Configuration management method and device for physical layer retimer and storage medium
The method of configuring multiple physical layer retimers in parallel through a multi-port controller solves the problem of long and low efficiency in configuration management in the prior art, and realizes efficient configuration management.
Patent Information
- Application Number
- CN202411943531.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-26
- Publication Date
- 2025-03-14
- Estimated Expiration
- 2044-12-26
AI Technical Summary
In the prior art, when configuring and managing multiple physical layer retimers, it takes a long time and is inefficient.
Connecting the substrate management controller through a multi-port controller, obtain the configuration information of each powered-up physical layer retimer, and write this information to the corresponding retimer registers to realize parallel configuration of multiple physical layer retimers.
This method can significantly shorten the configuration time and improve configuration efficiency, and is suitable for high-bandwidth interconnection scenarios of multi-GPU server nodes.
Smart Images

Figure CN119621193A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of cloud computing technology, and in particular to a configuration management method, device and computer-readable storage medium for a physical layer retimer. Background Art
[0002] In the related art, GPU (Graphic Processing Unit) server nodes are interconnected with switches via Ethernet through GPU to achieve external expansion of GPU server nodes. Since the bandwidth of the Ethernet interface for communication between GPU server nodes and switch nodes is large, the higher the link loss requirement, the GPU can be interconnected with switch nodes through PHY (Port Physical Layer) retimer, and the signal is reconstructed and forwarded through PHYretimer to complete the interconnection expansion of multi-GPU server clusters.
[0003] Since a GPU server node has multiple GPUs, external expansion also requires multiple PHY retimers. Currently, the configuration and management of PHY retimers is time-consuming and inefficient.
[0004] Therefore, how to efficiently configure and manage multiple PHY retimers in a server node is a problem that those skilled in the art need to solve. Summary of the invention
[0005] The purpose of the embodiments of the present invention is to provide a configuration management method, device and computer-readable storage medium for a physical layer retimer, which can improve the configuration efficiency of the physical layer retimer.
[0006] To solve the above technical problems, an embodiment of the present invention provides a configuration management method for a physical layer retimer, which is applied to a multi-port controller, wherein an input end of the multi-port controller is connected to an output end of a baseboard management controller, the multi-port controller includes a plurality of first communication ports and a plurality of second communication ports, each first communication port is used to connect a first end of a physical layer retimer, each second communication port is used to connect a first end of an optical module, a second end of the physical layer retimer is connected to a graphics processor, a third end of the physical layer retimer is linked to a second end of the optical module, and a fourth end of the physical layer timer is connected to a flash memory, and the method includes:
[0007] Obtain configuration information corresponding to each powered-on physical layer retimer;
[0008] Each configuration information is written into a retimer register corresponding to each powered-on physical layer retimer to configure the corresponding powered-on physical layer retimer.
[0009] In some embodiments, it also includes:
[0010] After the corresponding powered-on physical layer retimer is configured, register information of a retimer register corresponding to the configured physical layer retimer is obtained and / or register information of an optical module register corresponding to the optical module is obtained;
[0011] The register information of the retimer register and / or the register information of the optical module register is fed back to the baseboard management controller.
[0012] In some embodiments, obtaining configuration information corresponding to each powered-on physical layer retimer includes:
[0013] Obtaining identification information uniquely corresponding to each powered-on physical layer retimer;
[0014] For each piece of identification information, searching in the register of the multi-port controller according to the identification information whether there is configuration information corresponding to the identification information;
[0015] If configuration information corresponding to the identification information exists in the register of the multi-port controller, acquiring the configuration information;
[0016] When there is no configuration information corresponding to the identification information in the register of the multi-port controller, sending unsuccessful configuration feedback information to the baseboard management controller; so that the baseboard management controller determines the configuration information corresponding to the powered-on physical layer retimer according to the unsuccessful configuration feedback information; the unsuccessful configuration feedback information includes the identification information of the physical layer retimer and the unsuccessful configuration flag;
[0017] Get configuration information sent by the base management controller corresponding to the powered-up physical layer retimer.
[0018] In some embodiments, after acquiring the configuration information corresponding to the powered-on physical layer retimer sent by the basic management controller, the method further includes:
[0019] The configuration information corresponding to the powered-on physical layer retimer and the identification information of the powered-on physical layer retimer are correspondingly stored in a register of the multi-port controller.
[0020] In some embodiments, the configuration information includes one or more of an operating mode, a transmission rate, a bandwidth, and interface information of the physical layer retimer.
[0021] In some embodiments, the multi-port controller is a programmable logic device.
