Network card control method, system and electronic device

Abstract

This invention provides a network interface card (NIC) control method, system, and electronic device, relating to the field of NIC control. The method does not immediately start the link training process after the automatic negotiation process is completed. Instead, it first divides the NIC and the target host into tasks according to the target scheduling strategy to achieve dynamic task allocation negotiation. Based on the task decision instructions corresponding to the negotiation results, it determines the triggering time and working mode of the link training process, so that the link power consumption is dynamically matched with the actual task load, thereby reducing the NIC power consumption.

Description

Technical Field

[0001] This invention relates to the field of network interface card (NIC) control, and in particular to a NIC control method, system, and electronic device. Background Technology

[0002] Current 200G / 400G and higher speed Ethernet interfaces (such as IEEE 802.3) follow a standard process of Auto-Negotiation (AN) - Link Training (LT) - Normal Service Transmission. The link training process uses interactive training sequences to adjust parameters such as SerDes (SERializer / DESerializer) and CDR (Clock and Data Recovery) to ensure high-speed link stability.

[0003] In existing technologies, the link training process is entirely driven by physical layer events. Specifically, the AN / LT process is triggered when the device powers on and the port is physically connected, when the connection state changes during operation (plugging / unplugging, LOS, loss of lock, etc.), or when the bit error rate increases or the timing period arrives. This process is automatically executed by the PHY hardware or network card firmware without upper-layer software intervention and is not linked to the host-network card dynamic task allocation negotiation. Therefore, the existing mechanism has significant drawbacks. First, there is power waste in scenarios without services. In data centers, a large number of physical network card ports without services still execute the full AN / LT and make modules such as SerDes, FEC (Forward Error Correction), and MAC (Media Access Control) run at full power. Secondly, the link training process itself is high-power. SerDes sends and receives high-frequency training sequences at full amplitude, and frequent triggering will exacerbate power consumption and heat generation. Third, the software overhead is large when the negotiation is abnormal. When the negotiation fails or the mode rollback occurs, the link training process is still forcibly executed, which increases the processing latency. Fourth, it is not compatible with smart network cards. The link training process is strongly bound to the physical connection, and it does not take into account business needs and task division. It lacks power consumption awareness, which contradicts the on-demand offloading and energy-saving requirements of smart network cards. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a network card control method, system and electronic device. The method does not immediately start the link training process after the automatic negotiation process is completed. Instead, it first divides the network card and the target host into tasks according to the target scheduling strategy to realize dynamic task allocation negotiation. Based on the task decision instructions corresponding to the negotiation results, it determines the triggering time and working mode of the link training process, so that the link power consumption is dynamically matched with the actual task load, thereby reducing the power consumption of the network card.

[0005] In a first aspect, embodiments of the present invention provide a network interface card (NIC) control method, the method comprising: When the network card is detected to be in a physically connected state, the network card is controlled to execute the automatic negotiation process corresponding to its Ethernet communication standard set; Establish and configure the control channel between the network card and the target host based on the power consumption parameters of each functional module in the network card; Taking the unloading task in the network card as the decision target, the load data of the network card and the resource data of the target host are divided into task decision according to the target scheduling strategy through the control channel, and then the task decision instruction of the unloading task is obtained. After controlling the network card to execute the link training process corresponding to the Ethernet communication standard set using task decision instructions, the network card is controlled to execute the unloading task.

[0006] Optionally, the step of establishing and configuring a control channel between the network interface card (NIC) and the target host based on the power consumption parameters of each functional module in the NIC includes: Obtain the operating status data of the serializer module, deserializer module, forward error correction module, and media access control module in the network card; Based on the first operating status data between the serializer module and the deserializer module, the first power consumption parameter corresponding to the idle code interaction between the network card and the target host is obtained. The second power consumption parameter corresponding to the packet interaction between the network card and the target host is obtained by using the second operating status data between the forward error correction module and the media access control module. Using the first power consumption parameter and the second power consumption parameter as constraints, establish and configure the control channel between the network card and the target host.

[0007] Optionally, the steps of taking the offloading task in the network interface card (NIC) as the decision target, and dividing the NIC's load data and the target host's resource data according to the target scheduling strategy through the control channel to obtain the task decision instruction for the offloading task, include: The target scheduling strategy is determined based on the uninstallation processing strategy, pass-through processing strategy, and resource allocation strategy corresponding to the uninstallation task. The network interface card (NIC) load data is acquired in real time and transmitted to the target host via the control channel. Obtain the resource data of the target host, and control the target host according to the target scheduling policy. After making task decisions based on the resource data and load data, obtain the task negotiation result of the network card. The triggering timing and working mode of the uninstallation task are obtained through the task negotiation results, and the task decision instructions for the uninstallation task are generated using the triggering timing and working mode. The target host is controlled by the control channel to transmit task decision instructions to the network card.

[0008] Optionally, the steps of obtaining the triggering timing and working mode corresponding to the unloading task through task negotiation results, and generating task decision instructions for the unloading task using the triggering timing and working mode, include: Retrieve the host processing tasks, network card processing tasks, and negotiation failure tasks included in the task negotiation results; Based on the host processing task, determine the first working mode corresponding to the unloading task, and use the first triggering time corresponding to the target host in the first working mode to generate the first task decision instruction corresponding to the unloading task. Based on the network card processing task, determine the second working mode corresponding to the unloading task, and use the second triggering time corresponding to the network card in the second working mode to generate the second task decision instruction corresponding to the unloading task. Based on the negotiation failure task, determine the third working mode corresponding to the unloading task, determine the third triggering time corresponding to the network card through the failure handling strategy or rollback handling strategy corresponding to the third working mode, and use the third triggering time to generate the third task decision instruction corresponding to the unloading task.

[0009] Optionally, before controlling the network interface card to execute the link training process, the method further includes: If the task decision instruction is the first task decision instruction, then the control channel between the network card and the target host is updated using the current power consumption parameters of each functional module in the network card. If the task decision instruction is the third task decision instruction, then the task decision instruction for the unloading task is updated using the rollback processing strategy corresponding to the third task decision instruction.

