1659 results about "Network interface" patented technology

A coherent, intelligent, packet-switched memory fabric enables predictive, cache-coherent access across distributed compute, accelerator, and memory resources using a Memory-Fabric Transaction Layer Protocol (MF-TLP). MF-TLP defines routable packet formats for read, write, vectorized, atomic, reduction, collective, and predictive-prefetch transactions executed by memory-centric network interface controllers (MC-NICs). Each MC-NIC performs packet parsing, address translation, coherence management, and near-memory arithmetic or tensor operations while coordinating with MF-TLP-aware switches providing hierarchical directory control, multi-path routing, and in-network aggregation. Vectorized and multimodal packets encode multiple addresses or tensor offsets to reduce scatter / gather overhead, and programmable caching and quality-of-service modules manage tiered memory and tenant fairness. MF-TLP supports extension headers for predictive prefetch, collective coordination, and tenant governance, operating across hierarchical leaf-spine topologies using Ultra-Ethernet Transport, InfiniBand, or CXL fabrics. The system delivers scalable, low-latency, memory-centric orchestration for large-language-model training, multimodal AI, and data-intensive analytics.

A system and method for optimizing automated textual prompts in artificial intelligence (AI) conversational systems is disclosed. The system comprises a network interface, processors, and memory-storing instructions for performing operations to optimize prompts. These operations include receiving and preprocessing input data, tokenizing the data, verifying data authenticity, performing temporal analysis, calculating prompt complexity scores, and selectively expanding or refining prompts based on complexity thresholds. The system further incorporates context-aware optimization, multi-faceted prompt refinement, variation generation, and evaluation using machine learning models. Additional features include a technological hub with advanced processing capabilities, sensor-augmented input apparatus, device-specific prompt optimization, AI model selection, multimodal context integration, and an AI-driven creativity booster. The system provides interactive prompt visualization, certification, and uniqueness verification modules. This comprehensive approach ensures the generation of optimized, contextually relevant, and creative prompts for various AI applications while maintaining data integrity and user engagement.

The invention relates to the technical field of mobile communication transmission, and discloses a heterogeneous network multilink parallel mobile communication transmission method and system. A dynamic link aggregation module of the system collects link state parameters including signal strength, delay jitter and packet loss rate from a plurality of heterogeneous network interfaces in real time; a flow distribution decision module calculates available bandwidth weight of each link according to the data and generates a dynamic flow distribution strategy; the data fragment packaging module splits to-be-transmitted data into a plurality of data fragments according to the strategy, and adds a sequence identifier and a link identifier to each fragment; the multi-channel concurrent transmission module sends corresponding data fragments in parallel through a specified heterogeneous network link; and the receiving recombination verification module sorts the recombination data fragments according to the sequence identifiers, and detects the data integrity through a verification mechanism. According to the system, a set of complete multi-link parallel transmission solution is formed, and the problems existing in a traditional single link and an existing multi-link transmission technology can be effectively solved.

The invention relates to an optical data network (10-1) for connecting a plurality of data network nodes (N1-1, N2-1, N3-1, N4-1). The optical data network (10-1) comprises: i) the plurality of data network nodes (N1-1, N2-1, N3-1, N4-1), the data (99) comprising data packets (99p), wherein each data network node (N1-1, N2-1, N3-1, N4-1) is provided with a network interface (Nl); and ii) a data connection network (100-1) comprising at least one data switch (SW) and a plurality of optical connections (50) for allowing establishment of data links (DL1, DL2) between pairs of said plurality of data network nodes (N1-1, N2-1, N3-1, N4-1). At least one of the at least one data switch (SW) is a smart data switch (SSW) comprising at least one packet switch (PS, EPS) and at least one circuit switch (CS, OCS), At least two network interfaces are smart network interfaces (SN, SNC) comprising a data classifier (25) configured for classifying the data packets (99p) into at least two traffic classes including first traffic class (D1) and a second traffic class (D2) in accordance with a classification scheme. The smart data switch (SSW) and the network interfaces (SN, SNC) are configured to route the data packets (99p) of the first traffic class (D1) to the packet switch (PS, EPS) and to route the data packets (99p) of the second traffic class (D2) to the circuit switch (CS, OCS).