[0022] In some embodiments, it also includes:
[0023] Obtaining configuration change information sent by the baseboard management control; wherein the configuration change information includes identification information of the physical layer retimer and new configuration information;
[0024] Determine a physical layer retimer to be changed based on identification information in the configuration change information;
[0025] The new configuration information is written into the retimer register corresponding to the physical layer retimer to be changed, so as to reconfigure the physical layer retimer to be changed.
[0026] In some embodiments, it also includes:
[0027] The new configuration information is stored in the register of the multi-port controller, and the historical configuration information of the physical layer retimer to be changed is overwritten.
[0028] Another aspect of the embodiment of the present invention provides a configuration management device for a physical layer retimer, which is applied to a multi-port controller, wherein an input end of the multi-port controller is connected to an output end of a baseboard management controller, the multi-port controller includes a plurality of first communication ports and a plurality of second communication ports, each first communication port is used to connect to a first end of a physical layer retimer, each second communication port is used to connect to a first end of an optical module, a second end of the physical layer retimer is connected to a graphics processor, a third end of the physical layer retimer is linked to a second end of the optical module, and a fourth end of the physical layer timer is connected to a flash memory, and the device includes:
[0029] A first acquisition module is used to acquire configuration information corresponding to each powered-on physical layer retimer;
[0030] The configuration module is used to write various configuration information into the retimer register corresponding to each powered-on physical layer retimer, so as to configure the corresponding powered-on physical layer retimer.
[0031] On the other hand, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the configuration management method of the physical layer retimer as described above are implemented.
[0032] It can be seen from the above technical solution that the beneficial effects of the present invention are:
[0033] A configuration management method for a physical layer retimer is provided in an embodiment of the present invention, which is applied to a multi-port controller, wherein an input end of the multi-port controller is connected to an output end of a baseboard management controller, the multi-port controller includes multiple first communication ports and multiple second communication ports, each first communication port is used to connect to a first end of a physical layer retimer, each second communication port is used to connect to a first end of an optical module, the second end of the physical layer retimer is connected to a corresponding graphics processor, the third end of the physical layer retimer is linked to the second end of the optical module, and the fourth end of the physical layer timer is connected to a flash memory. The method includes: obtaining configuration information corresponding to each powered-on physical layer retimer; writing each configuration information into a retimer register corresponding to each powered-on physical layer retimer to configure the corresponding powered-on physical layer retimer.
[0034] It can be seen that the input end of the multi-port controller in the embodiment of the present invention is connected to the output end of the baseboard management controller, and the multi-port controller has multiple first communication ports for connecting each physical layer retimer and multiple second communication ports for connecting each optical module. The multi-port controller can determine each powered-on physical layer retimer through each first communication port, and obtain the configuration information corresponding to each powered-on physical layer retimer, and then write the configuration information into the retimer register corresponding to the powered-on physical layer retimer, thereby realizing the configuration of the powered-on physical layer retimer. Compared with the method in the related art that only one physical layer retimer can be accessed at the same time, the embodiment of the present invention can realize the configuration of multiple powered-on physical layer retimers in parallel through the multi-port controller, shortening the configuration time and improving the configuration efficiency.
[0035] In addition, the present invention also provides a corresponding implementation device and computer-readable storage medium for the configuration management method of the physical layer retimer, which further makes the method more practical, and the device, electronic device and computer-readable storage medium have corresponding advantages. BRIEF DESCRIPTION OF THE DRAWINGS
[0036] In order to more clearly illustrate the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0037] Figure 1 A schematic diagram of an external expansion interconnection relationship of a GPU server node in the related technology;
[0038] Figure 2 A schematic diagram of another GPU server node external expansion interconnection relationship in the related art;
[0039] Figure 3 A schematic diagram of a configuration management architecture of a physical layer retimer in the related art;
[0040] Figure 4 A flowchart of a configuration management method for a physical layer retimer provided by an embodiment of the present invention;
[0041] Figure 5 A schematic diagram of a configuration management architecture of a physical layer retimer provided in an embodiment of the present invention;
[0042] Figure 6 A flowchart of another configuration management method of a physical layer retimer provided by an embodiment of the present invention;
[0043] Figure 7 A flow chart of configuration change of a physical layer retimer provided in an embodiment of the present invention;
[0044] Figure 8 A structural diagram of a configuration management device for a physical layer retimer provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0046] The terms "including" and "having" in the specification of the present invention and the above-mentioned drawings, as well as any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device including a series of steps or units is not limited to the listed steps or units, but may include steps or units that are not listed.
[0047] In order to enable those skilled in the art to better understand the solution of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific implementation methods.
