Abstract: A method and apparatus for neural network based cross component prediction with scaling factors during encoding or decoding of an image frame or a video sequence, which may include training a deep neural network (DNN) cross component prediction (CCP) model with at least one or more scaling factors, wherein the at least one or more scaling factors are learned by optimizing a rate-distortion loss based on an input video sequence comprising a luma component, and reconstructing a chroma component based on the luma component using the trained DNN CCP model with the at least one or more scaling factors for chroma prediction. The trained DNN CCP may be updated for chroma prediction of the input video sequence using the one or more scaling factors, and performing chroma prediction of the input video sequence using the updated DNN CCP model with the one or more scaling factors.

Abstract: A method and apparatus for neural network based cross component prediction with low-bit precision during encoding or decoding of an image frame or a video sequence, which may include reconstructing a chroma component based on a received luma component using a pre-trained deep neural network (DNN) cross component prediction (CCP) model for chroma prediction, and updating a set of parameters of the pre-trained DNN CCP model with low-bit precision. The method may also include generating an updated DNN CCP model for chroma prediction with low-bit precision based on at least one video sequence, and using the updated DNN CCP model for cross component prediction of the at least one video sequence at reduced processing time.

Abstract: A method for load balancing and packet re-sequencing on a network includes an ingress node that determines an identifier of a destination chip based on a destination address of a packet flow and a forwarding table, allocates, based on the identifier of the destination chip, a channel identifier and a sequence number of each packet that is in the packet flow and that enters a channel, to generate an updated packet flow, where each packet in the updated packet flow includes a packet in a corresponding packet flow, a sequence number of the packet, and the channel identifier, and the updated packet flow is sent in a per-packet load sharing manner.

Abstract: A memory management method includes: determining that available storage space of a first memory in a network device is less than a first threshold, where the first threshold is greater than 0 and the first memory stores a first packet queue; and deleting at least one packet at the tail of the first packet queue from the first memory based on the available storage space of the first memory being less than the first threshold. When the available storage space of the first memory is less than the first threshold, a packet queue, namely, the first packet queue, is selected and a packet at the tail of the packet queue is deleted from the first memory.

Abstract: A packet forwarding method to shorten a transmission latency of an elephant flow is provided. In the method, for a first packet flow used as an elephant flow, a network device may receive a plurality of packets of the first packet flow, and determine a characteristic parameter of the first packet flow based on the plurality of packets, where the characteristic parameter of the first packet flow is used to indicate a transmission latency of the first packet flow. After determining the characteristic parameter of the first packet flow, the network device determines a forwarding policy of the first packet flow based on the characteristic parameter of the first packet flow. The forwarding policy of the first packet flow is used to indicate latency sensitivity of the first packet flow.

Abstract: A method for determining a packet dequeue rate includes allocating a plurality of consecutive blocks in a first memory to a first packet, storing the first packet and a first length in the plurality of blocks, where the first length is of a first packet queue and is obtained when the first packet is enqueued into the first packet queue, and determining, based on a first span and the first length stored, a first rate at which a packet in the first packet queue is dequeued, where the first span is equal to a difference between a second time and a first time, the first time is when the first packet is enqueued into the first packet queue, and the second time is when the first packet is dequeued from the first packet queue.

Abstract: Methods and apparatus to reduce stress on lasers in optical transceivers are disclosed. An apparatus comprising a printed circuit board (PCB) having a first side and a second side opposite the first side; and a first stiffener attached to the first side of the PCB; and a photonic integrated circuit (PIC) attached to the first stiffener. The first stiffener is between the PIC and the PCB. The apparatus also includes a second stiffener attached to the second side of the PCB.

Abstract: A distributed database architecture based on shared memory and multi-process includes a distributed database node. A system shared memory unit and a system process unit are built in a distributed database. The system shared memory unit includes a task stack information module and a shared cache module. A plurality of process tasks are built in the task stack information module. The process tasks include system information with various purposes in system process task information, and each system information corresponds to one process task. By using a system shared memory unit at a distributed database node, the number of user connections in the distributed database architecture does not have a corresponding relationship with the number of processes or threads. The number of processes or threads of the entire node does not increase as the number of user connections increases.

Abstract: This application relates to the field of data communication, and in particular, to a packet buffering method, an integrated circuit system, and a storage medium. The method can improve utilization of the on-chip buffer. The packet buffering method may be applied to a network device. The network device includes a first storage medium and a second storage medium. The first storage medium is a local buffer, and the second storage medium is an external buffer. The method may include: receiving a first packet, and identifying a queue number of the first packet, where the queue number indicates a queue for storing the first packet; querying a queue latency based on the queue number; determining a first latency threshold based on usage of the first storage medium; and buffering the first packet in the first storage medium or the second storage medium based on the queue latency and the first latency threshold.

Abstract: A distributed pessimistic lock based on HBase storage and a method for implementing a database pessimistic lock; the distributed pessimistic lock including a lock manager configured to be installed on a Region of a RegionServer node of a HBase system, the lock manager has a lock and unlock interface; and the distributed pessimistic lock, an operation transaction, and a lock holder form a cross linked list format; a horizontal dimension is an information of a current data row, a vertical dimension is an information of the operation transaction, and an intersection point between the horizontal dimension and the vertical dimension is the lock holder. By installing the lock manager on a node Region of a HBase storage system, the lock manager locks and unlocks a data operation of the HBase system with the distributed pessimistic lock.

