, relating to the technical field of servers. The server cabinet includes: a computing connector, a first switching connector, and a second switching connector. The computing connector is divided into a first pin set and a second pin set; there is a first connection relationship between the first pin set and the first switching connector, and the first connection relationship matches the processor information of a first processor with a target bandwidth; there is a second connection relationship between the second pin set and the second switching connector, and the second connection relationship matches the bandwidth difference and the first connection relationship. Through the present application, the technical problem of low utilization rate of connection resources of a server cabinet in the related art can be solved, and the technical effect of improving the utilization rate of connection resources of the server cabinet can be achieved.

Description

Technical Field

[0001] This application relates to the technical field of servers, and particularly to server cabinets and server systems. Background Art

[0002] Server cabinets are usually used to provide a communication environment for server processors deployed in servers through switching processors, so that the server processors can implement server functions. Current server cabinets are usually designed according to the specifications of the server processors supported by the servers. As the specifications of server processors supported by server cabinets are no longer limited to one type, using a server cabinet with a fixed connection form to handle server processors of multiple specifications is likely to cause waste of the connection resources of the server cabinet. Summary of the Invention

[0003] This application provides a server cabinet and a server system to at least solve the problem of low utilization rate of connection resources of server cabinets in related technologies.

[0004] This application provides a server cabinet, which includes: a computing connector, a first switching connector, and a second switching connector. The computing connector is divided into a first pin set and a second pin set; there is a first connection relationship between the first pin set and the first switching connector, and the first connection relationship matches the processor information of a first processor with a target bandwidth, where the first pin set corresponds to the signal of the target bandwidth; there is a second connection relationship between the second pin set and the second switching connector, and the second connection relationship matches the bandwidth difference and the first connection relationship, where the bandwidth difference is the difference between the reference bandwidth of a second processor and the target bandwidth, and the second pin set corresponds to the signal of the bandwidth difference.

[0005] This application also provides a server system, which includes: a computing node, a switching node, and the aforementioned server cabinet. The computing connector included in the server cabinet is connected to the server processor in the computing node, and the switching connector included in the server cabinet is connected to the switching processor in the switching node.

[0006] Through this application, the computing connectors in the server cabinet are divided into a first set of pins corresponding to signals of the target bandwidth and a second set of pins corresponding to signals of the bandwidth difference. The first connection relationship between the first set of pins and the first switching connector matches the processor information of the first processor with the target bandwidth. The second connection relationship between the second set of pins and the second switching connector matches the bandwidth difference between the reference bandwidth and the target bandwidth of the second connector and the aforementioned first connection relationship. Therefore, when dealing with server processors of different bandwidths, the server cabinet can provide a connection method that matches the connected server processor, avoiding the waste of connection resources of the server cabinet caused by using a fixed connection method. Thus, the technical problem of low utilization rate of the connection resources of the server cabinet in the related art can be solved, and the technical effect of improving the utilization rate of the connection resources of the server cabinet can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] In order to more clearly illustrate the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0008] Figure 1 is a schematic diagram of the front of the cabinet of an AI supernode whole cabinet server according to an embodiment of the present application;

[0009] Figure 2 is a schematic diagram of the back of the cabinet of an AI supernode whole cabinet server according to an embodiment of the present application;

[0010] Figure 3 is a connection schematic diagram of a computing node and a switching node of an AI supernode whole cabinet server according to an embodiment of the present application Figure 1 ;

[0011] Figure 4 is a connection schematic diagram of a computing node and a switching node of an AI supernode whole cabinet server according to an embodiment of the present application Figure 2 ;

[0012] Figure 5 is a connection schematic diagram of a computing node and a switching node of an AI supernode whole cabinet server according to an embodiment of the present application Figure 3 ;

[0013] Figure 6 is a schematic diagram of a server cabinet according to an embodiment of the present application;

[0014] Figure 7 is a connection schematic diagram of a computing connector according to an embodiment of the present application;

[0015] Figure 8 is a layout diagram of a connector according to an embodiment of the present application;

[0016] Figure 9 is a schematic diagram of a server system according to an embodiment of the present application;

[0017] Figure 10 is a schematic diagram of the ratio between the bandwidth of a large - bandwidth GPU and the bandwidth of a small - bandwidth GPU according to an embodiment of the present application;

[0018] Figure 11 is a schematic diagram of the connection relationship between a computing node and a switching node Figure 1 ;

[0019] Figure 12 is a schematic diagram of the connection relationship between a computing node and a switching node Figure 2 . Detailed implementation manners

[0020] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with 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 shall fall within the protection scope of the present application.

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

[0022] With the continuous development of artificial intelligence technology, the training and inference of AI (Artificial Intelligence) models have attracted increasing attention. As the key infrastructure supporting the training and inference of AI models, the new generation of computing cluster architectures are facing severe performance challenges. The marginal benefit of the performance of a single server processor is decreasing, approaching the physical limit, and it is difficult to meet the continuous growing demand for computing power of large AI models. At the same time, the horizontal expansion strategy of computing clusters by simply increasing the number of servers has also encountered bottlenecks in efficiency and scalability, facing challenges in aspects such as cost, data synchronization, and energy consumption, which hinder the efficient execution of large-scale parallel computing. In this context, the AI supernode whole cabinet system has emerged. It systematically integrates multiple groups of GPU (Graphics Processing Unit) computing units and switching units through an innovative topology architecture, comprehensively optimizes in aspects such as multi-source computing power integration, high-speed advanced interconnection, heat dissipation, high-power density power supply, and whole cabinet management, and provides a scalable high-bandwidth domain (HBD, HighBandwidth Domain) supernode system with 32 cards or more to meet the needs of cutting-edge large model training and inference.

[0023] Figure 1 is a schematic diagram of the front of the cabinet of an AI supernode whole cabinet server according to an embodiment of the present application. As Figure 1 shown, the number of Us represents the number of layers of the cabinet of the whole cabinet server, and the height of each U position can be adjusted. Generally, but not limited to, the square hole strips of each node of the cabinet can be set to three specifications: OU (48mm) or RU(44.45mm) or SU(46.5mm). The components of the server can be distributed to each layer of the whole cabinet server, for example, a management switch is deployed at the 44OU and 42 OU positions, a cable management tray is deployed at the 43 OU position, Powershelf (power shelf, used to provide centralized power management for the devices in the server rack) nodes are deployed at the 37-38 OU positions and the 5-6OU positions, Rack Stiffener (rack stiffener, used to increase the structural strength and stability of the server rack) nodes are deployed at the 36 OU position and the 7 OU position, and CDU (Cooling Distribution Unit, a cooling distribution unit used to manage and distribute coolant or cold air) nodes are deployed at the 1-4 OU positions. Figure 1 In the distribution shown, the 39-41 OU positions are not considered for what components to be deployed, so they are marked as the NA (Not Applicable) layer. In actual applications, specific functional components can be considered to be deployed in the NA layer according to the functional requirements of the server. The computing nodes are placed on the upper and lower sides of the cabinet respectively (for example Figure 1For the 28 OU - 35 OU and 8 OU - 15 OU deployments shown in Figure 1 (compute nodes are deployed), the switching nodes are placed in the middle of the cabinet (for example, Figure 1 for the 16 OU - 27 OU deployment shown in

[0024] Figure 2 is a schematic diagram of the back of the cabinet of an AI supernode whole - cabinet server according to an embodiment of the present application. As Figure 2 shown, the back of the whole - cabinet server mainly includes three components: Busbar (bus bar), Manifold (manifold), and Cable Tray (cable tray). Among them, the Busbar is located in the middle (or on one side), and the Powershelf nodes are connected to the Busbar to supply power to the whole cabinet. The Cable Tray is located on the left and right sides of the Busbar, and the Cable Tray is used to connect the high - speed signals of the compute nodes and the switching nodes according to the designed topological interconnection relationship. The Manifold is located on the left and right sides at the back of the cabinet, providing water inlet and outlet, and providing a liquid path for liquid cooling of the whole cabinet. The joint action of these components can ensure the efficient and safe operation of the power supply and cooling system of the whole - cabinet server.