[0048] It should be noted that with the rapid development of cloud computing applications, more and more people need to communicate through the Internet in their daily work and life, and the amount of network data is growing explosively. Servers, as core devices for processing and storing data, have increasingly higher requirements for performance and stability.
[0049] As a type of computer, the server has high-speed CPU (Central Processing Unit) computing power, powerful I / O (Input / Output) external data throughput capacity and better scalability. As the core component of the server, the CPU has also brought new changes to the server's architectural design. In recent years, AI technology has grown explosively, and matching AI applications have also emerged. Therefore, it is necessary to build a new AI data center. The construction of a new AI data center requires a large number of AI servers or GPU servers to cope with various AI reasoning and training application scenarios. Due to the geometric growth of the number of parameters in large AI models, the computing power of a single GPU server can no longer meet actual business needs, so GPU cluster interconnection has become a new technology trend in data centers.
[0050] In related technologies, the interconnection between GPU server nodes is mainly achieved through network cards, such as Figure 1 The diagram shows the external expansion interconnection relationship of the GPU server node. The CPU is interconnected through the PCIe bus and the PCIe (Peripheral Component Interconnect Express, a high-speed serial computer expansion bus standard) switch. The GPU and NIC (Network Interface Card) are also connected to the PCIe switch. Figure 1 The link diagram of a GPU is shown in Figure 1. Multiple GPUs together form a GPU server node. The external expansion interconnection of the GPU server node is realized through a network card. The network card is connected to the PCIe interface uplink, and is connected to the switch downlink through Ethernet. The physical layer interface selects the unavailable interface according to the unavailable external optical module or copper cable. This interconnection method is limited by the bandwidth of the network card, which limits the interconnection bandwidth between GPU nodes. At the same time, the data must also be converted and output from the network card, resulting in higher latency.
[0051] Figure 2 This is a diagram of the external expansion interconnection relationship of the new generation of GPU server nodes based on the UBB (Ultimate Bulletin Board) specification. The external expansion (i.e., scale-out) part of the GPU server is not interconnected through the NIC network card, but directly interconnected with the switch through Ethernet through the GPU itself. Since the Ethernet interface for communication between the GPU server node and the switch node has reached 400Gbps or even 800Gbps, and the single lane rate has reached 56Gbps or 100Gbps, the link loss requirement is relatively high, so it is necessary to connect the GPU and the switch through the PHY retimer to achieve signal reconstruction and forwarding, and complete the interconnection expansion of the multi-GPU server cluster.
[0052] That is to say, in the current GPU server external expansion architecture, the network card is used to expand the GPU server node, which is limited by the bandwidth of the network card PCIe interface, and the overall bandwidth is low and the latency is high. In order to improve the SI performance of the link and compensate for the link loss, PHY retimer can be used in the link. A complete GPU server node has multiple GPUs, and external expansion also requires multiple PHY retimers. Therefore, how to configure and manage multiple PHY retimers is a problem that needs to be solved.
[0053] like Figure 3 As shown, in the related technology, the PHY retimer is configured through BMC management, and the BMC is connected to the PHY retimer through a communication bus. Then, the external interface is connected through another interface to obtain the module information of the optical module. However, this management method is only applicable to scenarios with a small number of PHY retimers. For GPU server nodes with multiple GPUs, multiple PHY retimers are required for external expansion, and the configuration of each PHY retimer may not be available according to the actual link. The BMC needs to use multiple MUXs (multiplexers) or I2C (Inter-Integrated Circuit, a simple, bidirectional two-wire synchronous serial bus) switches to expand multiple channels to access devices. Even if multiple PHY retimers can be connected to the bus by setting different addresses, only one PHY retimer can be accessed at the same time. It takes a long time to configure all PHY retimers, and the configuration efficiency is low.
[0054] In view of this, an embodiment of the present invention provides a configuration management method for a physical layer retimer that can improve the efficiency of PHY retimer configuration management. Next, a configuration management method for a physical layer retimer provided by an embodiment of the present invention is described in detail. Figure 4 A flowchart of a configuration management method for a physical layer retimer provided in an embodiment of the present invention, the method is applied to a multi-port controller, the input end of the multi-port controller is connected to the output end of the baseboard management controller, the multi-port controller includes multiple first communication ports and multiple second communication ports, each first communication port is used to connect to a first end of a physical layer retimer, each second communication port is used to connect to a first end of an optical module, the second end of the physical layer retimer is connected to a graphics processor, the third end of the physical layer retimer is linked to the second end of the optical module, and the fourth end of the physical layer timer is connected to a flash memory.