A network interface module for coupling a host device to a switched network as a network node is described. The network interface module comprises a single half-duplex port for communicatively coupling to a shared bus of the switched network, at least one frame queue sized to store one multicast read frame received via the shared bus, and logic circuitry. The logic circuitry is configured to decode a read command for the interface module included in a payload of the multicast read frame that includes multiple read commands for other network nodes of the switched network, and transmit a response frame including read data on the shared bus when detecting the shared bus is available for transmitting.

A decentralized mesh-network platform provides real-time governance, supervision, and audit of large-language-model persona agents in regulated environments such as healthcare. Each computing node hosts processor circuitry, memory, a network interface, and an integrated suite comprising a User-Prompt Interface, Metadata Filter & Enrichment Engine, Persona Role Constructor, Policy Constraint Engine, gating logic module, Supervisor Review Interface, Session Memory Governance Module, Audit Logging Subsystem, cryptographically linked audit ledger, and Fallback Control Module. Devices communicate over a peer-to-peer channel, exchange role-switch tokens, and merge local ledgers into a composite audit graph when connectivity returns. The architecture supports continuous compliance and resilient operation across institutional boundaries, network partitions, and device failures, enabling conversational interaction with heterogeneous medical data while maintaining rigorous credential lineage, policy oversight, and forensic traceability.

A robotic surgical system network includes a plurality of robotic surgical systems. Each system includes robotic arms, sensors, a surgeon console, and a control system with an integrated AI module. A network interface is associated with robotic surgical system and provides secure data communication. A central or distributed data repository securely stores surgical data aggregated from the robotic surgical systems. The surgical data includes at least one of procedural data, sensor readings, imaging data, AI decision logs, surgical outcomes, or user interaction data. A training module utilizes aggregated surgical data to train or update AI models for the robotic surgical systems using unsupervised learning, transfer learning, or federated learning techniques. A cybersecurity module implements security measures for data transmission and system access, the measures comprising at least one of encryption, multi-factor authentication, or real-time threat detection.

An example system comprises a plurality of servers comprising respective network interface cards (NICs) connected by physical links in a physical topology, wherein each NIC of the plurality of NICs comprises an embedded switch and a processing unit coupled to the embedded switch; and an edge services controller configured to program the processing unit of a network interface card of the plurality of network interface cards to: receive, at a first network interface of the NIC, a data packet from a physical device; based on the data packet being received at the first network interface, modify the data packet to generate a modified data packet; and output the modified data packet to the physical device via a second network interface of the NIC.

The invention discloses an AI-driven RDMA network card hardware unloading acceleration method and system based on an RISC-V architecture, the system comprises an RISC-V processing unit, the RISC-V processing unit comprises a control module and a driving module, and the driving module stores RDMA exclusive acceleration instruction set extension; the AI reasoning unit is connected with the driving module and is used for expanding and running a pre-trained network optimization model based on an RDMA exclusive acceleration instruction set to analyze a network state and an application requirement, generating a hardware unloading strategy optimization instruction and sending the hardware unloading strategy optimization instruction to the control module; the control module generates a corresponding management command according to the hardware unloading strategy optimization instruction; and the hardware unloading unit comprises a management module and an RDMA component module, and the management module receives the management command and performs parameter configuration on the RDMA component module according to the management command. According to the invention, the burden of the CPU is reduced, and the throughput is improved.