[0010] Optionally, the task decision instructions control the network card to execute the link training process corresponding to the Ethernet communication standard set, including: The offloading task contained in the network card is determined by using the second triggering time in the second task decision instruction or the failure handling strategy in the third task decision instruction. Based on the unloading task control network card, the link training process corresponding to the Ethernet communication standard set is executed, so that the network card's corresponding equalizer coefficient parameters, clock data recovery parameters and forward error correction parameters are in the target convergence state.

[0011] Optionally, control the network card to perform offload tasks, including: The task decision command controls the serializer module, deserializer module, forward error correction module, and media access control module in the network card to operate at rated power. If an unloading task is detected within a preset idle time period and the target host has not initiated an update instruction for the task decision command, then the network card is controlled to execute the unloading task in the current working state.

[0012] Optionally, after the network interface card (NIC) performs the offload task, the method further includes: If the network interface card (NIC) does not have any update instructions for unloading tasks or for the target host to initiate task decision instructions within a preset idle time, the control channel between the NIC and the target host is updated using the current power consumption parameters of each functional module in the NIC.

[0013] Secondly, the present invention provides a network interface card (NIC) control system, the system comprising: Initialization module: Used to control the network card to execute the automatic negotiation process corresponding to its Ethernet communication standard set after detecting that the network card is in a physical connection state; Control Channel Construction Module: Used to establish and configure the control channel between the network card and the target host based on the power consumption parameters of each functional module in the network card; Task decision partitioning module: It is used to divide the load data of the network card and the resource data of the target host according to the target scheduling strategy through the control channel, taking the unloading task in the network card as the decision target, and then obtain the task decision instruction for the unloading task. Task execution control module: After controlling the network card to execute the link training process corresponding to the Ethernet communication standard set using task decision instructions, it controls the network card to execute the unloading task.

[0014] Thirdly, embodiments of the present invention also provide an electronic device, which includes a processor and a memory, wherein the memory stores computer-executable instructions that can be executed by the processor, and the processor executes the computer-executable instructions to implement the steps of the network card control method provided in the first aspect.

[0015] This invention provides a network interface card (NIC) control method, system, and electronic device. During the control of NIC link initialization behavior according to the Ethernet communication standard set, when the NIC is detected to be in a physical connection state, the method controls the NIC to execute the automatic negotiation process corresponding to its Ethernet communication standard set. Then, it uses the power consumption parameters of each functional module in the NIC to construct a control channel between the NIC and the target host. Subsequently, using the offloading task in the NIC as the decision target, the method divides the NIC's load data and the target host's resource data according to the target scheduling strategy through the control channel to obtain the task decision instruction for the offloading task. Finally, it uses the task decision instruction to control the NIC to execute the link training process corresponding to the Ethernet communication standard set, and then controls the NIC to execute the offloading task. This method does not immediately start the link training process after the automatic negotiation process is completed. Instead, it first divides the NIC and the target host according to the target scheduling strategy to achieve dynamic task allocation negotiation. Based on the task decision instruction corresponding to the negotiation result, it determines the triggering time and working mode of the link training process, enabling dynamic matching between link power consumption and actual task load, thereby reducing NIC power consumption.

[0016] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention are realized and obtained through the structures particularly pointed out in the description and the drawings.

[0017] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0018] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 A flowchart of a network interface card (NIC) control method provided in an embodiment of the present invention; Figure 2 A flowchart of another network card control method provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of a network card control system provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0020] icon: 100 - Initialization module; 200 - Control channel construction module; 300 - Task decision and partitioning module; 400 - Task execution control module; 101 - Processor; 102 - Memory; 103 - Bus; 104 - Communication interface. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] To facilitate understanding of this embodiment, a network interface card (NIC) control method disclosed in this embodiment will first be introduced. This method is as follows: Figure 1 As shown, it includes: Step S101: When the network card is detected to be in a physically connected state, control the network card to execute the automatic negotiation process corresponding to its Ethernet communication standard set.

[0023] Once the network interface card (NIC) port detects a valid physical connection (such as the establishment of a physical layer connection through optical module insertion or network cable connection), the host electronic device (embedded processor or hardware state machine) immediately triggers and controls the NIC to execute the automatic negotiation process (AN process) specified by the corresponding Ethernet communication standard set (such as the IEEE 802.3 standard). This process is completed autonomously by the NIC hardware without the need for intervention from the host upper-layer software. Its core is to negotiate basic parameters such as transmission rate (200G / 400G, etc.), FEC type, and duplex mode with the target host through lightweight interaction (low-speed pulses or DME messages) to determine a compatible basic link configuration. Unlike existing technologies, after completing the automatic negotiation step, this step does not immediately initiate the link training process (LT process). Instead, it controls the NIC to enter a task-waiting negotiation state, reserving time for subsequent dynamic task allocation.

[0024] Step S102: Establish and configure the control channel between the network card and the target host based on the power consumption parameters of each functional module in the network card.

[0025] After the auto-negotiation process is completed, a bidirectional lightweight control channel is constructed between the target host and the network card based on the real-time power consumption parameters of each functional module (SerDes, FEC, MAC, etc.). This control channel relies on the basic link configuration determined by auto-negotiation and adopts a low-power transmission mode. Specifically, it can be configured as follows: control SerDes to operate at a low amplitude, only sending idle codes or remaining silent, without performing complete training sequence interaction; at the same time, the receiving end only retains basic signal detection and low-speed packet parsing capabilities, and cannot demodulate high-speed service data; the FEC and MAC modules are switched to low-power standby state, retaining only basic packet reception and parsing capabilities.

[0026] The core purpose of building this control channel is to provide a stable, low-power interactive link for the subsequent dynamic task allocation negotiation between the target host and the network card, ensuring that the overall power consumption of the network card remains at a low level during the negotiation process and avoiding unnecessary power consumption.

