Cable module, server, server configuration method and product

Abstract

The present application relates to the field of server technology, and discloses a cable module, a server, a server configuration method and a product. The method includes: a board end and a first cable; the two sides of the board end form an electrical connection, wherein a first connector is fixed to one side of the board end, and a welding surface is provided on the other side of the board end; one end of the first cable is welded on the welding surface; a second connector is fixed to the other end of the first cable; the first connector and the second connector are respectively connected to a first data processing unit and a second data processing unit; the first data processing unit and the second data processing unit are used to perform artificial intelligence tasks. Using the cable module provided by the present application as a relay device in the server can reduce costs, improve the stability of the communication link, enable the server to adapt to more usage scenarios, and improve user experience.

Description

Technical Field

[0001] The present application relates to the field of server technology, and in particular to a cable module, a server, a server configuration method and a product. Background Art

[0002] With the development of artificial intelligence, AI (Artificial Intelligence) servers with integrated powerful data processing units are in short supply. In AI scenarios, servers are usually deployed in separate chassis. The server can be composed of a first chassis and a second chassis, and data processing units are respectively set in the first chassis and the second chassis. For example, the first chassis uses a general-purpose server with a CPU (Central Processing Unit) inside, and the second chassis only carries a GPU (Graphics Processing Unit) without a CPU. In order to achieve high-speed interconnection between different chassis and extend the signal transmission distance, relay devices are usually used in the link. Since relay devices have insertion loss, adding relay devices to the link will increase the total insertion loss of the link accordingly. In order to ensure the integrity of the transmission signal, the total insertion loss of the link (including cable insertion loss, pad insertion loss, device insertion loss, etc.) usually needs to be kept within a limited range.

[0003] In the related art, a retimer (Retimer card) or a passive card is used as a relay device. The retimer is an active device, and when used in a link, it requires a long time of parameter debugging, which is a complicated process and wastes costs. In addition, the retimer has the problem of excessive delay, which affects the link transmission performance. Although the passive card does not involve parameter debugging, the internal circuit is complex and the insertion loss is large. When using a passive card as a relay device, in order to ensure that the total insertion loss of the link is within a limited range, only shorter communication cables can be used, which affects the transmission performance of the link, limits the use scenarios of the server, and affects the user experience.

[0004] Given that the relay devices used in related technologies have problems such as difficulty in adjusting parameters, waste of costs and large insertion losses, how to improve the transmission performance of the server communication link and make the server adapt to more usage scenarios is a problem that needs to be solved. Summary of the invention

[0005] In view of this, the present application aims to propose a cable module, a server, a server configuration method and a product to improve the transmission performance of the server communication link and enable the server to adapt to more usage scenarios.

[0006] To achieve the above purpose, the technical solution of this application is as follows:

[0007] A first aspect of an embodiment of the present application provides a cable module, comprising: a board end and a first cable;

[0008] The two sides of the board end form an electrical connection, wherein a first connector is fixed to one side of the board end, and a welding surface is provided on the other side of the board end; one end of the first cable is welded to the welding surface;

[0009] A second connector is fixed to the other end of the first cable; the first connector and the second connector are connected to a first data processing unit and a second data processing unit, respectively; the first data processing unit and the second data processing unit are used to perform artificial intelligence tasks.

[0010] Optionally, the welding surface is arranged within a projection area of ​​the first connector on the board end;

[0011] At least one via hole is provided on the board end, and the via hole is provided within a projection area of ​​the welding surface on the board end;

[0012] The two sides of the board end are electrically connected through the via hole.

[0013] According to a second aspect of an embodiment of the present application, there is provided a server, comprising: a first chassis and a second chassis, wherein the first chassis communicates with the second chassis via a second cable;

[0014] The first chassis is provided with a central processing unit and the cable module provided by the first aspect of the embodiment of the present application; wherein the central processing unit is provided on the mainboard and is plugged with the second connector of the cable module through the second interface on the mainboard;

[0015] A graphics processor and an adapter card are deployed in the second chassis; wherein the graphics processor is connected to the adapter card via a high-speed connector; and the adapter card is connected to the cable module via the second cable.

[0016] Optionally, the adapter board is provided with a first interface;

[0017] The two ends of the second cable are fixed with first connectors, which are respectively connected to the first interface on the adapter card and the first connector of the cable module.

[0018] Optionally, the length of the second cable is not greater than a first threshold;

[0019] The first threshold is determined based on an insertion loss of a link between the central processing unit and the graphics processing unit.

[0020] Optionally, the length of the second cable is not less than a second threshold; the second threshold is determined based on a first position of the first chassis and a second position of the second chassis.

[0021] Optionally, the second cable is a fifth-generation high-speed serial computer expansion bus; and the adapter board is a high-speed serial computer expansion bus standard board.