[0025] Figure 3 is a connection schematic diagram of the compute nodes and the switching nodes of an AI supernode whole - cabinet server according to an embodiment of the present application. Figure 1 As Figure 3As shown in the figure, #1 to #N are N computing nodes. It is possible but not limited to deploy 1 CPU (Central Processing Unit) on each computing node. The CPU can be connected to 4 GPUs through a PCIe (Peripheral Component Interconnect Express) switch chip. Taking the example of deploying 16 computing nodes in an AI full rack server, there are a total of 16×4 = 64 GPUs in this AI full rack server. If N computing nodes are deployed in the AI full rack server, there are a total of 4×N GPUs in this AI full rack server. The computing nodes can also be deployed with DPU (Data Processing Unit) network cards, storage devices, and BMC (Baseboard Management Controller) management modules. The storage device can be, but is not limited to, an SSD (Solid State Drive). Figure 3 The shown Switch Node 1 to Switch Node M are M switch nodes. It is possible but not limited to have one MAC (Media Access Control) chip on each switch node. The high-speed signals of the switch nodes and the computing nodes are interconnected through a Cable Tray. In the AI full rack server, the computing nodes and the switch nodes are all connected to the Rack Cable Tray (cable bridge inside or near the rack) through high-density connectors to achieve Serdes (Serializer / Deserializer) high-speed interconnection between all GPUs in the AI full rack server. Figure 4 is a schematic diagram of the connection between the computing nodes and the switch nodes of an AI supernode full rack server according to an embodiment of the present application Figure 2 The above-mentioned Serdes high-speed interconnection method between all GPUs is as Figure 4 shown. It should be noted that the solution of full interconnection of computing nodes through switch nodes in the AI full rack server needs to ensure that the high-speed signal resources of each GPU in each computing node are evenly distributed to each switch node. This is a Scale up topology. Figure 5 is a schematic diagram of the connection between the computing nodes and the switch nodes of an AI supernode full rack server according to an embodiment of the present application Figure 3 such as Figure 5As shown, the computing nodes 1-8 are interconnected through the switching nodes 1-8. The connector of each computing node inside the AI whole cabinet server is connected to 4 GPUs. The signals at the same positions of the 8 computing nodes fill up one switch (i.e., the switching node). For example, Figure 5 the computing node signals and the switching node signals with the same numbers shown in it have a connection relationship.

[0026] For the above-mentioned AI whole cabinet server, when there are multiple specifications (i.e., bandwidths) of GPUs that can be connected in the computing node, if a fixed connection method between the computing node and the switching node is used regardless of the specification of the GPU connected to the computing node, it will lead to waste of the connection resources of the server cabinet. To solve the above problems, this application proposes a server cabinet and a server system.

[0027] To enable those skilled in the art of this technology to better understand the solution of this application, the following further details this application in conjunction with the accompanying drawings and specific embodiments.

[0028] The embodiment of this application provides a server cabinet, Figure 6 which is a schematic diagram of a server cabinet according to the embodiment of this application. As shown in Figure 6 it, the server cabinet includes:

[0029] a computing connector, a first switching connector, and a second switching connector. The computing connector is divided into a first pin set and a second pin set;

[0030] There is a first connection relationship between the first pin set and the first switching connector, and the first connection relationship matches the processor information of the first processor with a target bandwidth. Among them, the first pin set corresponds to the signal of the target bandwidth;

[0031] There is a second connection relationship between the second pin set and the second switching connector, and the second connection relationship matches the bandwidth difference and the first connection relationship. Among them, the bandwidth difference is the difference between the reference bandwidth of the second processor and the target bandwidth, and the second pin set corresponds to the signal of the bandwidth difference.

[0032] Based on the above, since the computing connectors in the server cabinet are divided into a first set of pins corresponding to the signals of the target bandwidth and a second set of pins corresponding to the signals of the bandwidth difference, the first connection relationship between the first set of pins and the first switching connector matches the processor information of the first processor with the target bandwidth, and the second connection relationship between the second set of pins and the second switching connector matches the bandwidth difference between the reference bandwidth and the target bandwidth of the second connector and the aforementioned first connection relationship. Therefore, when dealing with server processors of different bandwidths, the server cabinet can provide a connection method that matches the connected server processor, avoiding the waste of the connection resources of the server cabinet caused by using a fixed connection method. Thus, the technical problem of low utilization rate of the connection resources of the server cabinet in the related art can be solved, and the technical effect of improving the utilization rate of the connection resources of the server cabinet can be achieved.

[0033] Optionally, in this embodiment, the computing connector can be, but is not limited to, a connector for connecting a server processor. The server processor can be, but is not limited to, a device with computing capabilities. The server processor can be, but is not limited to, a GPU or a CPU, etc. The first switching connector and the second switching connector can be, but is not limited to, connectors for connecting a switching processor. The switching processor can be, but is not limited to, a device with signal / data switching capabilities. The switching processor can be, but is not limited to, a switching chip. The connector can be, but is not limited to, a device that can provide a path for data / signal transmission. The connector can be, but is not limited to, a PCIe connector or a high-density connector, etc.

[0034] Optionally, in this embodiment, the bandwidth specification of the computing connector can be, but is not limited to, determined according to the maximum processor bandwidth of the server processors supported by the server cabinet. For example, when the bandwidth specifications of the server processors supported by the server cabinet include bandwidth A, bandwidth B, and bandwidth C, where A > B > C, a connector with a bandwidth of A (or a bandwidth greater than A) needs to be selected as the computing connector of the server cabinet. Another example is when the bandwidth specifications of the server processors supported by the server cabinet include bandwidth D and bandwidth E, where D > E, a connector with a bandwidth of D (or a bandwidth greater than D) needs to be selected as the computing connector of the server cabinet.

[0035] Optionally, in this embodiment, the second switching connector and the first switching connector can be, but is not limited to, having the same bandwidth specification.

[0036] Optionally, in this embodiment, the first set of pins can be, but is not limited to, used to connect the first processor with the target bandwidth, and the combination of the first set of pins and the second set of pins can be, but is not limited to, used to connect the second processor with the reference bandwidth. Figure 7It is a connection schematic diagram of a computing connector according to an embodiment of the present application. As Figure 7 shown, the connector of the computing node (i.e., the computing connector) is divided into a shared bandwidth part (i.e., the first pin set) corresponding to the small bandwidth (i.e., the target bandwidth) signal and an extra bandwidth part (i.e., the second pin set) corresponding to the extra bandwidth signal of the large bandwidth (i.e., the reference bandwidth) compared to the small bandwidth. It is possible but not limited to determine the small bandwidth as the bandwidth of the server processor with the smallest bandwidth among multiple server processors to be connected to the server cabinet. In this case, the extra bandwidth is the difference between the bandwidths of the other server processors except the server processor with the smallest bandwidth among the multiple server processors and the small bandwidth, and the current extra bandwidth can be determined according to the bandwidths of the server processors connected to the current server cabinet.

[0037] Optionally, in this embodiment, the first connection relationship, that is, the connection relationship between the first pin set and the first switching connector, may but is not limited to include the first quantity of the computing connector, the second quantity of the first switching connector, and the first connection mode between the first quantity of the first pin set and the second quantity of the first switching connector, and the first connection mode may but is not limited to be characterized by the first ratio. The second connection relationship, that is, the connection relationship between the second pin set and the second switching connector, may but is not limited to include the third quantity of the second switching connector and the second connection mode between the first quantity of the second pin set and the third quantity of the second switching connector, and the second connection mode may but is not limited to be characterized by the second ratio.

[0038] Optionally, in this embodiment, the first connection relationship / second connection relationship can be implemented by connection with a Cable Tray (a structure for supporting and managing cables), for example.