[0027] Step S103: Taking the unloading task in the network card as the decision target, the load data of the network card and the resource data of the target host are divided into task decision according to the target scheduling strategy through the control channel, and the task decision instruction of the unloading task is obtained.

[0028] With the core decision-making objective being the hardware offloading tasks that the network interface card (NIC) can handle (such as RDMA remote direct memory access, data encryption, and storage virtualization), the NIC and the target host conduct bidirectional dynamic task allocation negotiation through the control channel established in step S102. Specifically, the NIC first reports its own hardware capabilities (supported offloading function types) and real-time load data (number of idle queues, bandwidth quota, computing power margin, etc.) to the target host; the target host synchronously feeds back its own resource data (CPU utilization, memory bandwidth, service load requirements, etc.). Both parties, according to the preset target scheduling strategy (the core being "on-demand allocation and optimal power consumption"), divide the data processing responsibilities, clarifying: which data streams are handled by the NIC hardware offloading, which data streams are passed through to the host CPU for processing, and the allocation ratio of resources on the NIC side (number of queues, bandwidth quota, etc.).

[0029] After multiple rounds of message interaction and iterative negotiation, a clear decision instruction for the offloading task is finally formed. This instruction includes whether link training needs to be started, the working mode of link training (full training / simplified training), and the specific execution parameters of the network card offloading task.

[0030] Step S104: After controlling the network card to execute the link training process corresponding to the Ethernet communication standard set using the task decision instruction, control the network card to execute the unloading task.

[0031] After receiving the task decision instruction generated in step S103, the network card is controlled to execute the link training process (LT process) specified in the corresponding Ethernet communication standard set according to the instruction content: If the decision instruction clearly indicates that there is a network card offload task (requiring high-speed link support), then the complete link training process is started. The network card and the target host dynamically adjust the SerDes equalizer coefficient, CDR clock data recovery parameters, FEC synchronization configuration, etc. through interactive training sequences until all parameters converge and the link signal quality meets the requirements for high-speed transmission; If the decision instruction clearly indicates that there is no offload task, then there is no need to start the complete link training, and the lightweight control channel mode is maintained to reduce the network card power consumption.

[0032] After the link training is completed (or no training is required), the control SerDes, FEC, MAC and other modules switch to the corresponding working state (if there is an offload task, it enters the full power working state, and if there is no offload task, it maintains the low power state). Then, according to the requirements of the task decision instruction, the corresponding hardware offload task is started and executed to complete the high-speed data transmission and reception and processing, and realize the dynamic matching between link power consumption and actual task load.

[0033] Optionally, step S102, which establishes and configures the control channel between the network card and the target host based on the power consumption parameters of each functional module in the network card, includes the following steps: Step S201: Obtain the operating status data of the serializer module, deserializer module, forward error correction module, and media access control module in the network card.

[0034] The host electronic device (embedded processor or hardware state machine) actively collects real-time operating status data of the core functional modules inside the network card, focusing on acquiring multi-dimensional operating condition information of the serializer, deserializer, forward error correction (FEC) module, and media access control (MAC) module. Specifically, this includes the current operating mode, signal transmission amplitude, operating clock frequency, load utilization, idle status indicator, and basic power consumption baseline value of each module. This comprehensive understanding of the operating characteristics of each underlying hardware module provides data support for subsequent power consumption parameter calculation and control channel construction.

[0035] Step S202: Based on the first operating status data between the serializer module and the deserializer module, obtain the first power consumption parameter corresponding to the idle code interaction between the network card and the target host.

[0036] Based on the collected first operating state data of the serializer and deserializer modules (collectively referred to as SerDes), and combined with the idle interaction scenario of the control channel, the first power consumption parameter corresponding to the idle code interaction or link silence state between the network card and the target host is accurately calculated and obtained. During this process, by analyzing the energy consumption characteristics of SerDes in low-amplitude operation scenarios without sending complete training sequences, its minimum stable power consumption threshold in the idle interaction phase is determined, ensuring that this power consumption parameter can maintain link connectivity while minimizing ineffective power consumption.

[0037] Step S203: Obtain the second power consumption parameter corresponding to the packet interaction between the network card and the target host through the second operating status data between the forward error correction module and the media access control module.

[0038] By combining the second operational status data of the Forward Error Correction (FEC) module and the Media Access Control (MAC) module, and considering the control message interaction scenarios required for dynamic task allocation negotiation between the host and the network interface card (NIC), the corresponding second power consumption parameters are further calculated and obtained. Specifically, the power consumption data of the FEC module in lightweight decoding mode and the MAC module in basic message parsing (processing only negotiation commands, not high-speed frame processing) mode are monitored to clarify the power consumption range during the transmission, reception, and parsing of negotiation messages between the two parties, ensuring that the power consumption during message interaction is controllable and matches the low power consumption requirements of the control channel.

[0039] Step S204: Establish and configure the control channel between the network card and the target host, using the first power consumption parameter and the second power consumption parameter as constraints.

[0040] Using the first power consumption parameter (idle interaction power consumption) obtained in step S202 and the second power consumption parameter (message interaction power consumption) obtained in step S203 as core constraints, and combined with the link basic configuration determined by the AN automatic negotiation process, the working modes of each functional module of the network card are uniformly configured. Specifically, this may include: configuring SerDes to a low-amplitude operating mode, limiting it to sending idle codes or remaining silent when idle; switching the FEC and MAC modules to a low-power standby state, retaining only basic message receiving and parsing capabilities; and optimizing the message transmission rate and interaction frequency of the control channel, ultimately forming a dedicated lightweight control channel that adapts to dynamic task negotiation between the host and the network card and operates stably with low power consumption, thus meeting the needs of negotiation message interaction while achieving optimal energy consumption control.

[0041] Optionally, step S103, which takes the offloading task in the network interface card (NIC) as the decision target and divides the NIC's load data and the target host's resource data according to the target scheduling strategy through the control channel to obtain the task decision instruction for the offloading task, includes the following steps: Step S301: Determine the target scheduling strategy based on the uninstallation processing strategy, pass-through processing strategy, and resource allocation strategy corresponding to the uninstallation task.

