Zte Corporation

Shenzhen Dapu Microelectronics Co., Ltd.

Tencent Technology (Shenzhen) Company Limited

Dapustor Corporation

Andritz China Ltd

Weijun Li

Bo Dai

Guanghui Yu

Yan Wang

Yijian Chen

Xiang Chen

Xun Yang

Zhisong Zuo

Li Jiang

Xin Wan Wu

Yafei Yang

Yunxin Huang

Sheng Che Chen

Pengfei Huang

Jian Zhong

Hongbo Wan

Wenjiang Li

Jixing Zhang

Yuanbing Liu

Florian Steinwender

Rongjun Qin

Xin Wu

A hardware accelerator for running an instruction set of a recurrent neural network, a data processing method, a system-level chip, and a medium are provided. The hardware accelerator is configured to process the instruction set. The instruction set includes: a data flow control instruction used for performing data flow control; a general-type computing instruction used for performing general-type computation to implement general-type computation in the recurrent neural network; a special-type computing instruction used for performing special-type computation to implement special-type computation in the recurrent neural network; an exponential shift instruction used for performing exponential shifting to implement data normalization during computation of the recurrent neural network; and a data transfer instruction used for performing data transfer to implement data transfer operations between different registers as well as data transfer operations between registers and memories during computation of the recurrent neural network.

Hardware accelerator for execution of instruction set of recurrent neural network, data processing method, system-level chip, and medium thereof

A data error correction method, apparatus, device, and readable storage medium are disclosed. The method includes: acquiring target data to be error-corrected; performing error correction on the target data using an error-correcting code to obtain first data; judging whether the performing of the error correction on the target data is successful; responsive to the performing of the error correction on the target data being not successful, correcting the target data using a target neural network to obtain second data, determining the second data as the target data, and continuing to perform the error correction on the target data again; and responsive to the performing of the error correction on the target data being successful, determining the first data as the error-corrected target data.

Data error correction method, apparatus, device, and readable storage medium

A data processing method, apparatus, device, and readable storage medium are provided. The method includes: acquiring historical I/O data (S), where the historical I/O data is data of a solid-state drive that is accessed within a preset time period; using a prediction model to learn the historical I/O data to obtain a prediction result (S), where the prediction result includes a data intensity of the solid-state drive to be accessed within a future window period, and the future window period is determined according to a cycle in which the solid-state drive is accessed; and managing the solid-state drive according to the prediction result (S).

Data processing method, apparatus, device, and readable storage medium

Disclosed is a method for enhancing service quality of a solid state drive and the solid state drive, and the method includes the steps of obtaining a remaining storage resource corresponding to a write I/O request periodically, and determining a resource waiting time of the write I/O request based on a resource consumption rate of the write I/O request. The method also includes setting a maximum scheduling delay time for the write I/O request according to the resource waiting time of the write I/O request. The method also includes identifying a type of unprocessed I/O requests and obtaining an arrival time of the unprocessed I/O requests, and sorting the unprocessed I/O requests in combination with the maximum scheduling delay time. The method further includes processing the unprocessed I/O requests according to the sort of the unprocessed I/O requests.

Method for enhancing quality of service of solid-state drive and solid-state drive

Disclosed is a method for data processing applied to a solid state drive, a computer device and a computer-readable storage medium. The method includes acquiring an interface protocol command received by the solid state drive. The method also includes parsing the interface protocol command to obtain I/O information from the interface protocol command. The I/O information includes at least an I/O timestamp, an I/O type, and an I/O size. The method further includes invoking machine learning based on the I/O information to predict I/O information of a first future time period, so that a processor of the solid state drive is configured to proactively execute management functions according to the prediction results.

Data processing method and apparatus for solid state disk

The present invention relates to a process for manufacturing a Yankee cylinder. The process comprises welding the first reinforcement flanges of two workpieces of Yankee cylinder section from inside of the cylinder case, preheating and welding the two workpieces of Yankee cylinder section from outside of the cylinder case, and then removing the first, second reinforcement flanges, radial inner end area of the weld and their surrounding material in the workpieces of the Yankee cylinder section, so that the joint of the two workpieces of Yankee cylinder section is machined to meet the final dimension requirements for the Yankee cylinder inner grooves. The present invention is also related to a workpiece of Yankee cylinder section used in the said process.

Workpiece of Yankee cylinder section and process for manufacturing a Yankee cylinder

A method for predicting logical blocks address (LBA) information, including: receiving, by a Solid State Drive (SSD), a trace sent from a host, wherein the host can acquire the trace in a reusable environment; determining, by the SSD, one or more LBAs received by the SSD according to the trace; obtaining, by the SSD, a distribution of the LBAs by learning the LBAs based on a preset learning algorithm; and predicting, by the SSD, one or more subsequent LBAs based on the distribution of the LBAs. As a result, it can perform heat classification and prediction of the following LBA used in the SSD by means of learning the LBA distribution of the SSD in a certain reusable environment of the host, thus to improve the hit rate of reading and writing and the efficiency of classification of hot and cold data in garbage collection.