[0022] Optionally, the server comprises: one first chassis and one second chassis;

[0023] The adapter board includes at least one upstream interface; the adapter board is connected to a corresponding number of second cables through the at least one upstream interface;

[0024] The first chassis includes at least one cable module, each cable module is connected to a second cable;

[0025] The central processing unit is connected to a cable module respectively through a plurality of second interfaces on the mainboard.

[0026] Optionally, the server includes: one first chassis and at least two second chassis;

[0027] Each adapter board includes at least one upstream interface; each adapter board is connected to a corresponding number of second cables through at least one upstream interface;

[0028] The first chassis includes at least two cable modules, each cable module is connected to a second cable;

[0029] The central processing unit is connected to a cable module respectively through a plurality of second interfaces on the mainboard.

[0030] Optionally, in the server, the number of second cables is not greater than the maximum number of cable modules that can be connected to the central processing unit.

[0031] According to a third aspect of an embodiment of the present application, a server configuration method is provided, the method comprising:

[0032] Based on the first chassis and the second chassis in the server provided by the second aspect of the embodiment of the present application, a target server is configured; the target server includes: a first chassis and at least one second chassis;

[0033] Detecting and obtaining the insertion loss of the cable module in the first chassis;

[0034] Based on the insertion loss of the cable module, a first threshold is calculated as an upper limit of the length of a second cable; the second cable is used to connect the first chassis and the second chassis;

[0035] Calculate a second threshold based on the distance between the first chassis and the second chassis as a lower limit of the length of the second cable;

[0036] Based on the first threshold and the second threshold, a length of the second cable is determined.

[0037] Optionally, the first chassis and the at least one second chassis are arranged in an up-and-down manner;

[0038] When the number of the second chassis is not less than 2, the first chassis is located between the two second chassis.

[0039] According to a fourth aspect of an embodiment of the present application, a computer program product is provided, including a computer program. When the computer program is executed by a processor, the steps in the server configuration method provided in the third aspect of the embodiment of the present application are implemented.

[0040] According to a fifth aspect of an embodiment of the present application, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the server configuration method provided in the third aspect of the embodiment of the present application are implemented.

[0041] According to the sixth aspect of an embodiment of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the server configuration method provided in the third aspect of the embodiment of the present application are implemented.

[0042] Compared with relay devices such as retimers and passive cards, the cable module provided in the present application uses welding to fix the first connector, the first cable and the second connector. In the server, the cable module serves as a relay device, and different data processing units are connected through the first connector and the second connector respectively, thereby reducing the number of detachable interfaces in the link, reducing the probability of link failure, and improving the stability of the link.

[0043] At the same time, compared with devices such as retimers and passive cards, cable modules also save complex internal wiring. Therefore, using cable modules as relay devices in server communication links greatly reduces the insertion loss of the link and improves the link transmission performance. While ensuring signal integrity, the server can use longer communication cables, which improves the server's adaptability to different usage scenarios and improves user experience.

[0044] In addition, the use of low-cost cable modules to replace components such as retimers and passive cards also reduces server costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the description of the embodiments of the present application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0046] Figure 1 is a structural schematic diagram of a cable module proposed in one embodiment of the present application;

[0047] Figure 2 It is one of the schematic diagrams of the internal architecture of a server in the related art;

[0048] Figure 3 This is the second schematic diagram of the internal architecture of the server in the related art;

[0049] Figure 4 It is a schematic diagram of the architecture of a server proposed in an embodiment of the present application;

[0050] Figure 5 is a flow chart of a server configuration method proposed in an embodiment of the present application;

[0051] FIG6 (a) is a schematic diagram of the layout of the first chassis and the second chassis of the server;

[0052] FIG6 (b) is a second schematic diagram of the layout of the first chassis and the second chassis of the server;

[0053] FIG6 (c) is a third schematic diagram of the layout of the first chassis and the second chassis of the server;

[0054] FIG6 (d) is a fourth schematic diagram of the layout of the first chassis and the second chassis of the server;

[0055] Figure 7 It is a schematic diagram of an electronic device proposed in an embodiment of the present application. DETAILED DESCRIPTION

[0056] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0057] It should be understood that the references to "one embodiment" or "an embodiment" throughout the specification mean that the specific features, structures, or characteristics associated with the embodiment are included in at least one embodiment of the present application. Therefore, the references to "in one embodiment" or "in an embodiment" appearing throughout the specification do not necessarily refer to the same embodiment. In addition, these specific features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0058] In the various embodiments of the present application, it should be understood that the size of the serial numbers of the following processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0059] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are merely examples of devices and methods consistent with some aspects as detailed in the present application.

[0060] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application may be combined with each other.