[0042] First, the core components of the target scheduling strategy are clarified. Combining the network interface card (NIC) hardware offloading capabilities with the host resource scheduling requirements, and based on the preset offloading processing strategy (which clarifies which types of service data can be offloaded by the NIC hardware, such as RDMA, data encryption, storage virtualization, etc.), the pass-through processing strategy (which defines the data streams that need to be passed through to the host CPU for processing, such as exception messages, management configuration messages, etc.), and the resource allocation strategy (which specifies the number of NIC queues, bandwidth quotas, and the allocation principles of computing resources), the target scheduling strategy for this dynamic task allocation negotiation is integrated and determined. The core principle is "optimal power consumption and on-demand allocation" to ensure that the scheduling strategy is both adapted to the NIC hardware capabilities and matched with the host resource load.

[0043] Step S302: Obtain the network card load data in real time and transmit the load data to the target host through the control channel.

[0044] The network card collects its own load data in real time through a host electronic device. This load data includes the types of offload tasks that the network card can currently handle, the real-time occupancy of hardware resources (such as the number of idle queues, bandwidth quota remaining, offload engine computing power utilization rate, and cache space size), and the operating status of each functional module. After the data collection is completed, the above load data is encapsulated into standard control messages through the control channel constructed in step S102 and transmitted to the target host in an orderly manner to ensure that the host can accurately grasp the hardware capabilities and real-time load limit of the network card.

[0045] Step S303: Obtain the resource data of the target host, and control the target host to perform task decision-making based on the resource data and load data according to the target scheduling policy, and obtain the task negotiation result of the network card.

[0046] The target host synchronously collects its own resource data, including current CPU utilization, memory bandwidth utilization, protocol stack processing latency, and available collaborative resources for the network interface card (NIC). Subsequently, according to the target scheduling strategy determined in step S301, the target host comprehensively analyzes its own resource data and the load data transmitted by the NIC, and makes task decisions to clarify the data processing division of labor and resource allocation scheme: that is, which data streams are offloaded by the NIC hardware, which data streams are passed through to the host CPU for processing, and the specific allocation ratio of resources on the NIC side. After multiple rounds of iterative negotiation and message exchange, the target host finally generates a clear NIC task negotiation result, which includes the existence of offload tasks, the specific type of offload task, and resource configuration parameters.

[0047] Step S304: Obtain the triggering time and working mode corresponding to the uninstallation task through the task negotiation results, and generate the task decision instruction for the uninstallation task using the triggering time and working mode.

[0048] The target host parses the task negotiation results and extracts key information directly related to the network card offloading task, including the triggering time of the offloading task (such as immediate start or delayed start) and the working mode of each functional module of the network card (such as whether full link training needs to be started later and the power consumption level of the module). Then, based on this key information, a standardized offloading task decision instruction is generated, which can directly guide the upper electronic device to complete the subsequent link training control and offloading task execution.

[0049] Step S305: Control the target host to transmit task decision instructions to the network card through the control channel.

[0050] The target host encapsulates the generated task decision instructions into control messages through the control channel and sends them to the network card. After receiving the messages, the network card verifies and parses the task decision instructions to confirm their integrity and validity, and completes the instruction reception confirmation.

[0051] This completes the task decision-making division between the host and the network interface card (NIC), providing clear instruction guidelines for subsequent NIC link training and offloading tasks.

[0052] Optionally, step S304, which involves obtaining the triggering timing and working mode corresponding to the unloading task through the task negotiation result, and generating the task decision instruction for the unloading task using the triggering timing and working mode, includes the following steps: Step S401: Obtain the host processing task, network card processing task, and negotiation failure task contained in the task negotiation result.

[0053] The host side comprehensively analyzes the task negotiation results formed by multiple rounds of interaction in the early stage, and accurately extracts the three types of core task information contained therein: host processing tasks (all data streams are passed through to the host CPU for processing, and there are no network card offloading tasks), network card processing tasks (there are clearly offloading tasks that need to be undertaken by the network card hardware, such as RDMA, data encryption, etc.), and negotiation failure tasks (the host and network card have not reached an agreement on task division and resource allocation, or an anomaly occurred during the negotiation process), which lays the foundation for subsequent generation of task decision instructions according to different scenarios.

[0054] Step S402: Determine the first working mode corresponding to the unloading task based on the host processing task, and generate the first task decision instruction corresponding to the unloading task using the first triggering time corresponding to the target host under the first working mode.

[0055] For the host processing task in the task negotiation result, it is clear that there is no need for network card offloading in this scenario. Therefore, the first working mode corresponding to the offloading task is determined, namely the control channel maintenance mode. In this mode, there is no need to start the link training process, and the various functional modules of the network card (SerDes, FEC, MAC, etc.) continue to maintain a low-power operation. At the same time, combined with the running requirements of the host processing task, the first triggering time is determined to be "not triggering any link training process and maintaining the lightweight mode for a long time". Based on this first working mode and the first triggering time, the first task decision instruction is generated. The core content of the instruction is: the network card maintains the control channel maintenance mode, does not perform link training, only responds to the task renegotiation request initiated by the host, and all data is transparently transmitted to the host for processing.

[0056] Step S403: Determine the second working mode corresponding to the unloading task based on the network card processing task, and generate the second task decision instruction corresponding to the unloading task using the second triggering time corresponding to the network card in the second working mode.

[0057] For the network interface card (NIC) processing task in the task negotiation results, it is clarified that the NIC needs to undertake specific hardware offloading services in this scenario, thereby determining the second working mode corresponding to the offloading task, namely the transmission mode. In this mode, the standard link training process needs to be started first. After the link parameters converge and the signal quality meets the standard, it switches to the full power consumption service mode. Combining the real-time requirements of the NIC offloading task, the second triggering time is determined to be "immediately start the complete link training process, and start the offloading task after training is completed". Based on this second working mode and the second triggering time, a second task decision instruction is generated. The instruction clearly includes: the start time of link training, the training parameter requirements, and the working status of each module of the NIC after training, as well as the specific execution parameters of the offloading task (such as queue allocation, bandwidth quota, etc.).