The invention relates to an activation value re-segmentation communication method for a heterogeneous AI acceleration chip server. The activation value re-segmentation communication method comprises the following steps: scanning through chips, network interface cards and PCIe levels in the AI acceleration chip server, extracting network topology information of the server, calculating delay and bandwidth from each chip to each network interface card, and generating an affinity matrix; based on the affinity matrix, each chip is bound to a network interface card in an exclusive or shared mode through maximum matching; based on the server network topology information, the chip-network interface card affinity and the network interface card and PCIe single line bandwidth, activation value communication mode matching is carried out, and communication modes comprise a transmitting / receiving mode, a transmitting / receiving + collecting mode and a transmitting / receiving + broadcasting mode; and double-level multi-channel communication between the chip and the network interface card is realized. The method has the advantages of remarkably improving the bandwidth utilization rate, being highly adaptive to topology difference, being high in heterogeneous compatibility and the like.

The invention provides an optimized kernel network and an implementation method thereof, and relates to the technical field of kernel networks, the kernel network comprises a fast data channel component XDP arranged on a data link layer; when the network card equipment receives a message data packet, the XDP analyzes and filters the received data packet, or encapsulates or de-encapsulates the received data packet, and modifies a field to forward the data packet; and the XDP matches and filters the received data packet with the rules in the BPF Maps, and discards the data packet when the data packet is illegal. The method comprises the steps that hardware initialization comprises the step of adding a zero copy processing flow when an NAPI polling point processing function is added; the application program configures an XDP socket; the xsk unit executes a BPF instruction, calls a network card driving unit to configure an XSK cache pool, and establishes a DMA mapping relationship between the device and the UMEM shared cache; and the network card drives zero-copy receiving and receiving of the data packet. According to the invention, the kernel network communication efficiency is improved.

The invention provides an uninterrupted redundant communication method and system, equipment and a medium, and belongs to the technical field of vehicle-mounted network communication.The method comprises the steps that data frames from two communication paths are received through two network interfaces respectively, CRC verification is carried out, wrong data frames are discarded, and the data frames passing verification are obtained; extracting the serial number of each data frame passing the verification as a target serial number, and comparing the target serial number with an expected serial number; if the target serial number is equal to the expected serial number, submitting the data frame to an upper application, and adding 1 to the expected serial number; wherein the data frame is generated by adding a globally unique and monotonically increasing serial number to each key application data frame by the sending end processing unit. According to the invention, fault switching delay is eliminated fundamentally, zero switching time is realized, and a crucial communication guarantee is provided for automatic driving of L4 and above.

The invention provides a software and hardware integrated intelligent terminal power transaction system and method, relates to the technical field of power market transaction, and solves the technical problems of data potential safety hazards and insufficient real-time performance caused by dependence on an external platform in a power transaction process in the prior art. The system specifically comprises a hardware architecture and a software function module, the hardware architecture comprises a processor, a redundant power supply, a storage device, a network interface, a security encryption chip and a display interface; the software function module comprises a core function layer, an interface layer and a security layer; wherein the core function layer is used for being responsible for processing and controlling the whole flow of power transaction; the interface layer is used for carrying out standardized and ordered external data interaction; and the security layer is used for guaranteeing the security of the terminal from multiple dimensions of communication, transaction, data and operation and maintenance. The method is used for electricity market transaction.

A network interface controller (NIC) capable of facilitating efficient data request management is provided. The NIC can be equipped with a command queue, a message chopping unit (MCU), and a traffic management logic block. During operation, the command queue can store a command issued via a host interface. The MCU can then determine a type of the command and a length of a response of the command. If the command is a data request, the traffic management logic block can determine whether the length of the response is within a threshold. If the length exceeds the threshold, the traffic management logic block can pace the command such that the response is within the threshold.