[0061] Figure 2 It is one of the schematic diagrams of the internal architecture of the server in the related technology. Figure 2 As shown, the server includes a first chassis and a second chassis, wherein a mainboard with a central processing unit (CPU) and a retimer board are arranged in the first chassis. A graphics processing unit (GPU) and a PCIe board are arranged in the second chassis. The interface on the CPU mainboard is connected to the connector of the retimer board through an internal high-speed cable, and the other end of the retimer is connected to the interface of the external high-speed cable and the PCIe board (carrying a switching chip) through a connector. A high-speed link from the central processing unit to the graphics processing unit is realized between the interface of the PCIe board and the graphics processing unit (GPU) through a high-density connector board.

[0062] In the server architecture, if the server is to be gradually stabilized, the equalization parameters of the retimer need to be debugged, which increases the configuration complexity and usage cost of the server. In addition, the relay device has a certain failure rate, which increases the security risk of the communication link. At the same time, the retimer has a delay and cannot meet the time-sensitive AI calculations.

[0063] Figure 3 This is the second schematic diagram of the internal architecture of the server in the related technology. Figure 3As shown, the server includes a first chassis and a second chassis, wherein a mainboard with a central processing unit deployed and a passive card are arranged in the first chassis. A graphics processor and a PCIe board loaded with a switching chip are arranged in the second chassis. The PCIe board and the graphics processor are connected through a high-speed connector. An internal high-speed cable is also arranged in the first chassis, and an external high-speed cable is arranged between the first chassis and the second chassis. One end of the internal high-speed cable is connected to the first internal port of the passive card, and the other end is connected to the second internal port on the mainboard; one end of the external high-speed cable is connected to the connector of the passive card through a connector, and the other end is connected to the interface on the PCIe board through a connector. This solution uses a passive card as a relay device in the link. Compared with the retimer, the passive card in this solution does not need to debug the equalization parameters, which reduces the use cost. However, due to the large insertion loss of the passive card, only a shorter external high-speed cable can be used in the link, which limits the use scenario of the server and affects the user experience.

[0064] The present application will be described in detail below with reference to the accompanying drawings and in combination with embodiments.

[0065] Figure 1 Schematic diagram of the structure of a cable module proposed in one embodiment of the present application. Figure 1 As shown, the cable module includes: a board end and a first cable;

[0066] The two sides of the board end form an electrical connection, wherein a first connector is fixed to one side of the board end, and a welding surface is provided on the other side of the board end; one end of the first cable is welded to the welding surface;

[0067] A second connector is fixed to the other end of the first cable; the first connector and the second connector are connected to a first data processing unit and a second data processing unit, respectively; the first data processing unit and the second data processing unit are used to perform artificial intelligence tasks.

[0068] like Figure 1 As shown, in this embodiment, the cable module includes a board end and a first cable, the front of the board end has a first connector, which can be used to plug in a communication device with the first connector, and the back of the board end is provided with a welding surface with a soldering pad, and one end of the first cable is soldered to the welding surface through the soldering pad. The front and back of the board end form an electrical connection, so that the first cable forms an electrical connection with the first connector. The other end of the first cable has a second connector, which can be used to plug in a communication device with the second connector. The first connector and the second connector can be connectors of the same type or different types.

[0069] In practical applications, the first connector and the second connector can be set according to the needs of the practical application. For example, in a PCIe (Peripheral Component Interconnect Express, a high-speed serial computer expansion bus standard) high-speed communication link, the cable module is deployed in the link as a relay device, and the eASM interface connector can be used as the first connector of the cable module, and the MCIO interface connector can be used as the second connector of the cable module.

[0070] Since the first connector and the second connector in the cable module are fixedly connected to both ends of the cable respectively, compared with the method of plugging the retimer and the passive card with the internal cable in the link in the related art, the detachable sockets and a large number of wiring in the link are saved, and the total insertion loss of the link is reduced, thereby improving the transmission performance of the link. When the cable module is used to replace the retimer or passive card, the server can use longer communication cables, thereby adapting to more usage scenarios and improving the user experience. In addition, reducing the detachable sockets in the link also reduces the probability of link failure and improves the stability of the link. Moreover, the use of low-cost cable modules to replace devices such as retimers and passive cards further reduces the cost of the server.

[0071] As an implementation manner of the present application, the welding surface is arranged within a projection area of ​​the first connector on the board end;

[0072] At least one via hole is provided on the board end, and the via hole is provided within a projection area of ​​the welding surface on the board end;

[0073] The two sides of the board end are electrically connected through the via hole.

[0074] In one embodiment, the first connector disposed on the front side of the board end of the cable module is positioned opposite to the welding surface disposed on the back side of the board end, and electrical connection is achieved through vias disposed on the board end. Specifically, the first connector is fixed to the front side of the board end by crimping and welding, and the welding surface is disposed on the back side of the board end and is located within the projection area of ​​the first connector on the board end. At least one via is disposed within the common projection area of ​​the first connector and the welding surface on the board end, and electrical connection is achieved between the first connector on the front side of the board end and the first cable welded on the welding surface on the back side of the board end through the via.