[0058] Step S404: Determine the third working mode corresponding to the unloading task based on the negotiation failure task, determine the third triggering time corresponding to the network card through the failure handling strategy or rollback handling strategy corresponding to the third working mode, and generate the third task decision instruction corresponding to the unloading task using the third triggering time.

[0059] For tasks that fail to negotiate in the task negotiation results, the task allocation scheme in which the two parties did not reach an agreement is clarified, and then the third working mode corresponding to the unloading task is determined, namely: negotiation retry mode or traditional mode rollback mode. Subsequently, according to the preset failure handling strategy or rollback handling strategy, the corresponding third triggering time is determined: if the failure handling strategy (waiting for retry) is adopted, the third triggering time is "maintain control channel mode, wait for a preset time and then initiate negotiation retry, and do not trigger link training temporarily"; if the rollback handling strategy (rollback to traditional mode) is adopted, the third triggering time is "immediately start the complete link training process and adapt to the traditional data transmission mode". Finally, based on the determined third working mode and third triggering time, a third task decision instruction is generated. The instruction clearly informs the network card of the current working mode, triggering action (retry negotiation or start link training) and related configuration parameters, ensuring that the network card can handle the negotiation failure scenario according to the preset strategy.

[0060] Optionally, before controlling the network interface card to execute the link training process, the method further includes the following steps: Step S501: If the task decision instruction is the first task decision instruction, then update the control channel between the network card and the target host using the current power consumption parameters of each functional module in the network card.

[0061] If the currently received task decision instruction is detected as the first task decision instruction (i.e., no network card offload task, all data is processed by the host CPU), the control channel update process is initiated. First, the current power consumption parameters of each core functional module (SerDes, FEC, MAC, etc.) inside the network card are collected in real time, including the actual operating power consumption of each module in lightweight control channel mode, the energy consumption data corresponding to the signal transmission amplitude, and the power consumption baseline value in idle state. Then, based on these real-time power consumption parameters, the parameters of the lightweight control channel established between the host and the network card are optimized and updated. The low-amplitude operating parameters of SerDes, the idle code transmission frequency, and the low-power standby configuration of the FEC and MAC modules are adjusted to ensure that the control channel further reduces ineffective power consumption while maintaining stable negotiation and interaction capabilities, adapts to the low-power operation requirements without offload tasks, and ensures a rapid response when the host initiates a renegotiation request.

[0062] Step S502: If the task decision instruction is the third task decision instruction, then update the task decision instruction of the unloading task using the rollback processing strategy corresponding to the third task decision instruction.

[0063] If the currently received task decision instruction is detected to be a third task decision instruction (i.e., negotiation failed or needs to fall back to the traditional mode), then the instruction must first be updated for compatibility. Specifically, the target host side re-examines the task execution requirements based on the fallback processing strategy specified in the third task decision instruction (such as falling back to the traditional Ethernet transmission mode), and adjusts the core parameters in the instruction: if the fallback strategy requires starting full link training to adapt to the traditional mode, then the triggering time in the instruction is updated to "start link training immediately" and the working mode is updated to "traditional business mode"; if the fallback strategy includes configuration related to retry negotiation, then parameters such as retry duration and number of retries are added, and the triggering time is adjusted to "maintain lightweight mode until the retry time node".

[0064] The above updates ensure that the task decision instructions and rollback processing strategies are fully matched, providing accurate and executable instructions for subsequent network card link training, mode switching, or negotiation retry, and avoiding process abnormalities caused by mismatch between instructions and strategies.

[0065] Optionally, the task decision instructions control the network card to execute the link training process corresponding to the Ethernet communication standard set, including the following steps: Step S601: Determine the unloading task contained in the network card using the second triggering timing in the second task decision instruction or the failure handling strategy in the third task decision instruction.

[0066] First, the received task decision instructions are precisely analyzed to clarify the instruction type: if it is a second task decision instruction (corresponding to a scenario where there is a network card offloading task), then based on the second triggering time specified in the instruction (i.e., the time when the network card offloading service needs to be met and high-speed link transmission needs to be started), the scope of offloading tasks that need to be undertaken by the network card is locked; if it is a third task decision instruction (corresponding to a scenario where negotiation fails and it is necessary to roll back to the traditional mode or retry negotiation), then based on the failure handling strategy specified in the instruction, the offloading tasks that still need to be executed by the network card (such as basic data transmission, simple service processing, etc.) are selected, and invalid tasks that do not need to be executed for link training are excluded, ensuring that subsequent link training is only carried out for the actual offloading tasks and avoiding resource waste.

[0067] In this step, it is crucial to analyze the task details in the instructions: clarify the specific type of the unloading task (such as RDMA-related services, data encryption processing, storage acceleration-related tasks, etc.), the task priority, and the corresponding resource allocation requirements. This will provide a clear basis for the targeted configuration of subsequent link training and ensure that the training process and the unloading task requirements are accurately matched.

[0068] Step S602: Based on the unloading task control network card, execute the link training process corresponding to the Ethernet communication standard set, so that the equalizer coefficient parameters, clock data recovery parameters and forward error correction parameters corresponding to the network card are in the target convergence state.

[0069] After determining the specific network interface card (NIC) offloading task, the host electronic device will initiate the complete link training process specified by the Ethernet communication standard to adapt to the offloading task's requirements for high-speed transmission and high reliability. During the training process, optimization adjustments will be made around three core parameters: First, for the SerDes module, adjust the signal transmission amplitude and rate to ensure that it matches the transmission requirements of the offloading task; Second, calibrate the equalizer coefficients to compensate for signal attenuation during high-speed transmission and improve link stability; Third, synchronize the encoding rules of the FEC (Forward Error Correction) module, unify the collaborative working mode of the FEC and MAC modules, and calibrate the CDR (Clock Data Recovery) parameters to achieve timing synchronization between the transmitting and receiving parties.