[0039] Optionally, in this embodiment, the processor information can be used to indicate, for example, the requirement situation of the first processor signal transmission. The processor information may but is not limited to include the quantity of the first processors and the bandwidth of the first processors, etc.

[0040] Optionally, in this embodiment, the matching of the first connection relationship with the processor information of the first processor with the target bandwidth includes, for example, that the first connection relationship is established according to the processor information. The matching of the first connection relationship with the processor information may but is not limited to mean that when connecting the first pin set and the first switching connector according to the first connection relationship, the waste of connector connection resources can be minimized while ensuring the full interconnection of the first processor.

[0041] Optionally, in this embodiment, the bandwidth difference is the difference between the reference bandwidth of the second processor and the target bandwidth of the first processor. The full-interconnection implementation solution of the server processors within the specified range of the server processors to be connected to the server cabinet can be designed by, but not limited to, regarding server processors with different bandwidths as the second processor. For example, if the bandwidth of server processor 1, which is the server processor with the minimum bandwidth that the server cabinet can support (or to be connected), is a, and the bandwidth of server processor 2, which is the server processor with the maximum bandwidth that the server cabinet can support (or to be connected), is b (a < b), server processor 1 can be regarded as the first processor, and server processors with bandwidths in the range of a to b (excluding a and including b) can be regarded as the second processor.

[0042] Optionally, in this embodiment, the matching of the second connection relationship with the bandwidth difference and the first connection relationship includes, but is not limited to, the second connection relationship being established based on the bandwidth difference and the first connection relationship. The matching of the second connection relationship with the bandwidth difference and the first connection relationship can, but is not limited to, mean that when connecting the first pin set and the first switching connector according to the first connection relationship and connecting the second pin set and the second switching connector according to the second connection relationship, the waste of connector connection resources can be minimized while ensuring the full interconnection of the second processor.

[0043] Optionally, in this embodiment, the second ratio included in the second connection relationship between the second pin set and the second switching connector is not limited to only being the first ratio included in the first connection relationship, but can be specifically set according to the bandwidth difference and the first connection relationship, making the topology structure within the server cabinet more flexible.

[0044] Optionally, in this embodiment, the server cabinet can, but is not limited to, be used to establish a full-interconnection architecture for multiple computing nodes included in the server system. For a full-interconnection architecture, one switching node must have a connection relationship with all computing nodes. The server system can include, but is not limited to, computing nodes and switching nodes. Server processors are provided within the computing nodes, and switching processors are provided within the switching nodes. Connectors are provided in the server cabinet corresponding to the switching nodes and the computing nodes. Through the connection of the connectors of the computing nodes and the connectors of the switching nodes, the signals of each computing node are relayed through the switching node to achieve full interconnection.

[0045] Optionally, in this embodiment, multiple computing connectors are deployed on multiple computing node chassis according to, but not limited to, heat dissipation requirements and power supply requirements, etc. The number of computing connectors deployed on each computing node chassis can, but is not limited to, be the same.

[0046] Optionally, in this embodiment, it includes, but is not limited to, deploying a plurality of first switching connectors on a plurality of switching node chassis according to heat dissipation requirements, power supply requirements, etc., and also deploying a plurality of second switching connectors on other multiple switching node chassis. The number of switching connectors deployed on each switching node chassis may, but is not limited to, be the same. To simplify the management of the switching node chassis, generally, a first switching connector and a second switching connector are not deployed on the same switching node chassis.

[0047] Optionally, in this embodiment, the switching nodes deployed on each switching node chassis may, but are not limited to, communicate with each other.

[0048] As an optional implementation manner, the first connection relationship includes: a first quantity of computing connectors, a second quantity of first switching connectors, and a first ratio. The first ratio is used to indicate the proportional relationship between the first pin set and the first switching connector and the proportional relationship between the first switching connector and the first pin set; the second quantity is calculated according to the first quantity, the target bandwidth, the switch bandwidth of the switching processor to which the first switching connector is to be connected, and the first ratio.

[0049] Optionally, in this embodiment, the first quantity of computing connectors is the number of all computing connectors in the server cabinet, which may, but is not limited to, be determined according to the number of first processors (or second processors).

[0050] Optionally, in this embodiment, the second quantity of first switching connectors is the number of all first switching connectors in the server cabinet.

[0051] Optionally, in this embodiment, the first ratio is used to indicate the proportional relationship between the first pin set and the first switching connector and the proportional relationship between the first switching connector and the first pin set. For example, when the first ratio is 3:3, it means that one first switching connector has a connection relationship with three first pin sets, and one first pin set has a connection relationship with three first switching connectors.

[0052] As an optional implementation manner, the server cabinet further includes a first switching node chassis, and the first switching connectors are deployed on the first switching node chassis; the number of first switching node chassis is the ratio of the second quantity to the first unit quantity, and the first unit quantity is the number of first switching connectors that each first switching node chassis allows to accommodate.

[0053] Optionally, in this embodiment, the first switching connectors may, but are not limited to, be evenly distributed on each first switching node chassis.

[0054] Optionally, in this embodiment, the first unit quantity may, but is not limited to, be determined from the perspective of heat dissipation requirements, etc.

[0055] Through the above content, by clarifying that the number of the first switching node chassis is the ratio of the second quantity (the number of the first switching connectors) to the first unit quantity (the number of connectors that each switching node chassis can accommodate), the rationalization of the number of the switching node chassis is ensured. This design avoids the waste of resources of the switching node chassis because the configuration of each switching node chassis is determined based on the number of connectors that it can effectively carry, thereby improving the overall resource utilization efficiency.

[0056] As an alternative implementation, the number of the computing connectors is the first quantity, where the first quantity is determined according to the number of the first processors. The server cabinet further includes computing node chassis, and the computing connectors are deployed on the computing node chassis; the number of the computing node chassis is the ratio of the first quantity to the second unit quantity, and the second unit quantity is the number of computing connectors that each computing node chassis is allowed to accommodate; all the computing node chassis are connected by each of the first switching node chassis under the first connection relationship.

[0057] Optionally, in this embodiment, the number of the computing connectors may but is not limited to be determined according to the number of the first processors. For example, the number of the computing connectors may be the number of the first processors.

[0058] Optionally, in this embodiment, the computing connectors may but are not limited to be evenly distributed on each of the computing node chassis.

[0059] Optionally, in this embodiment, the second unit quantity may be determined but is not limited to from perspectives such as heat dissipation requirements.

[0060] Through the above content, a full-interconnection network architecture in the case of connecting the first processors with the target bandwidth is achieved, ensuring efficient data exchange between the computing nodes.

[0061] As an alternative implementation, the second connection relationship includes: a second ratio and a third quantity of the second switching connectors, where the second ratio is used to indicate the proportional relationship between the second pin set and the second switching connectors and the proportional relationship between the second switching connectors and the second pin set; the second ratio is determined according to the first connection relationship, the target bandwidth, and the bandwidth difference; the third quantity is calculated according to the first quantity of the computing connectors and the second ratio.

[0062] Optionally, in this embodiment, the third quantity of the second switching connectors is the number of all the second switching connectors in the server cabinet.

[0063] Optionally, in this embodiment, the second ratio is used to indicate the ratio relationship between the second pin set and the second switching connector and the ratio relationship between the second switching connector and the second pin set. For example, when the second ratio is 2:1, it means that one second switching connector has a connection relationship with two second pin sets, and one second pin set has a connection relationship with one second switching connector.

[0064] As an optional implementation manner, the server cabinet further includes a second switching node chassis, and the second switching connectors are deployed in the second switching node chassis; the number of second switching node chassis is the ratio of the third number to the first unit number, and the first unit number is the number of first switching connectors that each first switching node chassis allows to accommodate.

[0065] Optionally, in this embodiment, the second switching connectors can be but are not limited to being evenly distributed on each second switching node chassis.

[0066] Optionally, in this embodiment, the first switching connectors and the second switching connectors can be but are not limited to being of the same specification attributes. Therefore, when calculating the number of second switching nodes, it is only necessary to calculate the ratio of the third number to the first unit number corresponding to the attributes of the first switching connectors.