[0055] It should be noted that if Figure 5As shown, in the embodiment of the present invention, the output end of the baseboard management controller is connected to the input end of the multi-port controller, wherein the multi-port controller can be a programmable logic device, for example, it can be implemented by CPLD (Complex Programmable Logic Device), and the multi-port controller can expand multiple first communication ports (for example, MDIO (Management Data Input / Output, serial communication bus) interface) and multiple second communication ports (for example, I2C (Inter-Integrated Circuit, two-wire serial bus) interface), each first communication port is connected to a physical layer retimer PHY retimer, and each second communication port is connected to an optical module (for example, OSFP (Octal Small Form-Factor Pluggable, optical module)), the optical module can be used to connect to the switch node, specifically, it can be connected to the OSFP interface in the switch node through the OSFP interface, so that the server node is connected to the switch node, of course, the optical module can also be connected through other interfaces (such as QSFP (Quad Small Form-factor Pluggable, four-channel SPF interface), QSFP-DD (Quad Small Form-Factor Pluggable DoubleDensity, quad-small form factor pluggable double-density interface)) is connected to the switch node, and the GPU and the physical layer of the PHY retimer, as well as the physical layer of the PHY retimer and the optical module OSFP can be connected through Serdes (serial-to-parallel converter). Among them, the BMC and CPLD can be interconnected through I2C or MDIO bus, or through higher-speed bus interfaces such as SPI. The CPLD can quickly and efficiently complete the configuration of the PHY retimer through multiple MDIO interfaces, avoiding the large timing difference between the CPU node and the GPU node, and the CPU node has been turned on and the GPU node configures the PHY retimer in turn, which causes a long waiting time.
[0056] Based on the configuration management architecture diagram of the physical layer retimer in the embodiment of the present invention, the method can be implemented by a multi-port controller, including:
[0057] S110: Obtain configuration information corresponding to each powered-on physical layer retimer;
[0058] It should be noted that, in the embodiment of the present invention, the multi-port controller can determine the powered-on physical layer retimer from each physical layer retimer PHY retimer connected to it, and further obtain the configuration information corresponding to each powered-on physical layer retimer.
[0059] It can be understood that in actual applications, after determining that the BMC is started, the multi-port controller can control the power-on of each physical layer retimer connected to it, and each physical layer retimer will generate a power-on feedback signal, wherein, when the mark in the power-on feedback signal is 1, it indicates that the corresponding physical layer retimer has been successfully powered on, and when the mark in the power-on feedback signal is 0, it indicates that the corresponding physical layer retimer has not been successfully powered on. Therefore, the physical layer retimer that has been powered on can be determined by identifying the value of the mark bit in the power-on feedback signal corresponding to each physical layer retimer.
[0060] In actual applications, the configuration information corresponding to each physical layer retimer can be stored in the register corresponding to the multi-port controller in advance. After determining each powered-on physical layer retimer, the configuration information corresponding to each powered-on physical layer retimer can be further determined in the register of the multi-port controller.
[0061] S120: Write each configuration information into the retimer register corresponding to each powered-on physical layer retimer to configure the corresponding powered-on physical layer retimer.
[0062] It should be noted that a retimer register is configured for each physical layer retimer in advance. For each physical layer retimer that has been powered on, after obtaining the configuration information corresponding to the physical layer retimer, the configuration information can be written into the retimer register of the corresponding physical layer retimer that has been powered on. In practical applications, each time a configuration information is obtained, the configuration information can be written into the corresponding retimer register; or after obtaining all the configuration information for each physical layer retimer that has been powered on, each configuration information can be written into the retimer register of the corresponding physical layer retimer that has been powered on and take effect, thereby completing the configuration of the corresponding physical layer retimer that has been powered on, so that the Ethernet link from the GPU to the switch is successfully connected. The specific time of writing is not particularly limited in the embodiment of the present invention, and can be determined according to actual needs.
[0063] In addition, the configuration information in the embodiment of the present invention may include one or more of the working mode, transmission rate, bandwidth and interface information of the physical layer retimer, so that the working mode, transmission rate, bandwidth or interface of the physical layer retimer is configured through the configuration information.
[0064] It should be noted that the physical layer retimer PHY retimer works at the physical layer. Its main function is to compensate for the Ethernet link loss so that the signal can be transmitted over a longer distance. In addition, it can also be used as a rate converter, such as converting the 56Gbps signal of the GPU node into a 112Gbps signal to interconnect with the switch. That is, there are two working modes, one is the retimer mode (retimer mode) and the other is the gearbox mode (gearbox mode). The specific working mode of the PHY retimer can be determined by the GPU model, the switch interface rate, and the system architecture, etc., and the embodiments of the present invention do not make special limitations on this.