The invention discloses a network card operation monitoring method and electronic equipment, and relates to the technical field of network card management, and the method comprises the steps: monitoring the operation state of a target network card, and if it is judged that the target network card currently has a target error event according to the operation state, and the current operation mode of the target network card is a data processor mode, executing the operation of the target network card; if yes, acquiring context data when the target network card operates based on the data processor mode, storing the context data in a preset shared storage area, activating the first function stack, and switching the current operation mode to the network interface card mode; and based on the context data in the preset shared storage area, establishing a network connection when the target network card operates based on the network interface card mode, and repairing a target error event of the target network card. The network card can be quickly switched to a stable working mode based on the network interface card when the mode based on the data processor fails, the service interruption time is shortened through self-repairing of the network card, and the operation and maintenance complexity and cost are remarkably reduced.

System and methods are disclosed relating to cloud edge based anomaly detection. In an example, an edge monitoring node can include a thermal camera to capture a thermal image of equipment that is under monitoring for an an anomaly event. The node further includes a machine learning (ML) model to process the thermal image of the equipment to detect the anomaly event. The node further includes a network interface to communicate the detected anomaly event over a network to a remote computing platform to determine one or more recommendations for proactive maintenance of the equipment.

An RDMA queue pair multiplexing method based on a shared memory efficiently maps a large number of logical queue pairs (LQPs) to a small number of physical queue pairs (PQPs), and combines shared memory zero-copy communication and O (1) active queue management, so that ten thousand-level LQPs are supported by a small number of PQPs, the overhead of a single-connection memory is reduced by about 60%-64%, and context jitter of a network interface card (NIC) is avoided; the connection establishment time delay and the short flow end-to-end time delay are greatly reduced; through O (1) active tracking and batch processing, CPU idling and system calling overhead are greatly reduced, and high throughput and expandability are achieved; the method can be implemented in a pure user mode, a kernel or NIC does not need to be modified, and the method can be immediately deployed in an existing RDMA environment.

The present application relates to a data replay method, a data processing unit, a network interface card, a device, and a storage medium. The method comprises: when a storage stack is restarted, acquiring a replay-pending request queue from a persistent memory, the replay-pending request queue being a queue obtained by mapping a transmission buffer area into the persistent memory, and the transmission buffer area being used for receiving a request from a host side; on the basis of the replay-pending request queue, acquiring a replay-pending request from the persistent memory; and resending the replay-pending request to a target acceleration component directed to a storage device, so as to instruct the storage device to store target data which resides in the persistent memory and corresponds to the replay-pending request. In the method, a request queue, a request, and data corresponding to the request are stored in the persistent memory in advance, and if the storage stack is restarted, the replay-pending request is recovered from the persistent memory, eliminating the need for DMA and encryption / decryption techniques, and shortening an interruption duration of a storage data path, thereby improving data storage efficiency on the host side.

The invention discloses a network interface card with autonomous access authentication, and provides an integrated hardware authentication mechanism. The EAP-TLS client protocol stack is integrated on the bottom layer of the network card, so that the defect that an authentication module depends on a host system, manual intervention and the like in the existing scheme is overcome, TLS handshake and certificate verification are independently completed on the network card side, and authentication interaction is automatically initiated. The network interface card serves as a unique identity certificate of an access network, a client certificate and a corresponding private key are embedded in the network interface card and are stored in a protected nonvolatile area, and data confidentiality is effectively enhanced. The network card is used as an identity certificate and network access equipment at the same time, and the integrated design reduces the dependence on the service capability of management personnel. A traditional physical network card and an automatic authentication protocol are fused, and the method is suitable for application environments with high requirements for access control and authentication safety, such as an enterprise intranet and a government affair private network.

A data processing method, a network interface card, an electronic device, and a storage medium. The method includes: determining a target dispatch queue to be dispatched from a host memory in response to a doorbell signal sent by a target host, where the doorbell signal indicates that there is a target message to be sent; determining a current target dispatch state of the target dispatch queue, where the target dispatch state is obtained based on an activity and a credit value corresponding to the target work queue element; determining whether a dispatch mechanism corresponding to the target dispatch queue is valid based on the target dispatch state, where the dispatch mechanism indicates whether the target dispatch queue is allowed to be dispatched; and performing a dispatch operation on the target work queue element and the target message in response to the dispatch mechanism being valid.