[0067] Through the above content, by clarifying that the number of second switching node chassis is the ratio of the third number (the number of second switching connectors) to the first unit number (the number of connectors that each switching node chassis can accommodate), the rationalization of the number of switching node chassis is ensured, and signal transmission failures caused by too many connectors being accommodated in the switching node chassis are avoided.

[0068] As an optional implementation manner, the number of computing connectors is the first number, where the first number is determined according to the number of first processors. The server cabinet further includes computing node chassis, and the computing connectors are deployed in the computing node chassis; the number of computing node chassis is the ratio of the first number to the second unit number, and the second unit number is the number of computing connectors that each computing node chassis allows to accommodate; each second switching node chassis is connected to all the computing node chassis under the second connection relationship.

[0069] Optionally, in this embodiment, the number of computing connectors can be but is not limited to being determined according to the number of first processors. For example, the number of computing connectors can be the same as the number of first processors.

[0070] Optionally, in this embodiment, the computing connectors can be but are not limited to being evenly distributed on each computing node chassis.

[0071] Optionally, in this embodiment, the second unit number can be determined but is not limited to from aspects such as heat dissipation requirements.

[0072] Through the above content, a full-interconnection connection of signals corresponding to the bandwidth difference is achieved. Combining the connection between the first set of pins and the first switching connector with the first connection relationship, a full-interconnection network architecture in the case of connecting a second processor with a reference bandwidth is realized, ensuring efficient data exchange between computing nodes.

[0073] As an alternative embodiment, the server cabinet may but is not limited to further include: a host computer and a display device. The host computer is connected to the display device. Among them, the host computer is used for: obtaining processor information and a reference bandwidth; matching the first connection relationship between the first set of pins and the first switching connector according to the processor information; matching the second connection relationship between the second set of pins and the second switching connector according to the bandwidth difference and the first connection relationship; the display device is used for: displaying the connector layout diagram of the server cabinet, where the connector layout diagram shows the computing connector, the first switching connector, the second switching connector, the first connection relationship, and the second connection relationship that divide the first set of pins and the second set of pins.

[0074] Optionally, in this embodiment, the host computer may but is not limited to be a computer device at a higher level in the control system. It is responsible for receiving data information from lower-level machines (such as field devices or sensors), processing, analyzing, and making decisions, and sending control instructions to the lower-level machines. In the server cabinet, the host computer plays a core role in network management and resource allocation. The host computer may but is not limited to be a high-performance server, or a high-performance computer, or a cloud computing platform, or an embedded control unit integrated in the server cabinet. The present application does not make any limitations in this regard.

[0075] Optionally, in this embodiment, the display device may but is not limited to be an industrial-grade monitor or a touch-screen operation panel connected to the host computer. The display device may also but is not limited to be a mobile device such as a tablet computer or a smart phone. The display device may but is not limited to communicate with the host computer through a wireless network.

[0076] Optionally, in this embodiment, the first connection relationship may but is not limited to include the first quantity of the computing connector, the second quantity of the first switching connector, and the first ratio.

[0077] Optionally, in this embodiment, matching the first connection relationship between the first pin set and the first switching connector according to the processor information includes, but is not limited to, determining a first ratio based on the target bandwidth of the first processor and the switch bandwidth of the switching processor to which the first switching connector is to be connected; determining a second number of the first switching connectors according to the first ratio and the first number of computing connectors, where the first number of computing connectors is determined according to the number of first processors. More specifically, determining the first ratio based on the target bandwidth of the first processor and the switch bandwidth of the switching processor to which the first switching connector is to be connected includes, but is not limited to, matching a first ratio for an available bandwidth combination from bandwidth combinations and ratios with a corresponding relationship, where the bandwidth combination is a combination of the bandwidth of the server processor and the bandwidth of the switching processor, and the available bandwidth combination includes the target bandwidth and the switch bandwidth. The corresponding relationship between the bandwidth combination and the ratio can be, but is not limited to, summarized based on historical data.

[0078] Optionally, in this embodiment, the second connection relationship may include, but is not limited to, a third number of the second switching connectors and a second ratio.

[0079] Optionally, in this embodiment, matching the second connection relationship between the second pin set and the second switching connector according to the bandwidth difference and the first connection relationship includes, but is not limited to, calculating a proportional relationship between the target bandwidth and the bandwidth difference to obtain an intermediate ratio; calculating the second ratio according to the intermediate ratio and the first ratio; calculating the third number of the second switching connectors according to the intermediate ratio and the second number of the first switching connectors. For example, when the target bandwidth is 3x, the bandwidth difference is 1x, the first ratio is 2:1, and the second number is y, the second ratio can be determined to be 6:1, and the third number is y / 3.

[0080] Optionally, in this embodiment, after the host computer determines the first connection relationship and the second connection relationship, the host computer may generate a connector layout diagram and send the generated connector layout diagram to a display device, and the display device displays the connector layout diagram after receiving the connector layout diagram sent by the host computer.

[0081] Optionally, in this embodiment, Figure 8 is a connector layout diagram according to an embodiment of the present application. As Figure 8As shown, it can be but is not limited to that in the connector layout diagram, 96 computing connectors divided into a first pin set and a second pin set, 48 first switching connectors, 16 second switching connectors are shown, the first ratio included in the first connection relationship is 2:1, the second ratio included in the second connection relationship is 6:1. In the connector layout diagram, the computing connector part with the same number has a connection relationship with the first switching connector part or the second switching connector part with the same number. For example, the computing connector part numbered 0111 has a connection relationship with the first switching connector part numbered 0111, and the computing connector part numbered 0112 has a connection relationship with the second switching connector part numbered 0112.

[0082] Optionally, in this embodiment, the connector layout diagram can be but is not limited to being matched with the deployment information of the server cabinet. Among them, the deployment information of the server cabinet is used to indicate the positional relationship between the computing connectors and the first switching connectors (or a combination of the first switching connectors and the second switching connectors) in the server cabinet. The positional relationship can be but is not limited to including that the first switching connector (or a combination of the first switching connectors and the second switching connectors) is located in the middle of the computing connectors or the first switching connector (or a combination of the first switching connectors and the second switching connectors) is located on one side of the computing connectors. As Figure 8 shown is a connector layout diagram in the case where the deployment information is used to indicate that the first switching connector (or a combination of the first switching connectors and the second switching connectors) in the server cabinet is located in the middle of the computing connectors.

[0083] Optionally, in this embodiment, the display device can be but is not limited to indicating to the operator to connect the first pin set and the first switching connector according to the first connection relationship, and connect the second pin set and the second switching connector according to the second connection relationship through the display of the connector layout diagram.

[0084] Optionally, in this embodiment, it can be but is not limited to that the server cabinet further includes a hardware control module. The hardware control module can be but is not limited to being connected to the upper computer. Among them, the hardware control module can be but is not limited to being used to connect the first pin set and the first switching connector according to the first connection relationship, and connect the second pin set and the second switching connector according to the second connection relationship.

[0085] As an alternative implementation, the host computer can be used for, but not limited to: determining a first quantity of computing connectors according to the quantity of first processors; calculating a second quantity of first switching connectors according to the first quantity, the target bandwidth, the switch bandwidth of the switching processors to which the first switching connectors are to be connected, and a first ratio, where the processor information includes: the quantity of first processors and the target bandwidth, and the first connection relationship includes: the first quantity, the second quantity, and the first ratio, and the first ratio is used to indicate the proportional relationship between the first pin set and the first switching connectors and the proportional relationship between the first switching connectors and the first pin set.

[0086] Optionally, in this embodiment, the first quantity can be, but not limited to, equal to the quantity of first processors.

[0087] As an alternative implementation, the host computer can be used for, but not limited to: calculating an initial quantity of first switching connectors according to the first quantity, the target bandwidth, the switch bandwidth, and an initial ratio; obtaining the first ratio according to the adjustment information of the initial ratio; and adjusting the initial quantity according to the first ratio to obtain the second quantity.