[0065] In some embodiments, the method may further include:
[0066] After the corresponding powered-on physical layer retimer is configured, register information of a retimer register corresponding to the configured physical layer retimer is obtained and / or register information of an optical module register corresponding to the optical module is obtained;
[0067] The register information of the retimer register and / or the register information of the optical module register is fed back to the baseboard management controller.
[0068] It should be noted that in actual applications, after the configuration of a certain powered-on physical layer retimer is completed, configuration feedback information corresponding to the powered-on physical layer retimer will be generated, and the configuration feedback information will be sent to the BMC, wherein the configuration feedback information includes the identification information of the powered-on physical layer retimer and the configuration completion mark. If the configuration completion mark is 1, it means that the configuration is completed, and if the configuration completion mark is 0, it means that the configuration is not completed, so that the BMC can timely understand the configuration of each powered-on physical layer retimer. In addition, during the operation of the server node, for each powered-on physical layer retimer that has been configured, the multi-port controller can obtain the register information of the retimer register corresponding to the physical layer retimer through each MDIO interface and each configured physical layer retimer, and report the register information of each retimer register to the BMC for processing. That is, the multi-port controller can simultaneously capture the register information of the retimer registers of multiple physical layer retimers, thereby obtaining the status information of multiple physical layer retimers, and sending it to the BMC for processing, so as to improve the management efficiency of the BMC for each physical layer retimer.
[0069] Of course, the multi-port controller can also obtain the register information of the optical module register of each optical module through each I2C bus, and report the register information of each optical module register to the BMC for processing. The embodiment of the present invention can batch obtain the register information of the retimer registers of multiple physical layer retimers, and batch obtain the register information of the optical module register of each optical module, thereby improving the management efficiency of the physical layer retimer and / or optical module.
[0070] In some embodiments, the process of obtaining configuration information corresponding to each powered-on physical layer retimer in S110 may include:
[0071] Obtaining identification information uniquely corresponding to each powered-on physical layer retimer;
[0072] For each piece of identification information, searching in the register of the multi-port controller according to the identification information whether there is configuration information corresponding to the identification information;
[0073] If configuration information corresponding to the identification information exists in the register of the multi-port controller, acquiring the configuration information;
[0074] When there is no configuration information corresponding to the identification information in the register of the multi-port controller, sending unsuccessful configuration feedback information to the baseboard management controller; so that the baseboard management controller determines the configuration information corresponding to the powered-on physical layer retimer according to the unsuccessful configuration feedback information; the unsuccessful configuration feedback information includes the identification information of the physical layer retimer and the unsuccessful configuration flag;
[0075] Get configuration information sent by the base management controller corresponding to the powered-up physical layer retimer.
[0076] It should be noted that if Figure 6As shown, in the embodiment of the present invention, after the BMC is powered on, each physical layer retimer can be controlled to power on, and the power-on feedback information of each physical layer retimer is used to determine whether the physical layer retimer is powered on successfully, and the physical layer retimer that has been powered on successfully is determined, and the multi-port controller (such as CPLD) configures the physical layer retimer that has been powered on successfully. In actual applications, after obtaining the configuration information of each physical layer retimer, the BMC can send the identification information and corresponding configuration information of the physical layer retimer to the register of the multi-port controller for storage. After determining the physical layer retimer that has been powered on, the identification information (such as including the identification code) of the physical layer retimer that has been powered on successfully can be obtained first, and then the configuration information of each physical layer retimer stored in the register of the multi-port controller can be matched. In the case that the configuration information corresponding to the identification information can be matched, the configuration information is obtained, and the configuration information is written into the retimer register of the corresponding physical layer retimer that has been powered on to take effect, thereby completing the configuration of the physical layer retimer that has been powered on.
[0077] If the configuration information corresponding to the identification information of the powered-on physical layer retimer cannot be matched among the various configuration information stored in the register of the multi-port controller, the multi-port controller can send configuration failure feedback information to the BMC, wherein the configuration failure feedback information may include the identification information of the physical layer retimer and an unsuccessful configuration mark. The BMC can determine that the corresponding physical layer retimer has not been configured successfully based on the identification information and the unsuccessful configuration mark in the configuration failure feedback information. At this time, the configuration information (hardware link information and architecture information, such as working mode, transmission rate, bandwidth and interface information, etc.) corresponding to the corresponding powered-on physical layer retimer can be determined from the various stored configuration information based on the identification information, and the configuration information and the corresponding identification information are sent to the multi-port controller. After obtaining the corresponding configuration information, the multi-port controller can configure the corresponding powered-on physical layer retimer.