[0075] In this embodiment, by aligning the welding surface with the shadow area of ​​the first connector on the board end and setting a via in the shadow area of ​​the welding surface on the board end, the shortest route from the first connector to the first cable is achieved on the board end, thereby further reducing the insertion loss of the cable module.

[0076] Based on the same inventive concept, an embodiment of the present application provides a server, comprising: a first chassis and a second chassis, wherein the first chassis communicates with the second chassis via a second cable;

[0077] The first chassis is provided with a central processing unit and a cable module as described in the above embodiment; wherein the central processing unit is provided on a mainboard and is plugged into the second connector of the cable module via a second interface on the mainboard;

[0078] A graphics processor and an adapter card are deployed in the second chassis; wherein the graphics processor is connected to the adapter card via a high-speed connector; and the adapter card is connected to the cable module via the second cable.

[0079] Figure 4 FIG. 1 is a schematic diagram of the server architecture proposed in one embodiment of the present application. Figure 4 As shown, in this embodiment, the server includes a first chassis and a second chassis, and the two chassis communicate through a second cable, and the second cable is located outside the first chassis and the second chassis. The first chassis is deployed with a central processing unit (CPU), and at least one cable module is arranged in the first chassis. A second interface adapted to the second connector is arranged on the mainboard where the central processing unit is located, and the central processing unit is connected to the second connector of the cable module through the second interface on the mainboard. Specifically, in the case of multiple cable modules, a plurality of second interfaces equal to the number of cable modules are arranged on the mainboard where the central processing unit is located, and each interface is respectively connected to the second connector of a cable module.

[0080] The first connector of the cable module is connected to the second cable outside the chassis, and a communication link between the first chassis and the second chassis is established through the second cable. The second chassis is deployed with a graphics processor (GPU), an adapter card and a high-speed connector, and the graphics processor and the adapter card communicate through the high-speed connector. The adapter card carries a switching chip, and the adapter card is provided with a first interface adapted to the first connector. The first connector at one end of the second cable is connected to the second chassis through the first interface to realize communication between the central processing unit in the first chassis and the graphics processor in the second chassis.

[0081] Optionally, the first connector and the second connector may be connectors with different types of interfaces, such as an eASM connector, a CDFP connector, an MCIO connector, etc. In this embodiment, an eASM connector is used as the first connector, and an MCIO connector is used as the second connector.

[0082] In addition, different data processing units can be deployed in the first chassis and the second chassis according to actual application requirements, such as central processing units, graphics processing units, field programmable gate arrays, and application-specific integrated circuits that can be used to process AI computing. There are many options for high-speed connectors that connect the adapter card and the GPU, including PCIe connectors, NVLink connectors, and CXL (Compute Express Link) connectors.

[0083] As an implementation mode of the present application, the adapter board is provided with a first interface;

[0084] The two ends of the second cable are fixed with first connectors, which are respectively connected to the first interface on the adapter card and the first connector of the cable module.

[0085] In one embodiment, both ends of the second cable are provided with first connectors of the same type, and the adapter card in the second chassis is provided with a first interface adapted to the first connector. The second cable is plugged into the first interface of the adapter card through the first connector, thereby establishing a communication link between the central processing unit in the first chassis and the graphics processing unit in the second chassis.

[0086] As an implementation manner of the present application, the length of the second cable is not greater than a first threshold;

[0087] The first threshold is determined based on an insertion loss of a link between the central processing unit and the graphics processing unit.

[0088] In one embodiment, the length of the second cable between the first chassis and the second chassis of the server is determined based on the insertion loss of the link between the central processing unit in the first chassis and the graphics processing unit in the second chassis. In order to ensure the signal integrity of the link transmission between the first chassis and the second chassis, the insertion loss value of the entire link cannot exceed the specification value. Therefore, in this embodiment, the first threshold is calculated based on the overall insertion loss of the transmission link between the two processors, that is, the upper limit of the length that the second cable can reach is determined. The following is an example of a PCIe high-speed link.

[0089] The insertion loss requirement of the PCIe Gen5 high-speed link is not more than the specification value of -36dB. When a passive card is used, the first threshold is calculated to be 300mm, that is, the length of the second cable can be up to 300mm. In this case, the layout position of the first chassis and the second chassis is very limited and cannot adapt to many application scenarios. Figure 6 (a) is one of the layout diagrams of the first chassis and the second chassis of the server. As shown in Figure 6 (a), in the scenario of using a passive card, the first chassis must be placed next to the second chassis, and the placement of the first chassis and the second chassis must ensure that the actual wiring length between the first connectors of the two chassis does not exceed 300mm.