The present application relates to a multi-path congestion control method and apparatus, a chip, a network interface card, and a computer device. The method comprises: under the condition that a token allocation request is received, on the basis of the priority information of M paths, determining N path combinations to be balanced; for each path combination, analyzing traffic to be balanced of a first path and a second path within the current period on the basis of respective round-trip delay of the first path and the second path within a previous period, and determining the number of tokens of the first path and the number of tokens of the second path on the basis of the traffic to be balanced; and feeding back the number of tokens of each path to hardware to instruct the hardware to transmit a message on the basis of the number of tokens of each path. Use of the present method can transfer traffic on a path having a large congestion degree to a path having a small congestion degree, and balance the messages on multiple paths within a next period, thereby achieving load balancing on different paths, and improving the message transmission efficiency and the bandwidth utilization rate of the whole link.

Methods and systems are described for routing based on a path-based protocol address. Data is detected, by a current node in a current location in a network path, in a data unit that is specified according to a network protocol and that includes a first path-based protocol address including a plurality of path segment identifiers that identify respective path segments of a network path for transmitting data from a source node to a destination node. A current-next path segment identifier is detected, based on the current location, that identifies a current-next path segment in the plurality for transmitting the data from the current node to a next node in the network path. A current-next network interface id determined, based on the current-next path segment identifier, that is included in the current-next path segment. The data is sent, via the current-next network interface, to the next node.

The present disclosure provides a method for processing a secret-shared adjacency list of a graph. The method includes partitioning the secret-shared adjacency list into blocks, adding a vertex identifier to each tuple in the secret-shared adjacency list, defining a boolean variable for each block, computing a secret-shared vector of row identifiers for each tuple, shuffling tuples and additional tuples, extracting components from the shuffled tuples to form a d-normalized replicated adjacency list, and storing the d-normalized replicated adjacency list in a non-volatile storage device. The method utilizes parallel processing capabilities and a hardware-based random number generator for shuffling. The method also includes securely transmitting and receiving encrypted data related to adjacency lists through a network interface.

The invention belongs to the technical field of server detection, and particularly relates to a server network function test method and system, a terminal and a storage medium, and the method comprises the steps: defining a test target, and configuring probe interface parameters and a resource scheduling strategy; the probe reports the CPU, the memory and the network interface utilization rate of the server to the control node in real time, and the control node adjusts the distribution of the test task by calculating the real-time load coefficient and the dynamic weight of the node according to the configured resource scheduling strategy; executing a multi-dimensional test based on the test points distributed after the test task is adjusted; through configuring a resource scheduling strategy in the step S1 and combining calculation of a node real-time load coefficient and a dynamic weight in the step S2, dynamic allocation of a test task is realized. When the node load reaches the threshold value, task migration is triggered according to the dynamic scheduling logic, resource waste caused by fixed node configuration is avoided, test resources are made to incline towards low-load nodes, the test period is remarkably shortened, and the resource utilization rate is increased.

A system receives, by a network interface card (NIC), inputs including an instruction to read or write a payload of a message, a tracker state indicating a round of processing, and a datatype descriptor defining organization of the message payload. The system identifies a current context and a processing state for the instruction. If the datatype descriptor indicates a first type, the system: obtains the current context associated with the first type from a host memory or a cache of the NIC; and creates direct memory access (DMA) instructions corresponding to the received instruction by executing operations in a nested loop. If the datatype descriptor indicates a second type, the system: obtains the current context associated with the second type by fetching vector entries from a buffer of the NIC; and creates the DMA instructions corresponding to the received instruction based on addresses and lengths in the vector entries.