[0078] In actual applications, after the configuration is completed, the GPU can establish a connection with the switch through the Ethernet interface, and the BMC can collect register information of each physical layer retimer and register information of the optical module through the multi-port controller.
[0079] In some embodiments, after acquiring the configuration information corresponding to the powered-on physical layer retimer sent by the basic management controller, the method may further include:
[0080] The configuration information corresponding to the powered-on physical layer retimer and the identification information of the powered-on physical layer retimer are correspondingly stored in a register of the multi-port controller.
[0081] It should be noted that in actual applications, after the multi-port controller obtains the configuration information corresponding to the powered-on physical layer retimer sent by the BMC, in addition to using the configuration information to configure the powered-on physical layer retimer, it can also store the identification information and configuration information of the powered-on physical layer retimer in the register of the multi-port controller, so that the corresponding configuration information can be directly obtained from the register to configure the physical layer retimer when it is powered on next time, thereby improving configuration efficiency.
[0082] In some embodiments, the method may further include:
[0083] Obtaining configuration change information sent by the baseboard management control; wherein the configuration change information includes identification information of the physical layer retimer and new configuration information;
[0084] Determine a physical layer retimer to be changed based on identification information in the configuration change information;
[0085] The new configuration information is written into the retimer register corresponding to the physical layer retimer to be changed, so as to reconfigure the physical layer retimer to be changed.
[0086] Understandably, please refer to Figure 7 When the user needs to change the configuration of the physical layer retimer, the user can input the configuration change information through the BMC, and the configuration change information may include the identification information of the physical layer retimer and the new configuration information. The BMC sends the configuration change information to the multi-port controller, and the multi-port controller can determine the corresponding physical layer retimer to be changed according to the identification information in the configuration change information, and then write the new configuration information into the retimer register of the physical layer retimer to be changed at the position corresponding to the physical layer retimer to be changed, and overwrite the old configuration information of the physical layer retimer to be changed, and then control the reset of the physical layer retimer to be changed to take effect, thereby realizing the configuration change of the physical layer retimer to be changed.
[0087] In addition, after the change is completed, the new configuration information can also be stored in the register of the multi-port controller, and the historical configuration information of the physical layer retimer to be changed can be overwritten to update the configuration information in the register so that the latest configuration information can be used to configure the corresponding physical layer retimer at the next power-on.
[0088] It can be seen that the input end of the multi-port controller in the embodiment of the present invention is connected to the output end of the baseboard management controller, and the multi-port controller has multiple first communication ports for connecting each physical layer retimer and multiple second communication ports for connecting each optical module. The multi-port controller can determine each powered-on physical layer retimer through each first communication port, and obtain the configuration information corresponding to each powered-on physical layer retimer, and then write the configuration information into the retimer register corresponding to the powered-on physical layer retimer, thereby realizing the configuration of the powered-on physical layer retimer. Compared with the method in the related art that only one physical layer retimer can be accessed at the same time, the embodiment of the present invention can realize the configuration of multiple powered-on physical layer retimers in parallel through the multi-port controller, shortening the configuration time and improving the configuration efficiency.
[0089] The present invention also provides a corresponding device for the configuration management method of the physical layer retimer, which further makes the method more practical. Among them, the device can be described from the perspective of functional modules and hardware. The configuration management device of the physical layer retimer provided by the present invention is introduced below. The device is used to implement the configuration management method of the physical layer retimer provided by the present invention. In this embodiment, the configuration management device of the physical layer retimer may include or be divided into one or more program modules, and the one or more program modules are stored in a storage medium and executed by one or more processors to complete the configuration management method of the physical layer retimer disclosed in the above embodiment. The program module referred to in the present invention refers to a series of computer program instruction segments that can complete specific functions, which is more suitable for describing the execution process of the configuration management device of the physical layer retimer in the storage medium than the program itself. The following description will specifically introduce the functions of each program module of this embodiment. The configuration management device of the physical layer retimer described below and the configuration management method based on the physical layer retimer described above can be referred to each other.