[0090] In this embodiment, the actual insertion loss of the passive card is obtained through actual measurement, including the routing insertion loss (2 inches) of the passive card -1.87 dB / 16Ghz, the insertion loss of the via and the pad -2.1 dB / 16Ghz, and the device insertion loss is reduced to -0.3 dB / 16Ghz after replacing with the cable module, that is, in this embodiment, the cable module is used to replace the passive card, and the insertion loss value of 3.67dB / 16Ghz can be left within the specification value range to increase the length of the second cable. The insertion loss of the 300mm cable is measured to be -3.65 dB / 16Ghz, and the length of the second cable can be roughly increased to about 601mm. The length of the second cable determined by actual testing can reach 605mm (insertion loss value -7.32 dB / 16Ghz).

[0091] In this embodiment, based on the insertion loss of the link between the central processing unit of the first chassis and the graphics processing unit of the second chassis, the maximum length that the second cable can reach is determined, so as to ensure the integrity of the signal during transmission and increase the length of the second cable, so as to make the layout of the first chassis and the second chassis more flexible and adapt to different application scenarios.

[0092] In one embodiment, a server includes a first chassis and a second chassis, and a mainboard with a central processing unit deployed therein and a cable module are arranged in the first chassis. A graphics processing unit and a PCIe board carrying a switching chip are arranged in the second chassis. The PCIe board and the graphics processing unit are connected via a high-speed connector. The PCIe board is used to realize communication between the central processing unit and the graphics processing unit. One end of the cable module in the first chassis is plugged into the MCIO interface on the mainboard via an MCIO connector, and the other end is connected to a high-speed cable between the first chassis and the second chassis via an eASM connector. The other end of the high-speed cable also uses an eASM connector to be plugged into the eASM interface on the PCIe board in the second chassis, so that the central processing unit and the graphics processing unit are interconnected, and a communication link exists between the first chassis and the second chassis.

[0093] In one embodiment, a server includes a first chassis and a plurality of second chassis, wherein a mainboard with a central processing unit deployed therein and a cable module are arranged in the first chassis. A graphics processor and a PCIe board carrying a switching chip are arranged in each second chassis. The PCIe board and the graphics processor are connected via a high-speed connector. The PCIe board is used to realize communication between the central processing unit and the graphics processor. A plurality of cable modules are arranged in the first chassis, wherein one end of each cable module is plugged into an MCIO interface on the mainboard via an MCIO connector, and the other end is connected to a high-speed cable located between the first chassis and the second chassis via an eASM connector. The other end of each high-speed cable is also plugged into an eASM interface on the PCIe board in the second chassis using an eASM connector, so that the central processing unit and the graphics processor are interconnected, and there are multiple communication links between the first chassis and the second chassis.

[0094] As an implementation manner of the present application, the length of the second cable is not less than a second threshold; the second threshold is determined based on a first position where the first chassis is located and a second position where the second chassis is located.

[0095] In one embodiment, the length of the second cable between the first chassis and the second chassis of the server is determined based on the link insertion loss between the central processing unit in the first chassis and the graphics processing unit in the second chassis, as well as the first position of the first chassis and the second position of the second chassis. In this embodiment, determining the length of the second cable requires ensuring the integrity of the transmission signal in the link and the layout of the server in the application scenario.

[0096] Specifically, in order to ensure the signal integrity of the link transmission between the first chassis and the second chassis, the insertion loss value of the entire link cannot exceed the specification value. Therefore, the first threshold is calculated according to the overall insertion loss of the transmission link between the two processors, that is, the upper limit of the length that the second cable can reach is determined.

[0097] In order to meet the equipment layout requirements of the current application scenario, it is also necessary to determine the minimum length of the second cable between the first chassis and the second chassis, that is, the second threshold, based on the location of the first chassis (i.e., the first location) and the location of the second chassis (i.e., the second location). In actual applications, the second threshold is determined based on the distance between the first chassis and the second chassis processed based on the length of the second cable. It is also necessary to consider the interface location on the server, the layout of other external cables, and reserve redundant length.

[0098] Finally, based on the first threshold and the second threshold, the length of the second cable is determined. Specifically, the first threshold is used as the upper limit of the length of the second cable, and the second threshold is used as the lower limit of the length of the second cable, so as to determine the actual length of the second cable.

[0099] In this embodiment, the upper limit of the length of the second cable is determined based on the overall insertion loss on the communication link between the central processing unit in the first chassis and the graphics processing unit in the second chassis, and the lower limit of the length of the second cable is determined based on the actual layout requirements of the first chassis and the second chassis in the application scenario. This allows the length of the second cable to be flexibly adjusted while ensuring the integrity of the transmission signal in the link and the layout requirements of different application scenarios.

[0100] As an implementation manner of the present application, the second cable is a fifth-generation high-speed serial computer expansion bus; and the adapter board is a high-speed serial computer expansion bus standard board.

[0101] In one embodiment, a high-speed cable of the PCIe Gen5 standard is selected as the second cable connecting the first chassis and the second chassis, and accordingly, a PCIe adapter card of a corresponding type is selected as the adapter card in the second chassis. Accordingly, a high-speed connector between the adapter card and the CPU adopts a PCIe connector.