Method for predicting LBA information, and SSD

The present disclosure provides a method, system, client, server, and storage medium for creating an instant messaging discussion group. The method may include: sending, by a first client, a request for creating an instant messaging discussion group to a server, and receiving a creation identifier allocated by the server; broadcasting, by the first client, the creation identifier, so that another client other than the first client in an instant messaging application system initiates authentication from the server according to the creation identifier; and receiving, by the first client, a successfully authenticated client list delivered by the server, and selecting at least one client from the client list to create an instant messaging discussion group. The present disclosure provides improved process of creating an instant messaging discussion group, increases creation efficiency, and enhances convenience for a user to use an instant messaging application.

Method, system, storage medium for creating instant messaging discussion group

The present disclosure discloses a method and system for transmitting downlink control information, for solving the problem of how to indicate a transmission scheme when a cyclic redundancy check code in the downlink control channel is scrambled with the semi-persistently scheduled cell radio network temporary identifier. In order to support the bi-flow beam forming technology in the LTE-A R9, the present disclosure provides a downlink control information Format 1A, which may, when the cyclic redundancy check code in the downlink control channel is scrambled with the semi-persistently scheduled cell radio network temporary identifier, indicates a method for transmitting downlink control information, for example, single-layer transmission, transmit diversity, or a transmission scheme simultaneously involving single-layer transmission and transmit diversity, which guarantees flexibility of system schedule. The present disclosure adds no signaling overhead and has good adaptability with the LTE R8.

Method and system for transmitting downlink control information

A method for sending and detecting downlink control information is provided. A base station carries the downlink control information in a physical downlink control channel, and sends said physical downlink control channel through component carriers. Said base station bears first type downlink control information on an anchor component carrier, and bears third type downlink control information on a prime component carrier, wherein said first type downlink control information bears one or more types of uplink scheduling information, downlink scheduling information and uplink power control information of its component carrier, said third type downlink control information bears indication information of the second type downlink control information which bears the uplink scheduling information and/or the downlink scheduling information of one or more component carriers. A terminal detects the first type downlink control information on the anchor component carries and detects the third type downlink control information on the prime component carrier.

Method for sending and detecting downlink control information

The present disclosure discloses a method and system for transmitting a position reference signal. Wherein the method includes: frequency domain positions for transmitting a position reference signal are n physical resource blocks, and the value of n is obtained according to signaling; time domain positions for transmitting the position reference signal are the remaining orthogonal frequency division multiplexing symbols in a subframe, except for orthogonal frequency division multiplexing symbols for transmitting a physical downlink control channel and orthogonal frequency division multiplexing symbols for transmitting a cell-specific reference signal; and the position reference signal is transmitted according to the frequency domain position and the time domain position. By adopting the method and the system of the present disclosure, the transmitting of the position reference signal is realized, and the positioning precision of a UE is ensured.

Method and system for transmitting position reference signal

The present disclosure provides a method and system for transmitting a position reference signal, which includes: acquiring a currently needed Position Reference Signal (PRS) sequence with a length of 2&#xd7;N, wherein Nis the PRS bandwidth configured by high layer signaling, which is represented by a unit of a resource block; determining the position of the PRS sequence in a physical resource block; and transmitting the acquired PRS sequence on the determined position. With the method of the present disclosure, a currently needed PRS sequence is acquired directly, or at first, a PRS sequence with the maximum length is generated, and then the currently needed PRS sequence is intercepted from the PRS sequence with the maximum length. The present disclosure ensures that an effective PRS sequence can be acquired in any circumstance, thereby ensuring the realization of the positioning function of the PRS.

The present invention discloses a method and a system for transmitting downlink control information, wherein when a base station is not configured with a carrier indicator field, the base station transmits a physical downlink control channel bearing the corresponding downlink control information on each downlink component carrier in a user equipment's physical downlink shared channel (PDSCH) carrier set, and the user equipment detects the corresponding downlink control information on each downlink component carrier in the configured PDSCH carrier set. When a base station is configured with a carrier indicator field, the base station transmits the physical downlink control channel bearing the corresponding downlink control information on each downlink component carrier in the configured physical downlink control channel (PDCCH) carrier set, and the user equipment detects the corresponding downlink control information on each downlink component carrier in the PDCCH carrier set.

Growing community of inventors

Shenzhen, China

Weijun Li