[0090] From the perspective of functional modules, see Figure 8 , Figure 8 A structural diagram of a configuration management device for a physical layer retimer provided by the present invention in a specific implementation manner, the device may include:
[0091] Another aspect of the embodiment of the present invention provides a configuration management device for a physical layer retimer, which is applied to a multi-port controller. The input end of the multi-port controller is connected to the output end of the baseboard management controller. The multi-port controller includes a plurality of first communication ports and a plurality of second communication ports. Each first communication port is used to connect to a first end of a physical layer retimer, and each second communication port is used to connect to a first end of an optical module. The second end of the physical layer retimer is connected to a graphics processor, the third end of the physical layer retimer is linked to the second end of the optical module, and the fourth end of the physical layer timer is connected to a flash memory. The device includes:
[0092] The first acquisition module 11 is used to acquire configuration information corresponding to each powered-on physical layer retimer;
[0093] The configuration module 12 is used to write various configuration information into the retimer register corresponding to each powered-on physical layer retimer, so as to configure the corresponding powered-on physical layer retimer.
[0094] In some embodiments, the apparatus further comprises:
[0095] A second acquisition module is used to obtain the register information of the retimer register corresponding to the configured physical layer retimer and / or obtain the register information of the optical module register corresponding to the optical module after the corresponding powered-on physical layer retimer is configured;
[0096] The feedback module is used to feed back the register information of the retimer register and / or the register information of the optical module register to the baseboard management controller.
[0097] In some embodiments, the first acquisition module 11 includes:
[0098] A first acquisition unit, configured to acquire identification information uniquely corresponding to each powered-on physical layer retimer;
[0099] A query unit, configured to search, for each piece of identification information, a register of the multi-port controller for configuration information corresponding to the identification information;
[0100] A second acquiring unit, configured to acquire the configuration information if the configuration information corresponding to the identification information exists in the register of the multi-port controller;
[0101] A feedback unit is used to send unsuccessful configuration feedback information to a baseboard management controller when there is no configuration information corresponding to the identification information in the register of the multi-port controller; so that the baseboard management controller determines the configuration information corresponding to the powered-on physical layer retimer according to the unsuccessful configuration feedback information; the unsuccessful configuration feedback information includes the identification information of the physical layer retimer and the unsuccessful configuration flag;
[0102] The third acquisition unit is used to acquire the configuration information corresponding to the powered-on physical layer retimer and sent by the basic management controller.
[0103] In some embodiments, the apparatus further comprises:
[0104] The first storage module is used to store configuration information corresponding to the powered-on physical layer retimer and identification information of the powered-on physical layer retimer in a register of the multi-port controller.
[0105] In some embodiments, the configuration information includes one or more of an operating mode, a transmission rate, a bandwidth, and interface information of the physical layer retimer.
[0106] In some embodiments, the multi-port controller is a programmable logic device.
[0107] In some embodiments, the apparatus further comprises:
[0108] A third acquisition module is used to acquire configuration change information sent by the baseboard management control; wherein the configuration change information includes identification information of the physical layer retimer and new configuration information;
[0109] A change module, used for determining a physical layer retimer to be changed based on identification information in the configuration change information;
[0110] The writing module is used to write new configuration information into the retimer register corresponding to the physical layer retimer to be changed, so as to reconfigure the physical layer retimer to be changed.
[0111] In some embodiments, the apparatus further comprises:
[0112] The second storage module is used to store the new configuration information in the register of the multi-port controller and overwrite the historical configuration information of the physical layer retimer to be changed.
[0113] It should be noted that the configuration management device of the physical layer retimer provided in the embodiment of the present invention has the same beneficial effects as the configuration management method of the physical layer retimer provided in the above-mentioned embodiment, and for the specific introduction of the configuration management method of the physical layer retimer involved in the embodiment of the present invention, please refer to the above-mentioned embodiment, and the present invention will not be repeated here.
[0114] Based on the above embodiments, an embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the configuration management method of the physical layer retimer as described above are implemented.
[0115] Based on the above embodiments, an embodiment of the present invention further provides a computer program product, including a computer program / instruction, which implements the steps of the above-mentioned physical layer retimer configuration management method when executed by a processor.
[0116] The above is a detailed introduction to a configuration management method, device and computer-readable storage medium for a physical layer retimer provided in an embodiment of the present invention. The various embodiments in the specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same and similar parts between the various embodiments can be referenced to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.
[0117] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.
[0118] The above is a detailed introduction to the configuration management method, device and computer-readable storage medium of a physical layer retimer provided by the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea. It should be pointed out that for ordinary technicians in this technical field, without departing from the principles of the present invention, several improvements and modifications can be made to the present invention, and these improvements and modifications also fall within the scope of protection of the claims of the present invention.