[0102] As an implementation manner of the present application, the server includes: one first chassis and one second chassis;

[0103] The adapter board includes at least one upstream interface; the adapter board is connected to a corresponding number of second cables through the at least one upstream interface;

[0104] The first chassis includes at least one cable module, each cable module is connected to a second cable;

[0105] The central processing unit is connected to a cable module respectively through a plurality of second interfaces on the mainboard.

[0106] In one embodiment, the server includes a first chassis and a second chassis. There are multiple communication links between the first chassis and the second chassis. Specifically, a central processing unit and multiple cable modules are deployed in the first chassis. A plurality of second interfaces, the same number as the number of cable modules, are provided on a mainboard where the central processing unit is located, and each interface is respectively connected to a second connector of a cable module.

[0107] The first connector of each cable module is connected to a second cable outside the chassis, and a communication link between the first chassis and the second chassis is established through the second cable. A graphics processor, an adapter board and a high-speed connector are deployed in the second chassis, and the graphics processor and the adapter board communicate through the high-speed connector. In the second chassis, the adapter board includes multiple upstream interfaces (first interfaces adapted to the first connector), wherein each upstream interface is plugged with a second cable. The first interface is connected to the second chassis to realize communication between the central processing unit in the first chassis and the graphics processor in the second chassis.

[0108] In this embodiment, multiple communication links are set between the central processing unit and the graphics processing unit in the server to improve the data transmission performance of the server. Cable modules are deployed on each communication link to replace relay devices such as retimers or passive cards, thereby reducing link insertion loss and further improving transmission performance.

[0109] As an implementation manner of the present application, the server includes: one first chassis and at least two second chassis;

[0110] Each adapter board includes at least one upstream interface; each adapter board is connected to a corresponding number of second cables through at least one upstream interface;

[0111] The first chassis includes at least two cable modules, each cable module is connected to a second cable;

[0112] The central processing unit is connected to a cable module respectively through a plurality of second interfaces on the mainboard.

[0113] In the above embodiment, the server includes a first chassis and multiple second chassis. There are one or more communication links between the first chassis and each second chassis. Specifically, a central processing unit and multiple cable modules are deployed in the first chassis. On the mainboard where the central processing unit is located, multiple second interfaces are provided, which are the same in number as the cable modules, and each interface is respectively connected to a second connector of a cable module.

[0114] The first connector of each cable module is connected to a second cable outside the chassis, and a communication link between the first chassis and each second chassis is established through the second cable. A graphics processor, an adapter board and a high-speed connector are deployed in the second chassis, and the graphics processor and the adapter board communicate through the high-speed connector. In each second chassis, the adapter board includes multiple upstream interfaces (first interfaces adapted to the first connector). In each second chassis, the adapter board is respectively plugged into a second cable through one or more upstream interfaces to realize communication between the central processing unit in the first chassis and the graphics processors in multiple second chassis.

[0115] In this embodiment, multiple communication links are set between the central processing unit and multiple graphics processors in the server to improve the data parallel processing performance of the server. Cable modules are deployed on each communication link to replace relay devices such as retimers or passive cards, thereby reducing link insertion loss, improving transmission performance, and further improving data processing efficiency.

[0116] As an implementation manner of the present application, in the server, the number of second cables is not greater than the maximum value of the cable modules that can be connected to the central processing unit.

[0117] In the above embodiment, the number of cable modules that can be connected to the central processing unit of the first chassis in the server has an upper limit. In view of the fact that the number of second cables in the server is not greater than the number of cable modules deployed in the first chassis, the upper limit of the number of cable modules can be used as the upper limit of the number of second cables, and the total number of uplink interfaces occupied by the adapter cards in all second chassis in the server is the same as the number of second cables, that is, it is also not greater than the upper limit. For example, when 8 cable modules are deployed in the first chassis in the server, the number of second cables between the first chassis and the second chassis is not greater than 8.

[0118] Based on this, according to the number of second chassis in the server, the communication link between the first chassis and each second chassis is flexibly established in an equal or unequal manner, thereby forming different server architectures to adapt to diverse application scenarios.

[0119] For example, in a case where the server includes a first chassis and two second chassis, eight cable modules are deployed in the first chassis, each of which is connected to eight second cables. All second cables are connected to adapter cards in the two second chassis in an evenly distributed manner, wherein each adapter card is connected to four second cables via four uplink interfaces.

[0120] In the case where the server includes a first chassis and three second chassis, eight cable modules are deployed in the first chassis and are respectively connected to eight second cables. All second cables are respectively connected to adapter cards in the three second chassis, wherein an adapter card of one second chassis is connected to four second cables via four uplink interfaces, and adapter cards of two second chassis are respectively connected to two second cables via two uplink interfaces.

[0121] In one embodiment, the server includes: a first chassis and at least a plurality of second chassis. The plurality of second chassis form at least one computing unit, and each computing unit includes at least two first chassis. The first chassis is connected to any second chassis in each computing unit via at least one second cable, and the second chassis in each computing unit are connected to each other via at least one second cable.