Abstract: Embodiments of this application provide a packet processing method and apparatus, and a communications device. The method includes the following operations: After obtaining first network cost information of a received first packet, a network device may determine a first queue for the first packet based on the first network cost information, and buffer the first packet to the first queue, where the first network cost information is used to indicate a cost for transmitting the first packet to an egress node. Therefore, the network device considers costs for transmitting packets to the egress node, divides queues based on values of the costs that are indicated, and performs differentiated enqueuing operations on the received packets considering the values of the costs.

Abstract: A distributed database architecture based on shared memory and multi-process includes a distributed database node. A system shared memory unit and a system process unit are built in a distributed database. The system shared memory unit includes a task stack information module and a shared cache module. A plurality of process tasks are built in the task stack information module. The process tasks include system information with various purposes in system process task information, and each system information corresponds to one process task. By using a system shared memory unit at a distributed database node, the number of user connections in the distributed database architecture does not have a corresponding relationship with the number of processes or threads. The number of processes or threads of the entire node does not increase as the number of user connections increases.

Abstract: A distributed pessimistic lock based on HBase storage and a method for implementing a database pessimistic lock; the distributed pessimistic lock including a lock manager configured to be installed on a Region of a RegionServer node of a HBase system, the lock manager has a lock and unlock interface; and the distributed pessimistic lock, an operation transaction, and a lock holder form a cross linked list format; a horizontal dimension is an information of a current data row, a vertical dimension is an information of the operation transaction, and an intersection point between the horizontal dimension and the vertical dimension is the lock holder. By installing the lock manager on a node Region of a HBase storage system, the lock manager locks and unlocks a data operation of the HBase system with the distributed pessimistic lock.

Abstract: A memory management method includes: determining that available storage space of a first memory in a network device is less than a first threshold, where the first threshold is greater than 0 and the first memory stores a first packet queue; and deleting at least one packet at the tail of the first packet queue from the first memory based on the available storage space of the first memory being less than the first threshold. When the available storage space of the first memory is less than the first threshold, a packet queue, namely, the first packet queue, is selected and a packet at the tail of the packet queue is deleted from the first memory.

Abstract: A packet forwarding method to shorten a transmission latency of an elephant flow is provided. In the method, for a first packet flow used as an elephant flow, a network device may receive a plurality of packets of the first packet flow, and determine a characteristic parameter of the first packet flow based on the plurality of packets, where the characteristic parameter of the first packet flow is used to indicate a transmission latency of the first packet flow. After determining the characteristic parameter of the first packet flow, the network device determines a forwarding policy of the first packet flow based on the characteristic parameter of the first packet flow. The forwarding policy of the first packet flow is used to indicate latency sensitivity of the first packet flow.

Abstract: A method for load balancing and packet re-sequencing on a network includes an ingress node that determines an identifier of a destination chip based on a destination address of a packet flow and a forwarding table, allocates, based on the identifier of the destination chip, a channel identifier and a sequence number of each packet that is in the packet flow and that enters a channel, to generate an updated packet flow, where each packet in the updated packet flow includes a packet in a corresponding packet flow, a sequence number of the packet, and the channel identifier, and the updated packet flow is sent in a per-packet load sharing manner.

Abstract: A method for determining a packet dequeue rate includes allocating a plurality of consecutive blocks in a first memory to a first packet, storing the first packet and a first length in the plurality of blocks, where the first length is of a first packet queue and is obtained when the first packet is enqueued into the first packet queue, and determining, based on a first span and the first length stored, a first rate at which a packet in the first packet queue is dequeued, where the first span is equal to a difference between a second time and a first time, the first time is when the first packet is enqueued into the first packet queue, and the second time is when the first packet is dequeued from the first packet queue.

Abstract: A solar generator can include a photon-enhanced thermionic emission generator with a cathode to receive solar radiation. The photon-enhanced thermionic emission generator can include an anode that in conjunction with the cathode generates a first current and waste heat from the solar radiation. A thermoelectric generator can be thermally coupled to the anode and can convert the waste heat from the anode into a second current. A circuit can connect to the photon-enhanced thermionic emission generator and to the thermoelectric generator and can combine the first and the second currents into an output current.

Abstract: A solar generator can include a photon-enhanced thermionic emission generator with a cathode to receive solar radiation. The photon-enhanced thermionic emission generator can include an anode that in conjunction with the cathode generates a first current and waste heat from the solar radiation. A thermoelectric generator can be thermally coupled to the anode and can convert the waste heat from the anode into a second current. A circuit can connect to the photon-enhanced thermionic emission generator and to the thermoelectric generator and can combine the first and the second currents into an output current.

Abstract: A solar generator can include a photon-enhanced thermionic emission generator with a cathode to receive solar radiation. The photon-enhanced thermionic emission generator can include an anode that in conjunction with the cathode generates a first current and waste heat from the solar radiation. A thermoelectric generator can be thermally coupled to the anode and can convert the waste heat from the anode into a second current. A circuit can connect to the photon-enhanced thermionic emission generator and to the thermoelectric generator and can combine the first and the second currents into an output current.