[0088] Optionally, in this embodiment, the initial ratio can be, but not limited to, 1:1, that is, each first switching connector is connected to one computing connector (first pin set), and each computing connector (first pin set) is connected to one first switching connector.

[0089] Optionally, when the initial ratio is used, one first switching connector can be, but not limited to, connected to one switching processor. Therefore, the initial quantity of first switching connectors can be obtained by calculating the value obtained by dividing the product of the first quantity and the target bandwidth by the switch bandwidth.

[0090] Optionally, in this embodiment, when the initial quantity is determined, the ratio of the initial quantity to a first unit quantity can be calculated to determine the initial number of first switching node chassis, where the first unit quantity is the number of first switching connectors that each first switching node chassis allows to accommodate; comparing the initial number of first switching node chassis with a first switching node chassis threshold; when the initial number of first switching node chassis is greater than the first switching node chassis threshold, determining that the adjustment information is to increase the initial ratio; when the initial number of first switching node chassis is less than or equal to the first switching node chassis threshold, determining that the adjustment information indicates that the initial ratio is not adjusted. In this way, the situation where the number of required first switching node chassis is too large to be accommodated in the server cabinet is avoided.

[0091] Optionally, in this embodiment, expanding the initial ratio means expanding the number of computing connectors (the first pin set) connected to a first switching connector. For example, expanding the connection of one computing connector (the first pin set) to one first switching connector in the initial ratio to the connection of two computing connectors (the first pin set) to one first switching connector.

[0092] Optionally, in this embodiment, when expanding the initial ratio, it is also necessary to consider the number of first switching connectors / second switching connectors that a switching node chassis can accommodate and ensure the establishment of a full-interconnection architecture.

[0093] Optionally, in this embodiment, the adjustment information can be but is not limited to being determined according to a ratio adjustment model. The ratio adjustment model is a more refined model that takes into account more factors, such as the actual data transmission requirements between computing nodes and switching nodes, system performance bottlenecks, energy consumption efficiency, etc. The system learns from historical data and predicts the optimal first ratio based on specific information of the first processor, such as processor utilization rate, task type, etc., to improve signal transmission efficiency and resource utilization.

[0094] Through the above content, by introducing the adjustment of the initial ratio, more refined resource management is achieved. Different from the fixed ratio design, this method can dynamically adjust the first ratio and the number of first switching connectors according to the number of first processors on the current computing node and the target bandwidth, ensuring a more precise matching of computing resources and network resources, avoiding resource waste, and at the same time improving the efficiency of resource utilization.

[0095] As an optional implementation manner, the host computer can be but is not limited to also being used for: when the initial ratio is used to indicate that each computing connector is connected to one first switching connector and each first switching connector is connected to one computing connector, calculating the product of the first quantity and the target bandwidth to obtain a first value; calculating the ratio of the first value to the switch bandwidth to obtain an initial quantity.

[0096] Optionally, in this embodiment, when using the initial ratio, if one first switching connector is connected to one switching processor, then the bandwidth of one first switching connector is the switch bandwidth.

[0097] As an optional implementation manner, the host computer can be but is not limited to also being used for: when the adjustment information is used to indicate that the initial ratio is adjusted and the adjustment method of the initial ratio, adjusting the initial ratio to the first ratio according to the adjustment method; when the adjustment information is used to indicate that the initial ratio is not adjusted, determining the initial ratio as the first ratio.

[0098] Optionally, in this embodiment, it is possible but not limited to obtaining an adjustment method for the initial ratio when it is detected that the initial number of the first switching connectors is greater than or equal to the first switching connector number threshold, where the first switching connector number threshold can be used but not limited to indicating the maximum number of the first switching connectors that the server cabinet can accommodate. It is possible but not limited to determining the first switching connector number threshold as the product of the first unit quantity and the first switching node number threshold, where the first switching node number threshold can be used but not limited to indicating the maximum number of the first switching node chassis that the server cabinet can accommodate. The server cabinet may also include an input device, which is connected to the upper computer. The upper computer can obtain the adjustment method for the initial ratio in the following ways: controlling the display device of the server cabinet to display a ratio adjustment prompt, where the ratio adjustment prompt is used to instruct the operator to input the adjustment method for the initial ratio; receiving the adjustment method for the initial ratio input by the operator through the input device.

[0099] Based on the above content, combined with the actual situation, choose whether to adjust the initial ratio. When there is no need to adjust the initial ratio, select the initial ratio as the first ratio, ensuring that the fully interconnected architecture always holds when connecting the first processor while guaranteeing a flexible topological structure.

[0100] As an optional implementation manner, the upper computer can also be used but not limited to: determining a second ratio according to the first connection relationship, the target bandwidth, and the bandwidth difference, where the second ratio is used to indicate the proportional relationship between the second pin set and the second switching connector and the proportional relationship between the second switching connector and the second pin set; calculating a third number of the second switching connectors according to the first number of the computing connectors and the second ratio, where the second connection relationship includes: the second ratio and the third number.

[0101] Based on the above content, by determining the second ratio according to the bandwidth difference and the first connection relationship, it can be ensured that when the server cabinet faces processors with different bandwidths, its network connection and data transmission can be flexibly adapted. This dynamic adjustment ability enables the server system to effectively utilize the existing network resources when processing GPUs with small bandwidths, and when processing GPUs with large bandwidths, through the reasonable allocation of the second switching connectors, it ensures that the additional bandwidth requirements are met, thereby improving the adaptability of the server cabinet to variable computing requirements.

[0102] As an optional implementation manner, the upper computer can also be used but not limited to: determining the connector bandwidth of the first switching connector according to the first connection relationship and the target bandwidth; determining the second ratio according to the connector bandwidth and the bandwidth difference.

[0103] Optionally, in this embodiment, when the first ratio is not the initial ratio, due to the requirement of the connection bandwidth, the number of switching processors connected by a first switching connector may change and may no longer be one. Then, the connector bandwidth of the first switching connector may not be the switch bandwidth of the switching processor. Therefore, it is necessary to determine the connector bandwidth of the first switching connector according to the first connection relationship and the target bandwidth.

[0104] Optionally, in this embodiment, the second ratio is determined according to the connector bandwidth and the bandwidth difference, that is, it is determined how many signals with the bandwidth difference can be accommodated by the connector bandwidth of a second switching connector.

[0105] Optionally, in this embodiment, the second ratio can be determined according to the connector bandwidth and the bandwidth difference in the following ways but is not limited thereto: calculate the ratio of the connector bandwidth of the first switching connector to the bandwidth difference to obtain the second ratio.

[0106] As an optional implementation manner, the host computer can be used for but is not limited to: extracting the set number of the set of the first pin sets connected by each first switching connector from the first connection relationship; determining the product of the set number and the target bandwidth as the connector bandwidth.

[0107] Optionally, in this embodiment, extracting the set number of the set of the first pin sets connected by each first switching connector from the first connection relationship includes but is not limited to extracting the proportional relationship between the first switching connector and the first pin set from the first ratio.

[0108] Through the above content, determining the connector bandwidth according to the first connection relationship ensures the accuracy of the connector bandwidth value and, to a certain extent, ensures the accurate establishment of the second connection relationship.

[0109] As an optional implementation manner, the server cabinet further includes a computing node chassis and a first switching node chassis. The computing connector is deployed on the computing node chassis, and the first switching connector is deployed on the first switching node chassis. The computing node chassis is used to install computing nodes, and the first switching node chassis is used to install first switching nodes. The host computer is further used for: calculating the ratio of the second number of the first switching connectors to the first unit number to obtain the number of the first switching node chassis, where the first unit number is the number of first switching connectors that each first switching node chassis allows to accommodate, and the ratio of the first number to the second unit number is the number of the computing node chassis, and the second unit number is the number of computing connectors that each computing node chassis allows to accommodate.