[0122] In this embodiment, the server is used to perform large-scale AI operations, wherein the central processing unit inside the first chassis, after receiving the target operation task, processes the target operation task according to the number of computing units in the server, generates subtasks equal to the number of computing units, and distributes each subtask to each computing unit. The graphics processor in each computing unit performs parallel operations on the received subtasks and returns the calculation results to the central processing unit. By using cable modules instead of retimers or passive cards in large server architectures, the overall insertion loss of the link is significantly reduced, the server performance is improved, and the resources consumed by equipment operations are greatly saved, reducing the equipment development and use costs.

[0123] Based on the same inventive concept, an embodiment of the present application provides a server configuration method. Figure 5 , Figure 5 1 is a flow chart of a server configuration method proposed in an embodiment of the present application. Figure 5 As shown, the method includes:

[0124] S1: Based on the first chassis and the second chassis provided in the above embodiment, configuring a target server; the target server includes: a first chassis and at least one second chassis;

[0125] S2: Detect and obtain the insertion loss of the cable module in the first chassis;

[0126] S3: Calculate a first threshold based on the insertion loss of the cable module as an upper limit of the length of a second cable; the second cable is used to connect the first chassis and the second chassis;

[0127] S4: Calculate a second threshold based on the distance between the first chassis and the second chassis as a lower limit of the length of the second cable;

[0128] S5: Determine the length of the second cable based on the first threshold and the second threshold.

[0129] In this embodiment, when configuring the server, first determine the number of second chassis that need to be deployed in the server. Determine the insertion loss of the cable module used, and further calculate the insertion loss of the entire link between the first chassis and each second chassis when the cable module is deployed. Based on the overall insertion loss on each link, calculate the first threshold as the upper limit of the length of the second cable. Then, based on the scenario in which the server needs to be applied, determine the layout of the server. Determine the specific positions of the first chassis and each second chassis according to the layout of the server, and further determine the distance between the first chassis and each second chassis. Based on the distance between the first chassis and each second chassis, calculate the second threshold as the lower limit of the length of the second cable. Finally, based on the upper and lower limits of the length of the second cable, determine the specific length of the second cable.

[0130] As an implementation manner of the present application, the first chassis and the at least one second chassis are arranged in an upper and lower manner;

[0131] When the number of the second chassis is not less than 2, the first chassis is located between the two second chassis.

[0132] In one embodiment, the server is placed in a cabinet, wherein the first chassis and the second chassis are placed one above the other. In actual applications, the first chassis and the second chassis may be from different suppliers, so the positions of the sockets of the external cable (the second cable) are different. The following example is used for illustration.

[0133] This embodiment shows four different layouts of the first chassis and the second chassis that can be achieved by the present application scheme. Specifically, Figure 6 (a) is one of the layout diagrams of the first chassis and the second chassis of the server; Figure 6 (b) is the second layout diagram of the first chassis and the second chassis of the server; Figure 6 (c) is the third layout diagram of the first chassis and the second chassis of the server; Figure 6 (d) is the fourth layout diagram of the first chassis and the second chassis of the server. Among them, Figure 6 (a) shows the layout method used in the relevant technology, in which the first chassis with a smaller height is placed above the second chassis with a larger height, the sockets of the first chassis and the second chassis are both deployed on the top of the chassis, the sockets for plugging in the external second cables on the two chassis are close to each other, and the required length of the second cable is relatively short.

[0134] Compared with the layout shown in FIG6(a), the layouts shown in FIG6(b), FIG6(c), and FIG6(d) are as follows:

[0135] In FIG6 (b), the second chassis with a larger height is placed above the first chassis with a smaller height, and the sockets of the first chassis and the second chassis are both deployed on the top of the chassis;

[0136] In FIG6 (c), the first chassis with a smaller height is placed above the second chassis with a larger height, the socket of the first chassis is deployed at the top of the chassis, and the socket of the second chassis is deployed at a lower position in the chassis;

[0137] In Figure 6 (d), the first chassis with a smaller height is placed above two second chassis with a larger height. The sockets of the first chassis and the second chassis are both deployed on the top of the chassis. If there are more second chassis, they are placed above the current topmost second chassis or below the bottommost second chassis.

[0138] Since the sockets on the chassis in Figures 6 (b), 6 (c), and 6 (d) are located far away, their layout requires a longer second cable. Relay devices such as retimers and passive cards are used in related technologies, but the above three layouts cannot be achieved due to the limitation of the overall insertion loss of the link. The cable module provided in the present application can reduce the insertion loss of the link, and thus a longer second cable can be used, so that the server can adapt to the application scenarios shown in Figures 6 (b), 6 (c), and 6 (d) above.