The invention relates to the technical field of communication machine room monitoring, and discloses a communication machine room intelligent monitoring system based on edge computing and an implementation method thereof. The method comprises the steps that an edge gateway is connected with intelligent equipment through a network interface, and a monitoring unit is connected with traditional equipment through a serial port interface; when new equipment is accessed, identifying a protocol type through the edge gateway and generating communication parameter configuration; serial port equipment data is acquired according to the communication parameter configuration, the data acquisition time deviation between the edge gateway and the monitoring unit is calculated, when the data acquisition time deviation exceeds a preset threshold value, the edge gateway acquires standard UTC time, and a first local clock of the edge gateway and a second local clock of the monitoring unit are calibrated as the standard UTC time. According to the communication machine room monitoring system, a collaborative division mechanism of the edge gateway and the embedded monitoring unit is established, so that full-coverage acquisition of the network equipment and the serial port equipment is realized, and the intelligent level and the operation and maintenance efficiency of the communication machine room monitoring system are further improved.

The present application relates to the technical field of computers. Disclosed are a computing system, a model training method and apparatus, and a device, a medium and a product. The artificial intelligence computing system comprises a plurality of computing nodes. Each computing node comprises a host, a computing device and a network interface card, wherein the computing device comprises a first hardware interface, a first device coherence engine and a first computing core, and is connected to the host by means of the first hardware interface; the network interface card comprises a second hardware interface and a second device coherence engine, and is connected to the host by means of the second hardware interface; and by means of a cache flush operation, the computing device transmits gradient data with a network interface card memory of the network interface card, and the network interface card is used for synchronizing the gradient data with other computing nodes. The present application improves the efficiency of transmitting gradient data between a computing device and a network interface card, thus improving the training efficiency for an artificial intelligence model.

The invention provides a dynamic self-adaptive data stream accelerated transmission method and device oriented to a credential environment. The method comprises the following steps: carrying out capability detection on a central processing unit, an operating system, a network interface and a password acceleration unit in a credential computing node to obtain environment capability information, and organizing the environment capability information into a credential environment capability portrait; analyzing the service type, the data scale, the timeliness level and the security level of the data stream, and generating a data stream portrait corresponding to the data stream in combination with a preset classification rule; acquiring running state information in a cross-layer manner and organizing the running state information into running state vectors; based on the credential environment capability portrait, the data stream portrait and the operation state vector, executing a preset strategy decision process, and determining a transmission strategy parameter set for the data stream; and sending the data fragments according to the transmission strategy parameter set. The resource utilization rate of the creative environment can be improved, the multi-type data stream transmission efficiency is improved, and the manual tuning overhead of transmission parameters is reduced.

A telecommunications switch-type platform positioned between a private-branch exchange and a data network dynamically hosts artificial-intelligence agents that interpret, transcribe, translate and transliterate live call traffic while the session is in progress. Machine-readable instructions stored in local memory instantiate, migrate and terminate the agents on demand across dedicated processing resources such as field-programmable gate arrays, application-specific integrated circuits or system-on-chip devices, thereby maintaining end-to-end latency below 250 milliseconds. An integrated call-audio recorder captures bidirectional media streams for secure archiving without interrupting service. The platform exposes a network interface that passes both signalling and media traffic, enabling seamless deployment inside call-centre or enterprise environments and compatibility with softswitch architectures conforming to class-4 Computational Interpretation of Communication using Artificial Intelligence standards. The same functional stack is deliverable as a computer-implemented method and as a non-transitory machine-readable medium storing the instructions executed by the processing resources.

Provided is a system for network access control. The system includes an authentication server configured to perform a basic authentication procedure for a user by communicating with at least one user terminal, a service server configured to provide a service to at least one user terminal passing through the basic authentication procedure, a collection device configured to acquire traffic data of the at least one user terminal passing through the basic authentication procedure from at least one of the service server and a network interface connected to the service server, acquire a traffic dataset by refining the acquired traffic data in accordance with correlation, and extract time-series feature points of the traffic dataset, and an artificial intelligence (AI) management device configured to train at least one AI model to define a traffic character template of the user on the basis of the time-series feature points.