[0110] Optionally, in this embodiment, the computing connectors in the server cabinet may be, but are not limited to, evenly deployed on the computing node chassis of the server cabinet, and the first switching connectors in the server cabinet may be, but are not limited to, evenly deployed on the first switching node chassis of the server cabinet.

[0111] As an optional implementation manner, the display device is further configured to: divide the computing connectors on the connector layout diagram according to the number of computing node chassis, and divide the first switching connectors according to the number of the first switching node chassis, where all the computing node chassis are connected under the first connection relationship by each of the first switching node chassis.

[0112] Through the above visualization division method, the node network architecture design in the server cabinet becomes more intuitive. When operating to plan or adjust the node network layout, the connection relationship between the computing nodes and the first switching nodes can be visually seen, avoiding complex network calculations and simulations, and simplifying the design process.

[0113] As an optional implementation manner, the server cabinet further includes a computing node chassis and a second switching node chassis. The computing connectors are deployed on the computing node chassis, and the second switching connectors are deployed on the second switching node chassis. The computing node chassis is used to install computing nodes, and the second switching node chassis is used to install second switching nodes. The host computer is further configured to: calculate the ratio of the third quantity of the second switching connectors to the first unit quantity to obtain the number of the second switching node chassis.

[0114] Optionally, in this embodiment, the second switching connectors in the server cabinet may be, but are not limited to, evenly deployed on the second switching node chassis of the server cabinet.

[0115] Optionally, in this embodiment, the deployment scheme of the second switching connectors on the second switching node chassis in the server cabinet may be, but is not limited to, the same as the deployment scheme of the first switching connectors on the first switching node chassis in the server cabinet, that is, the number of the first switching connectors / second switching connectors deployed on the first switching node chassis and the second switching chassis in the server cabinet may be, but is not limited to, equal, and the interconnection signal resources of the first switching connectors deployed on the first switching node chassis in the server cabinet may be, but is not limited to, equal to the interconnection signal resources of the second switching connectors deployed on the second switching node chassis in the server cabinet.

[0116] As an optional implementation manner, the display device is further configured to: divide the second switching connectors on the connector layout diagram according to the number of the second switching node chassis, where all the computing node chassis are connected under the second connection relationship by each of the second switching node chassis.

[0117] Optionally, in this embodiment, a full-interconnection architecture of multiple computing nodes may be implemented through the relay of the first switching node and / or the second switching node, but is not limited thereto. The full-interconnection structure means that a switching node must be connected to all computing nodes. Therefore, but not limited thereto, in the connector layout diagram, the first pin set may be connected to the first switching connector according to the first connection relationship, and it may be predicted whether the full-interconnection architecture of multiple computing nodes is established after the second pin set and the second switching connector are connected according to the preliminary connection method of the second connection relationship. If it is not established, the preliminary connection method between the second pin set and the second switching connector may be adjusted according to the second connection relationship to obtain the final connection method; in the connector layout diagram, the second pin set and the second switching connector are connected according to the final connection method.

[0118] An embodiment of the present application further provides a server system. Figure 9 It is a schematic diagram of a server system according to an embodiment of the present application. As Figure 9 shown, the server system includes: computing nodes, switching nodes, and the aforementioned server cabinet. The computing connectors included in the server cabinet are connected to the server processors in the computing nodes, and the switching connectors included in the server cabinet are connected to the switching processors in the switching nodes.

[0119] Optionally, in this embodiment, the server processor may but is not limited to include a first processor and a second processor, and the server processors may but are not limited to be evenly deployed on multiple computing nodes.

[0120] Optionally, in this embodiment, the switching connector may but is not limited to include a first switching connector and a second switching connector.

[0121] Optionally, in this embodiment, the switching node may but is not limited to include a first switching node and a second switching node.

[0122] Optionally, in this embodiment, the switching processors may but are not limited to be evenly deployed on one or more switching nodes.

[0123] Optionally, in this embodiment, the server processor may but is not limited to be an electronic circuit or device capable of executing instructions. For example, it may be a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), an ASIC (Application-Specific Integrated Circuit), or an FPGA (Field-Programmable Gate Array).

[0124] Optionally, in this embodiment, the switching processor may be, but is not limited to, a device responsible for packet forwarding, processing, and management. For example, it may be a network switching chip or a routing processor.

[0125] Optionally, in this embodiment, each switching node (including the first switching node and the second switching node) may be, but is not limited to, interconnected. Coupled with the foregoing solution, all the first switching nodes are connected to all the computing nodes under the first connection relationship, and all the second switching nodes are also connected to all the computing nodes under the second connection relationship. This enables all the computing nodes to be fully connected throughout the server cabinet, so that in the case of partial switching node failures, each computing node can still communicate through other switching nodes.

[0126] As an alternative embodiment, the present application also provides a method for non-fixed ratio of the Cable Tray of a whole cabinet server, which solves the problem in the related art that when designing a system where different GPUs (i.e., server processors) share a set of computing nodes and a Cable Tray system, due to the large difference in bandwidth between different GPUs, it may affect the design of the system connection topology.

[0127] If a solution that evaluates and designs the Cable Tray system according to the GPU with the maximum bandwidth (i.e., the second processor) is adopted for different GPUs with different designed bandwidths, then when installing a GPU with a small bandwidth later, it is necessary to reduce the connectors (i.e., computing connectors) and cables of the computing nodes (i.e., computing node chassis) on the basis of the existing design, and remove the corresponding switching nodes; or only reduce the cables and remove the corresponding switching nodes. The disadvantage of this design solution is that when the connection relationship ratio (i.e., the first ratio and the second ratio) of the connectors of the computing nodes and the switching nodes (i.e., the first switching node chassis and the second switching node chassis) (i.e., the first connector and the second connector) exceeds 1:1, the ratio on each connector is fixed, and the number of corresponding connectors changes with the ratio. For example, when changing from a 1:1 ratio to a k:k ratio (k > 1), the number of connectors of the corresponding computing nodes will also change from the original M to M / k; the number of connectors of the switching nodes will also change from the original N to N / k. If M / k or N / k is not an integer, it will affect the existing design.

[0128] The method adopted in this embodiment, where the ratio of the connection relationship between the connectors of the computing nodes and the connectors of the switching nodes is different when installing a GPU with a small bandwidth (i.e., the first processor) and a GPU with a large bandwidth (i.e., the second processor), this method can be called the Cable Tray non-fixed ratio method, which is much more flexible in design compared to the method of adopting the same ratio.

[0129] Optionally, in this embodiment, it includes, but is not limited to, first designing the computing connectors within the computing nodes. Within the same computing connector, all signals of the low-bandwidth GPUs and high-bandwidth GPUs are included, where the common signals of the low-bandwidth GPUs and high-bandwidth GPUs (i.e., the signals of the target bandwidth) occupy some of the PINs of the computing connector, and the additional signals of the high-bandwidth GPUs (i.e., the signals of the bandwidth difference) occupy the other PINs of the computing connector.

[0130] Optionally, in this embodiment, it includes, but is not limited to, after determining the design scheme of the computing connector, first designing the number of GPUs with the minimum bandwidth (i.e., the number of the first processors), the bandwidth (i.e., the target bandwidth), etc., and determining the connector situation of the corresponding switching nodes, etc. At this time, the ratio design of the connectors of the computing nodes and the connectors of the switching nodes is only formed according to the actual situation of the low-bandwidth GPUs, and the situation of the high-bandwidth GPUs is not considered at this time.

[0131] Optionally, in this embodiment, let the bandwidth of the low-bandwidth GPU (i.e., the target bandwidth) be A, the number of GPUs within a single computing node be N, and the number of computing nodes be P; let the bandwidth of the switching chip (i.e., the bandwidth of the switch) be B, the number of switching chips within a single switching node be M, and the number of switching nodes be Q. In the case where the ratio design of the connectors of the computing nodes and the connectors of the switching nodes is 1:1 (i.e., the initial ratio), the total bandwidth of the computing nodes = A × N × P = the total bandwidth of the switching nodes = B × M × Q. Since the bandwidth of a single small GPU is fixed, the number of GPUs within a single computing node and the number of computing nodes are also fixed according to the original requirements, the bandwidth of the switching chip is also fixed, and the number of switching chips within a single switching node needs to be fixed according to the actual heat dissipation, power supply, etc. Therefore, it can also be basically determined, and thus the number of switching nodes can also be determined accordingly.