[0139] Based on the same inventive concept, an embodiment of the present application provides a computer program product, including a computer program. When the computer program is executed by a processor, the steps in the server configuration method described in any of the above embodiments of the present application are implemented.

[0140] Based on the same inventive concept, an embodiment of the present application provides a readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the server configuration method described in any of the above embodiments of the present application are implemented.

[0141] Based on the same inventive concept, an embodiment of the present application provides an electronic device. Figure 7 Schematic diagram of an electronic device according to an embodiment of the present application. Figure 7 As shown, the electronic device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the server configuration method described in any of the above embodiments of the present application are implemented.

[0142] Regarding the product in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0143] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0144] For the method embodiments, for the sake of simplicity, they are all described as a series of action combinations, but those skilled in the art should be aware that the present application is not limited by the order of the actions described, because according to the present application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in the specification are all preferred embodiments, and the actions and components involved are not necessarily required by the present application.

[0145] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, devices, or computer program products. Therefore, the embodiments of the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the embodiments of the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0146] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, terminal devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0147] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0148] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0149] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they are aware of the basic creative concepts. Therefore, the present application is interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present application.

[0150] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements.

[0151] The cable module, server, server configuration method and product provided by the present application are introduced in detail above. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A server, characterized in that: include: A first chassis and a second chassis, the first chassis communicating with the second chassis via a second cable; A central processing unit and a cable module are deployed in the first chassis; The cable module replaces the passive card and is used as a relay device in the server; wherein the central processor is deployed on the mainboard and is plugged with the second connector of the cable module through the second interface on the mainboard; A graphics processor and an adapter card are arranged in the second chassis; wherein the graphics processor is connected to the adapter card via a high-speed connector; and the adapter card is connected to the cable module via the second cable; The cable module includes: a board end and a first cable; two sides of the board end form an electrical connection, wherein a first connector is fixed to one side of the board end, and a welding surface is provided on the other side of the board end; one end of the first cable is welded on the welding surface; a second connector is fixed to the other end of the first cable; the first connector is connected to the second cable, and connected to the graphics processor through the second cable; the second connector is connected to the central processing unit.

2. The server according to claim 1, characterized in that: The welding surface is arranged in a projection area of ​​the first connector on the board end; At least one via hole is provided on the board end, and the via hole is provided within a projection area of ​​the welding surface on the board end; The two sides of the board end are electrically connected through the via hole.

3. The server according to claim 1, characterized in that: The adapter board is provided with a first interface; The two ends of the second cable are fixed with first connectors, which are respectively connected to the first interface on the adapter card and the first connector of the cable module.

4. The server according to claim 1, characterized in that: The length of the second cable is not greater than a first threshold; The first threshold is determined based on an insertion loss of a link between the central processing unit and the graphics processing unit.

5. The server according to claim 4, characterized in that: The length of the second cable is not less than a second threshold; the second threshold is determined based on a first position where the first chassis is located and a second position where the second chassis is located.

6. The server according to claim 1, characterized in that: The second cable is a fifth-generation high-speed serial computer expansion bus; the adapter board is a high-speed serial computer expansion bus standard board.

7. The server according to claim 1, characterized in that: The server comprises: a first chassis and a second chassis; The adapter board includes at least one upstream interface; the adapter board is connected to a corresponding number of second cables via the at least one upstream interface; The first chassis includes at least one cable module, each cable module is connected to a second cable; The central processing unit is connected to a cable module respectively through a plurality of second interfaces on the mainboard.

8. The server according to claim 1, characterized in that: The server comprises: one first chassis and at least two second chassis; Each adapter board includes at least one upstream interface; each adapter board is connected to a corresponding number of second cables through at least one upstream interface; The first chassis includes at least two cable modules, each cable module is connected to a second cable; The central processing unit is connected to a cable module respectively through a plurality of second interfaces on the mainboard.

9. The server according to claim 7 or 8, characterized in that: In the server, the number of second cables is not greater than the maximum number of cable modules that can be connected to the central processing unit.

10. A server configuration method, characterized in that: include: Based on the first chassis and the second chassis as described in any one of claims 1 to 9, configuring a target server; The target server comprises: a first chassis and at least one second chassis; Detecting and obtaining the insertion loss of the cable module in the first chassis; Based on the insertion loss of the cable module, a first threshold is calculated as an upper limit of the length of a second cable; the second cable is used to connect the first chassis and the second chassis; Calculate a second threshold based on the distance between the first chassis and the second chassis as a lower limit of the length of the second cable; Based on the first threshold and the second threshold, a length of the second cable is determined.

11. The server configuration method according to claim 10, characterized in that: The first chassis and the at least one second chassis are arranged in an up-and-down manner; When the number of the second chassis is not less than 2, the first chassis is located between the two second chassis.

12. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps in the method of claim 10 or 11 are implemented.

13. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps in the method of claim 10 or 11 are implemented.

14. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps in the method according to claim 10 or 11 are implemented.

Images