Claims
1. A configuration management method for a physical layer retimer, characterized in that: Applied to a multi-port controller, wherein the input end of the multi-port controller is connected to the output end of the baseboard management controller, the multi-port controller includes a plurality of first communication ports and a plurality of second communication ports, each of the first communication ports is used to connect to a first end of a physical layer retimer, each of the second communication ports is used to connect to a first end of an optical module, the second end of the physical layer retimer is connected to a corresponding graphics processor, the third end of the physical layer retimer is linked to the second end of the optical module, the fourth end of the physical layer timer is connected to a flash memory, and the method includes: Obtain configuration information corresponding to each powered-on physical layer retimer; Each of the configuration information is written into a retimer register corresponding to each of the powered-on physical layer retimers to configure the corresponding powered-on physical layer retimer.
2. The configuration management method of the physical layer retimer according to claim 1, characterized in that: Also includes: After the corresponding powered-on physical layer retimer is configured, register information of a retimer register corresponding to the configured physical layer retimer is obtained and / or register information of an optical module register corresponding to the optical module is obtained; The register information of the retimer register and / or the register information of the optical module register is fed back to the baseboard management controller.
3. The configuration management method of the physical layer retimer according to claim 1, characterized in that: Get the configuration information of each powered-on physical layer retimer, including: Acquire identification information uniquely corresponding to each of the powered-on physical layer retimers; For each piece of identification information, searching in a register of the multi-port controller according to the identification information whether there is configuration information corresponding to the identification information; If configuration information corresponding to the identification information exists in a register of the multi-port controller, acquiring the configuration information; In the case that the configuration information corresponding to the identification information does not exist in the register of the multi-port controller, sending configuration failure feedback information to the baseboard management controller; so that the baseboard management controller determines the configuration information corresponding to the powered-on physical layer retimer according to the configuration failure feedback information; the configuration failure feedback information includes the identification information of the physical layer retimer and the unsuccessful configuration flag; Acquire configuration information sent by the basic management controller and corresponding to the powered-on physical layer retimer.
4. The configuration management method of the physical layer retimer according to claim 3, characterized in that: After acquiring the configuration information corresponding to the powered-on physical layer retimer sent by the basic management controller, the method further includes: The configuration information corresponding to the powered-on physical layer retimer and the identification information of the powered-on physical layer retimer are correspondingly stored in the register of the multi-port controller.
5. The configuration management method of the physical layer retimer according to claim 1, characterized in that: The configuration information includes one or more of an operating mode, a transmission rate, a bandwidth, and interface information of a physical layer retimer.
6. The configuration management method of the physical layer retimer according to claim 1, characterized in that: The multi-port controller is a programmable logic device.
7. The configuration management method of a physical layer retimer according to any one of claims 1 to 6, characterized in that: Also includes: Acquire configuration change information sent by the baseboard management control; wherein the configuration change information includes identification information of the physical layer retimer and new configuration information; Determine a physical layer retimer to be changed based on identification information in the configuration change information; The new configuration information is written into a retimer register corresponding to the physical layer retimer to be changed, so as to reconfigure the physical layer retimer to be changed.
8. The configuration management method of the physical layer retimer according to claim 7, characterized in that: Also includes: The new configuration information is stored in a register of the multi-port controller, and the historical configuration information of the physical layer retimer to be changed is overwritten.
9. A configuration management device for a physical layer retimer, characterized in that: Applied to a multi-port controller, wherein the input end of the multi-port controller is connected to the output end of the baseboard management controller, the multi-port controller includes a plurality of first communication ports and a plurality of second communication ports, each of the first communication ports is used to connect to a first end of a physical layer retimer, each of the second communication ports is used to connect to a first end of an optical module, the second end of the physical layer retimer is connected to a corresponding graphics processor, the third end of the physical layer retimer is linked to the second end of the optical module, the fourth end of the physical layer timer is connected to a flash memory, and the device includes: A first acquisition module is used to acquire configuration information corresponding to each powered-on physical layer retimer; The configuration module is used to write each of the configuration information into the retimer register corresponding to each of the powered-on physical layer retimers, so as to configure the corresponding powered-on physical layer retimer.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the configuration management method of the physical layer retimer according to any one of claims 1 to 7 are implemented.
Citation Information
Patent Citations
Method, system and device for setting universal substrate scale out and storage medium
CN115509311A
Retimer mode control method, system and device, equipment and storage medium
CN115858433A
PCIe (Peripheral Component Interconnect Express) bandwidth configuration system and method for retimer chip
CN117234603A
Method and system for matching retimer with downlink equipment
CN117472833A
Firmware configuration updating method, device and system, electronic equipment and storage medium
CN117687645A
Cited By
Retimer configuration method and electronic equipment
CN122160249A
A method for configuring a re-timer and an electronic device
CN122160249B