[0132] Optionally, in this embodiment, Figure 10 It is a schematic diagram of the ratio between the bandwidth of the high-bandwidth GPU and the bandwidth of the low-bandwidth GPU according to the embodiment of the present application. The method given in the present application can be applied, but is not limited to, in the GPU scenarios as Figure 10 shown.

[0133] It can be, but is not limited to, first designing the connection relationship between the computing connector and the first switching connector according to the initial ratio. In the initial design, one connector of a single computing node is only interconnected with one connector of a single switching node (which can also be called a switch), and one connector of a single switching node will also only be interconnected with one connector of a single computing node. Therefore, in the original design, the connection relationship between the connectors of the computing nodes and the connectors of the switching nodes is a 1:1 ratio.

[0134] In actual use, if a 1:1 ratio is adopted, then the number of connectors of the computing node / the number of connectors of the switching node = Q / P. Under the 1:1 ratio, the number of connectors of the computing node or the switching node may be too large, and such a large number of connectors cannot be accommodated in a single computing node or switching node. Therefore, the ratio will be adjusted in actual use to reduce the number of connectors.

[0135] Figure 11 It is a schematic diagram of the connection relationship between a computing node and a switching node according to an embodiment of the present application Figure 1 . As Figure 11 shown, when designing the connection method according to the small-bandwidth GPU, there are a total of 16 computing nodes (i.e., computing node 1 - computing node 16), and the number of GPUs (i.e., the total amount of the first processor / the second processor / the server processor) is fixed. Then, according to the switch chip bandwidth (i.e., the switch bandwidth of the switching processor), after evaluating the comprehensive design, the number of switch chips (i.e., switching processors) and the number of switches (i.e., switching nodes) are determined. After that, the number of basic connectors (i.e., computing connectors and the first switching connectors) of the computing node and the switch can be determined (at this time, the ratio is 1:1). Then, according to the actual situation, the ratio is adjusted, and finally the final number of connectors of the computing node and the switch is determined. As Figure 11 shown, when the computing node uses a small-bandwidth GPU (i.e., the first processor), the ratio relationship between the connectors of the computing node (i.e., computing connectors) and the connectors of the switching node (i.e., the first switching connectors) is 2:1 (i.e., the first ratio), that is, one connector of each computing node corresponds to one connector of one switching node, and one connector of each switching node corresponds to one connector of each of two computing nodes. Figure 11 As shown, each computing node includes 6 computing connectors, and there are a total of 16 computing nodes. After applying the method given in the present application, 6 switching nodes are required to achieve full interconnection of the computing nodes, and each switching node includes 8 first switching connectors. The connectors of each computing node are divided into a first pin set (i.e., the part numbered XXX1) and a second pin set (i.e., the part numbered XXX2). The first pin set corresponds to the signal of the small-bandwidth GPU bandwidth, and the second pin set corresponds to the signal of the additional bandwidth of the large-bandwidth GPU compared with the small-bandwidth GPU. Figure 11 There is a connection relationship between the computing connector parts with the same number in

[0136] and the first switching connector parts with the same number. For example, there is a connection relationship between the computing connector part in the computing node numbered 0111 and the first switching connector part in the switching node numbered 0111. After determining the connection relationship of the small-bandwidth GPU part (i.e., the first connection relationship), the second connection relationship is determined according to the first connection relationship, the small GPU bandwidth, and the additional bandwidth (the bandwidth difference between the large GPU bandwidth and the small GPU bandwidth). In asFigure 11 When the bandwidth of the small-bandwidth GPU shown is 3 / 4 of that of the large-bandwidth GPU, the connectors of the computing nodes should be divided into 4 parts. Among them, the shared bandwidth of the small-bandwidth GPU and the large-bandwidth GPU occupies 3 / 4 of the connectors, and the additional bandwidth of the large-bandwidth GPU occupies the other 1 / 4. The ratio of the connectors of the small-bandwidth GPU to the switch is 2:1 (i.e., the first ratio), that is, the bandwidth of the connectors of every two computing nodes × 3 / 4 of the bandwidth of the connectors of the computing nodes = the bandwidth of the connectors of each switching node. Plus the small GPU bandwidth: additional bandwidth = 3:1. Therefore, it can be learned that the ratio of the additional part of the large-bandwidth GPU to the switch is 6:1 (i.e., the second ratio), that is, the connectors of every 6 computing nodes (the additional bandwidth part of the large-bandwidth GPU) are connected to 1 connector of the switch.

[0137] Then calculate the additional bandwidth when all 16 computing nodes are installed with large-bandwidth GPUs, and obtain the additional number of switches required. 16 computing nodes × 6 connectors per computing node / the ratio of the connectors of the additional bandwidth part of the computing node to the switch, where every 6 connectors of the computing nodes are connected to 1 connector of the switch = the number of connectors of the switches required = 16 switch connectors (i.e., the third quantity of the second switching connector). Divide by the number of connectors per switch (i.e., the first unit quantity) to get that 2 additional switches are needed, and the ratio is 6:1. Therefore, for the additional part of the large-bandwidth GPU, 2 additional switches need to be added in the cabinet for support.

[0138] Since the connectors of each computing node are exactly 6, one additional switch node can be connected to the additional part of the 6 connectors of each computing node. However, in this case, each additional switch corresponds to 8 computing nodes alone and cannot be connected to all 16 computing nodes. Such an architecture is not a fully interconnected architecture.

[0139] This involves considering the specific connection method from the perspective of full interconnection after determining the second ratio and the third quantity of the second switching connector. Since the additional bandwidth of the large-bandwidth GPU of the computing node is often less than the shared bandwidth of the large and small GPUs. Therefore, when dealing with the additional bandwidth part, the ratio of the connectors of the computing node to the connectors of the switching node is often very large (i.e., the connectors of many computing nodes correspond to the connectors of one switching node, which may be equal to or more than the number of connectors owned by each computing node). In this case, attention needs to be paid to ensuring the realization of the fully interconnected architecture. Therefore, when designing the connection topology of the additional bandwidth part according to the 6:1 ratio, the connectors of the computing nodes selected need to be carefully considered. The principle is that each additional switching node can be connected to each computing node.

[0140] Therefore, the extra parts of the three connectors of each computing node + the extra parts of the three connectors of the next computing node can be used to connect the extra parts (i.e., the second pin set) and the connectors of the switch (i.e., the second switching connectors). In this way, each extra switch can be fully interconnected with 16 computing nodes. Figure 12 It is a schematic diagram of the connection relationship between a computing node and a switching node according to an embodiment of the present application Figure 2 , Figure 12 There is a connection relationship between the connector parts of the computing nodes with the same number and the connector parts of the switching node in []. Referring to the above design, the topological connection relationship as shown in Figure 12 can be finally obtained.

[0141] Through the above method of non-fixed ratio of the cable tray, when GPUs with different bandwidths are installed in the computing nodes, the ratio of the common PIN (pin) parts of the connectors of the computing nodes and the ratio of the connectors of the switch are different from the ratio of the extra bandwidth parts of the connectors of the computing nodes and the connectors of the switch. Thus, it is possible to flexibly perform a compatibility design according to different GPU bandwidths without affecting the existing topological design.