[0070] The entire training process will continue until all key parameters reach the target convergence state: the equalizer coefficients are adjusted to the optimal level to effectively compensate for channel loss; the FEC encoding and decoding rules are fully synchronized, enabling efficient error correction; and the CDR clock is perfectly matched with the data transmission rhythm, with no timing deviation. At this point, the link signal quality and transmission stability can meet the high-speed transmission requirements of the offloading task, laying a solid foundation for the subsequent efficient execution of the offloading task by the network card and the realization of high-speed data interaction.

[0071] Optionally, control the network card to perform the offloading task, including the following steps: Step S701: Use task decision instructions to control the serializer module, deserializer module, forward error correction module and media access control module in the network card to operate at rated power.

[0072] Based on the previously generated task decision instructions, the core functional modules of the network interface card (NIC) enter their rated power operating state. This includes four core modules: the controller serializer (SerDes), deserializer, forward error correction (FEC), and media access control (MAC). These modules operate according to preset rated power conditions, ensuring that their operation matches the offloading task requirements and high-speed data transmission needs. Specifically, the serializer module operates at its rated amplitude, performing normal signal transmission and reception; the forward error correction module is constantly operational, handling signal errors during data transmission in real time; and the MAC module efficiently completes Ethernet frame encapsulation, parsing, and verification, ensuring data transmission accuracy. All modules work collaboratively to provide hardware support for the smooth execution of subsequent offloading tasks, while simultaneously ensuring the stability and efficiency of data transmission.

[0073] Step S702: If it is detected that the network card has an unloading task within a preset idle time and the target host has not initiated an update instruction for the task decision instruction, then control the network card to execute the unloading task in the current working state.

[0074] After confirming the validity of the task decision instructions and that all functional modules are operating at their rated power, the task execution status of the network interface card (NIC) is continuously monitored. If, within a preset idle period, the NIC still has a clear unloading task (no task interruption, no abnormal signal), and the target host does not send any update notifications regarding task decision instructions (i.e., no task adjustment, mode switching, or other related instructions), the NIC is controlled to stably execute the unloading task according to the predetermined unloading task requirements in the current operating mode. During this process, the operating status, data transmission rate, and error correction status of each functional module are monitored in real time to ensure the efficient and stable progress of the unloading task. Key parameters during task execution (such as data transmission volume and error correction success rate) are also recorded to provide data support for subsequent task optimization. If a task abnormality is detected, timely feedback and adjustments to the operating parameters will be made to ensure the successful completion of the unloading task.

[0075] Optionally, after the network interface card (NIC) executes the unloading task, the method further includes: if it is detected that the NIC does not have an unloading task or an update instruction for the target host to initiate a task decision within a preset idle time, then the control channel between the NIC and the target host is updated using the current power consumption parameters of each functional module in the NIC.

[0076] The task load status of the network interface card (NIC) and host command feedback are continuously monitored. If one of the following two situations is detected, a control channel update operation is triggered: First, the NIC does not detect any offload tasks to be executed within a preset idle time (this time can be pre-configured according to link reliability and power consumption optimization requirements), meaning there is currently no high-speed service data to be processed; Second, the target host actively initiates an update notification for task decision commands, including commands related to task division adjustment, addition / cancellation of offload tasks, and changes in resource allocation parameters. When either of the above situations is met, step S102 is re-executed. The host electronic device collects the current actual power consumption parameters of its internal core functional modules (serializer, deserializer, forward error correction module, and media access control module) in real time. Based on this, the parameters of the lightweight control channel established between the NIC and the target host are optimized and updated to ensure that the control channel is adapted to the low-power operation requirements of no offload tasks or after task updates, while ensuring the stability of subsequent task renegotiation or command interaction between the host and the NIC.

[0077] Regarding the above content, please refer to Figure 2 Another network card control method shown includes the following steps: Step 1: Complete the physical layer automatic negotiation (AN).

[0078] After the physical connection of the ports is established, the automatic negotiation process of the IEEE 802.3 standard is executed to confirm the basic physical layer capabilities supported by both parties, such as the data rate, FEC type, and PAM4 modulation scheme. This step is completed by the PHY hardware and can follow the existing standard process.

[0079] Step 2: After AN is completed, LT will not be started immediately, and the lightweight control channel mode will be entered.

[0080] After AN is completed, the host electronic device (such as an embedded processor or hardware state machine) configures modules such as SerDes (serializer / deserializer), FEC (forward error correction), and MAC into lightweight control channel mode. In this mode: SerDes (serializer / deserializer) operates at a low amplitude, sending idle codes or remaining silent, without performing a complete training sequence exchange; Power consumption is significantly lower than in full-amplitude training mode or full-rate service mode. This mode is sufficient to support the control message exchange required for dynamic task allocation negotiation between the host and the network card.

[0081] Step 3: Perform dynamic task allocation negotiation between the host and the smart network card.

[0082] Through a lightweight control channel, the host CPU and the smart network interface card (NIC) dynamically negotiate to determine the division of data processing responsibilities. The negotiation may include: which data streams will be offloaded by the NIC (e.g., remote direct memory access, encryption, storage virtualization); which data streams will be passed through to the host CPU for processing; and the allocation ratio of NIC resources (e.g., queue size, bandwidth quota).

[0083] This negotiation process may involve multiple message exchanges and requires waiting for a decision from the host side.

[0084] Step 4: Determine the timing of link training based on the negotiation results.

[0085] The higher-level electronic device parses the negotiation result and executes the following branch judgment: Branch A: The negotiation result is that there is no network card offloading task (all data is handled by the host); if the negotiation result is that there is no task offloading and all data is handled by the host CPU, then maintain the lightweight control channel mode (return to step 2), do not trigger link training, keep the network card port in a low-power state, and only respond to task renegotiation requests that the host may initiate later.

[0086] Branch B: The negotiation result indicates that the network card offloading task exists; if the negotiation result clearly indicates that the network card needs to undertake data processing tasks (i.e., there is a business traffic requirement), then proceed to step 5 to trigger the complete link training.

[0087] Branch C: Negotiation fails or rollback is required; if negotiation fails or both parties decide to roll back to traditional mode, then choose according to the pre-configured strategy: keep lightweight mode and wait for retry, or trigger link training to support traditional data channels.

[0088] Step 5: Trigger full link training (LT).

[0089] When the negotiation result indicates that a network card offloading task exists, the standard link training process is initiated. Through training sequence exchange, the convergence of parameters such as equalizer coefficients, CDR (clock data recovery), and FEC (forward error correction) synchronization is completed to ensure that the link achieves reliable high-speed signal quality.

[0090] Step 6: Enter normal business transmission mode.

[0091] After the link training is completed, modules such as SerDes, FEC, and MAC enter full-power operation and begin normal high-speed data transmission and reception, carrying the offloaded tasks determined through negotiation.

[0092] Step 7: When the task is idle or undergoing negotiated updates, return to lightweight mode.

[0093] During service transmission, the network card continuously monitors the task load status. When one of the following conditions is met, it actively exits the normal service mode and falls back to the lightweight control channel mode (returning to step 2), namely: no network card (port) unloads task data within a preset idle time (e.g., 10ms, which can be configured according to link reliability requirements); or a task allocation renegotiation request is received from the host, and the task allocation needs to be renegotiation.

[0094] The above-mentioned network card control method has the following technical effects: 1. Reduce the power consumption of the network card during the task negotiation phase; During the dynamic task allocation negotiation process (typically lasting from a few milliseconds to hundreds of milliseconds), this solution keeps the network interface card (NIC) port in a lightweight control channel mode, avoiding forced high-power link training. Taking a 200G NIC (port) as an example, the power consumption of a single link training session is approximately 5-8W, lasting approximately 50-200ms. This solution can completely save this power consumption during each negotiation process; the cumulative energy saving effect is even more significant in scenarios such as negotiation failure and retries.

[0095] 2. Achieve precise matching between power consumption and task load; When the negotiation result is "no network card offload task" or when the port repeatedly triggers AN due to cable plugging / unplugging or link interruption, this solution does not trigger link training, and the network card port remains in a low power consumption state. This avoids the energy waste of "full power standby even when there is no service" in the traditional solution, and makes the power consumption curve of the network card truly reflect the actual task load it undertakes.

[0096] 3. Improve the overall efficiency of the dynamic task allocation mechanism; This solution incorporates physical layer link state management into the network interface card's task negotiation framework, enabling the decision of "who processes the data" to not only optimize computing resource utilization but also simultaneously optimize power consumption resource utilization, providing a new dimension for data center-level energy efficiency optimization.

[0097] 4. Compatible with existing standards and easy to implement; This solution inserts a state judgment and task negotiation step between AN and LT without modifying the standard procedures of AN and LT themselves. It can be implemented through network card firmware upgrades or hardware state machine optimizations and is transparent to upper-layer applications.

[0098] Corresponding to the above-described network card control method embodiments, this invention also provides a network card control system, such as... Figure 3 As shown, the system includes: Initialization module 100: Used to control the network card to execute the automatic negotiation process corresponding to its Ethernet communication standard set after detecting that the network card is in a physical connection state; Control channel construction module 200: used to establish and configure the control channel between the network card and the target host based on the power consumption parameters of each functional module in the network card; Task decision partitioning module 300: It is used to divide the load data of the network card and the resource data of the target host according to the target scheduling strategy through the control channel, taking the unloading task in the network card as the decision target, and then obtain the task decision instruction for the unloading task. Task execution control module 400: After controlling the network card to execute the link training process corresponding to the Ethernet communication standard set using task decision instructions, it controls the network card to execute the unloading task.

[0099] As can be seen from the above network card control system, the system does not immediately start the link training process after the automatic negotiation process is completed. Instead, it first divides the network card and the target host into tasks according to the target scheduling strategy to achieve dynamic task allocation negotiation. Based on the task decision instructions corresponding to the negotiation results, it determines the triggering time and working mode of the link training process, so that the link power consumption is dynamically matched with the actual task load, thereby reducing the network card power consumption.

[0100] The network card control system provided in this embodiment of the invention has the same implementation principle and technical effect as the aforementioned network card control method embodiment. For the sake of brevity, any parts not mentioned in the system embodiment can be referred to the corresponding content in the aforementioned network card control method embodiment.

[0101] This embodiment also provides an electronic device, the structural schematic diagram of which is shown below. Figure 4As shown, the device includes a processor 101 and a memory 102; wherein, the memory 102 is used to store one or more computer instructions, which are executed by the processor to implement the steps of the above-described network card control method.

[0102] Figure 4 The electronic device shown also includes a bus 103 and a communication interface 104, with the processor 101, communication interface 104 and memory 102 connected via the bus 103.

[0103] The memory 102 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device. The bus 103 may be an ISA bus, PCI bus, or EISA bus, etc. The bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 4 The symbol is represented by a single double-headed arrow, but this does not mean that there is only one bus or one type of bus.

[0104] The communication interface 104 is used to connect to at least one user terminal and other network units through a network interface, and to send encapsulated IPv4 packets or IPv4 packets to the user terminal through the network interface.

[0105] Processor 101 may be an integrated circuit chip with signal processing capabilities. In implementation, each step of the above method can be completed by the integrated logic circuitry in the hardware of processor 101 or by instructions in software form. The processor 101 can be a general-purpose processor, including a Central Processing Unit (CPU), a Network Processor (NP), etc.; it can also be a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this disclosure. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this disclosure can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory 102. The processor 101 reads the information in memory 102 and, in conjunction with its hardware, completes the steps of the method described in the foregoing embodiments.

[0106] This invention also provides a storage medium storing a computer program, which, when executed by a processor, performs the steps of the network card control method described in the foregoing embodiments.