[0142] Those skilled in the art can further realize that the units and algorithm steps of each example described in combination with the embodiments disclosed in this article can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0143] Specific examples are used in this article to elaborate on the principles and implementation manners 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 in this technical field, without departing from the principle of the present application, several improvements and modifications can 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 server cabinet, characterized in that, Including: A computing connector, a first switching connector, and a second switching connector, wherein the computing connector is divided into a first pin set and a second pin set; There is a first connection relationship between the first pin set and the first switching connector, and the first connection relationship matches the processor information of a first processor with a target bandwidth. Among them, the first pin set corresponds to the signal of the target bandwidth, and the first connection relationship includes: a first quantity of the computing connector, a second quantity of the first switching connector, and a first ratio. The first ratio is used to indicate the proportional relationship between the first pin set connecting the first switching connector and the proportional relationship between the first switching connector connecting the first pin set. The second quantity is calculated based on the first quantity, the target bandwidth, the switch bandwidth of the switching processor to be connected by the first switching connector, and the first ratio; There is a second connection relationship between the second pin set and the second switching connector, and the second connection relationship matches the bandwidth difference and the first connection relationship. Among them, the bandwidth difference is the difference between the reference bandwidth of a second processor and the target bandwidth. The second pin set corresponds to the signal of the bandwidth difference, and the second connection relationship includes: a second ratio and a third quantity of the second switching connector. The second ratio is used to indicate the proportional relationship between the second pin set connecting the second switching connector and the proportional relationship between the second switching connector connecting the second pin set. The second ratio is determined based on the first connection relationship, the target bandwidth, and the bandwidth difference.

2. The server cabinet according to claim 1, wherein, The server cabinet further includes a first switching node chassis, and the first switching connector is deployed on the first switching node chassis; The quantity of the first switching node chassis is the ratio of the second quantity to a first unit quantity, and the first unit quantity is the quantity of the first switching connectors that each first switching node chassis allows to accommodate; 3. The server cabinet according to claim 2, wherein The quantity of the computing connector is a first quantity, wherein the first quantity is determined based on the quantity of the first processor. The server cabinet further includes a computing node chassis, and the computing connector is deployed on the computing node chassis; The quantity of the computing node chassis is the ratio of the first quantity to a second unit quantity, and the second unit quantity is the quantity of the computing connectors that each computing node chassis allows to accommodate; Under the first connection relationship, each of the first switching node chassis is connected to all of the computing node chassis.

4. The server cabinet according to claim 1, wherein The third quantity is calculated based on the first quantity of the computing connector and the second ratio.

5. The server cabinet according to claim 4, characterized in that, The server cabinet further includes a second switching node chassis, and the second switching connector is deployed in the second switching node chassis; The quantity of the second switching node chassis is the ratio of the third quantity to a first unit quantity, and the first unit quantity is the quantity of the first switching connectors that each first switching node chassis allows to accommodate.

6. The server cabinet according to claim 5, wherein The number of the computing connectors is a first number, where the first number is determined according to the number of the first processors. The server cabinet further includes a computing node chassis, and the computing connectors are deployed in the computing node chassis; The number of the computing node chassis is the ratio of the first number to a second unit number, where the second unit number is the number of the computing connectors that each computing node chassis allows to accommodate; Each of the second switching node chassis is connected to all of the computing node chassis under the second connection relationship.

7. The server cabinet according to claim 1, characterized in that, The server cabinet further includes: a host computer and a display device, the host computer is connected to the display device, where, The host computer is configured to: obtain the processor information and the reference bandwidth; match the first connection relationship between the first pin set and the first switching connector according to the processor information; match the second connection relationship between the second pin set and the second switching connector according to the bandwidth difference and the first connection relationship; The display device is configured to: display a connector layout diagram of the server cabinet, where the connector layout diagram shows the computing connectors that divide the first pin set and the second pin set, the first switching connector, the second switching connector, the first connection relationship, and the second connection relationship.

8. The server cabinet according to claim 7, characterized in that, The host computer is further configured to: Determine the first number of the computing connectors according to the number of the first processors; Calculate a second number of the first switching connectors according to the first number, the target bandwidth, the switch bandwidth of the switching processors to which the first switching connectors are to be connected, and a first ratio, where the processor information includes: the number of the first processors and the target bandwidth, the first connection relationship includes: the first number, the second number, and the first ratio, and the first ratio is used to indicate the proportional relationship between the first pin set and the first switching connector and the proportional relationship between the first switching connector and the first pin set.

9. The server cabinet according to claim 8, wherein The host computer is further configured to: Calculate an initial number of the first switching connectors according to the first number, the target bandwidth, the switch bandwidth, and an initial ratio; Obtain the first ratio according to the adjustment information of the initial ratio; Adjust the initial number according to the first ratio to obtain the second number.

10. The server cabinet according to claim 9, characterized in that, The host computer is further configured to: When the initial ratio is used to indicate that each computing connector is connected to one first switching connector and each first switching connector is connected to one computing connector, calculate the product of the first number and the target bandwidth to obtain a first value; Calculate the ratio of the first value to the switch bandwidth to obtain the initial number.

11. The server cabinet according to claim 9, characterized in that, The host computer is further configured to: When the adjustment information is used to indicate that the initial ratio is adjusted and the adjustment manner of the initial ratio, adjust the initial ratio to the first ratio according to the adjustment manner; In the case where the adjustment information is used to indicate that the initial ratio is not adjusted, determine the initial ratio as the first ratio.

12. The server cabinet according to claim 7, characterized in that, The host computer is further configured to: Determine a second ratio according to the first connection relationship, the target bandwidth, and the bandwidth difference, where the second ratio is used to indicate the proportional relationship between the second pin set and the second switching connector and the proportional relationship between the second switching connector and the second pin set; Calculate a third quantity of the second switching connector according to the first quantity of the computing connector and the second ratio, where the second connection relationship includes: the second ratio and the third quantity.

13. The server cabinet according to claim 12, characterized in that, The host computer is further configured to: Determine the connector bandwidth of the first switching connector according to the first connection relationship and the target bandwidth; Determine the second ratio according to the connector bandwidth and the bandwidth difference.

14. The server cabinet according to claim 13, characterized in that, The host computer is further configured to: Extract the set quantity of the first pin set connected to each of the first switching connectors from the first connection relationship; Determine the product of the set quantity and the target bandwidth as the connector bandwidth.

15. The server cabinet according to claim 8, wherein, The server cabinet further includes a computing node chassis and a first switching node chassis. The computing connectors are deployed on the computing node chassis, and the first switching connectors are deployed on the first switching node chassis. The computing node chassis is used to install computing nodes, and the first switching node chassis is used to install first switching nodes. The host computer is further configured to: Calculate the ratio of the second quantity of the first switching connector to the first unit quantity to obtain the quantity of the first switching node chassis, where the first unit quantity is the quantity of the first switching connectors that each first switching node chassis is allowed to accommodate, and the ratio of the first quantity to the second unit quantity is the quantity of the computing node chassis, and the second unit quantity is the quantity of the computing connectors that each computing node chassis is allowed to accommodate.

16. The server cabinet according to claim 15, wherein The display device is further configured to: Divide the computing connectors on the connector layout diagram according to the quantity of the computing node chassis, and divide the first switching connectors according to the quantity of the first switching node chassis, where each of the first switching node chassis is connected to all of the computing node chassis under the first connection relationship.

17. The server cabinet according to claim 15, characterized in that, The server cabinet further includes a computing node chassis and a second switching node chassis. The computing connectors are deployed on the computing node chassis, and the second switching connectors are deployed on the second switching node chassis. The computing node chassis is used to install computing nodes, and the second switching node chassis is used to install second switching nodes. The host computer is further configured to: Calculate the ratio of the third quantity of the second switching connector to the first unit quantity to obtain the quantity of the second switching node chassis.

18. The server cabinet according to claim 17, wherein The display device is further configured to: Divide the second switching connectors on the connector layout diagram according to the quantity of the second switching node chassis, where each of the second switching node chassis is connected to all of the computing node chassis under the second connection relationship.

19. A server system, characterized in that, Includes: A computing node, a switching node, and a server cabinet as recited in any one of claims 1 to 18, wherein a computing connector included in the server cabinet connects a server processor in the computing node, and a switching connector included in the server cabinet connects a switching processor in the switching node.

Images