[0107] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, devices, and methods can be implemented in other ways. The system embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the coupling or direct coupling or communication connection shown or discussed may be through some communication interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0108] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0109] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0110] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a processor-executable, non-volatile, computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, electronic device, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0111] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the technical scope disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A network interface card (NIC) control method, characterized in that, The method includes: When the network card is detected to be in a physically connected state, the network card is controlled to execute the automatic negotiation process corresponding to its Ethernet communication standard set; A control channel between the network card and the target host is established and configured based on the power consumption parameters of each functional module in the network card. Taking the unloading task in the network card as the decision target, the load data of the network card and the resource data of the target host are divided into task decision according to the target scheduling strategy through the control channel, and then the task decision instruction of the unloading task is obtained. After controlling the network card to execute the link training process corresponding to the Ethernet communication standard set using the task decision instructions, the network card is then controlled to execute the offloading task.

2. The network card control method according to claim 1, characterized in that, The steps of establishing and configuring a control channel between the network interface card (NIC) and the target host based on the power consumption parameters of each functional module in the NIC include: Obtain the operating status data of the serializer module, deserializer module, forward error correction module, and media access control module in the network card; Based on the first operating status data between the serializer module and the deserializer module, the first power consumption parameter corresponding to the idle code interaction between the network card and the target host is obtained. The second power consumption parameter corresponding to the packet interaction between the network card and the target host is obtained by using the second operating status data between the forward error correction module and the media access control module. Using the first power consumption parameter and the second power consumption parameter as constraints, a control channel between the network card and the target host is established and configured.

3. The network card control method according to claim 1, characterized in that, The steps of taking the offloading task in the network interface card (NIC) as the decision target, dividing the load data of the NIC and the resource data of the target host according to the target scheduling strategy through the control channel to obtain the task decision instruction for the offloading task include: The target scheduling strategy is determined based on the uninstallation processing strategy, pass-through processing strategy, and resource allocation strategy corresponding to the uninstallation task. The load data of the network card is acquired in real time, and the load data is transmitted to the target host through the control channel; The resource data of the target host is obtained, and the target host is controlled according to the target scheduling policy to perform task decision-making and division based on the resource data and the load data, so as to obtain the task negotiation result of the network card. The triggering timing and working mode corresponding to the uninstallation task are obtained through the task negotiation results, and the task decision instructions for the uninstallation task are generated using the triggering timing and the working mode. The target host is controlled through the control channel to transmit the task decision instructions to the network card.

4. The network card control method according to claim 3, characterized in that, The steps of obtaining the triggering timing and working mode corresponding to the uninstallation task through the task negotiation result, and generating the task decision instruction for the uninstallation task using the triggering timing and the working mode, include: Obtain the host processing task, network card processing task, and negotiation failure task included in the task negotiation result; Based on the host processing task, a first working mode corresponding to the unloading task is determined, and a first task decision instruction corresponding to the unloading task is generated using the first triggering timing corresponding to the target host under the first working mode. Based on the network interface card (NIC) processing task, determine the second working mode corresponding to the unloading task, and use the second triggering timing corresponding to the NIC in the second working mode to generate the second task decision instruction corresponding to the unloading task. Based on the negotiation failure task, the third working mode corresponding to the unloading task is determined. The third triggering time corresponding to the network card is determined by the failure handling strategy or rollback handling strategy corresponding to the third working mode. The third task decision instruction corresponding to the unloading task is generated using the third triggering time.

5. The network card control method according to claim 4, characterized in that, Before controlling the network interface card to execute the link training process, the method further includes: If the task decision instruction is the first task decision instruction, then the control channel between the network card and the target host is updated using the current power consumption parameters of each functional module in the network card; If the task decision instruction is the third task decision instruction, then the task decision instruction for the unloaded task is updated using the rollback processing strategy corresponding to the third task decision instruction.

6. The network card control method according to claim 4, characterized in that, The task decision instructions control the network interface card to execute the link training process corresponding to the Ethernet communication standard set, including: The unloading task contained in the network card is determined by using the second triggering timing in the second task decision instruction or the failure handling strategy in the third task decision instruction; Based on the unloading task, the network card is controlled to execute the link training process corresponding to the Ethernet communication standard set, so that the equalizer coefficient parameters, clock data recovery parameters and forward error correction parameters corresponding to the network card are in the target convergence state.

7. The network card control method according to claim 2, characterized in that, Controlling the network interface card to execute the offloading task includes: The task decision instructions are used to control the serializer module, the deserializer module, the forward error correction module, and the media access control module in the network card to operate at rated power. If it is detected that the network interface card (NIC) has the uninstallation task within a preset idle time and the target host has not initiated an update instruction for the task decision instruction, then the NIC is controlled to execute the uninstallation task in the current working state.

8. The network card control method according to claim 7, characterized in that, After controlling the network card to execute the offloading task, the method further includes: If it is detected that the network interface card (NIC) does not have an unload task or the target host initiates an update instruction for the task decision within a preset idle time, the control channel between the NIC and the target host is updated using the current power consumption parameters of each functional module in the NIC.

9. A network card control system, characterized in that, The system includes: Initialization module: Used to control the network card to execute the automatic negotiation process corresponding to its Ethernet communication standard set after detecting that the network card is in a physical connection state; Control channel construction module: used to establish and configure the control channel between the network card and the target host based on the power consumption parameters of each functional module in the network card; Task decision partitioning module: Used to divide the load data of the network card and the resource data of the target host according to the target scheduling strategy through the control channel, taking the unloading task in the network card as the decision target, and then obtain the task decision instruction of the unloading task. Task execution control module: After controlling the network card to execute the link training process corresponding to the Ethernet communication standard set using the task decision instructions, it controls the network card to execute the unloading task.

10. An electronic device, characterized in that, The electronic device includes a processor and a memory, the memory storing computer-executable instructions that can be executed by the processor, the processor executing the computer-executable instructions to implement the steps of the network interface card control method according to any one of claims 1 to 8.

Images