Huawei Technologies Co., Limited

Tencent Technology (shenzhen) Company Limited

Research Institute of Petroleum Processing, Sinopec

China Petrochemical Corporation

Febris Bio-tech Limited

Dai, Huanqin

Zhang, Lixin

Smith, Garth Selwyn

Lixin Zhang

Ke Zhang

Xiaoyu Yu

Yuwei Wang

Rui Hou

Hua Li

Jianfeng Zhan

Chen Chen Zheng

Ning Zhang

Boyan Tu

Yi Zhang

Wei Chen

Wei Wei

Ye Li

Mingyu Chen

Bo Zhang

Qing Zeng

Shuanglong Chen

Zifang Guo

Lei Fan

Leyu Lin

Binzhang Fu

Feng Xia

Jianbin Xu

Yu Meng

Zhenhua Jing

Yungang Bao

Jin Xiong

Dejun Jiang

Lele Zhang

Jianping Zhu

Rui Ren

Yuan Rao

Jiuyue Ma

Zhe Feng

Jianlin Gao

Weimin Wu

Xiufeng Sui

Long Fu

Chongkang Tan

Ruen Wang

Garth Selwyn Smith

Xiaolan Shi

Wentao Bao

Xiangcong Zeng

Mu Yuan

Jianwen Liu

Ruihua Hou

Liuhang Zhang

Huanqin Dai

Biwei Xie

Jing Zhang

Hang Yin

Haibing Ding

Zunwei Li

Ying Chai

Jingwen Shi

Zhuo Kang

Yixing Gao

Xihan Zhang

This application discloses a pooling processing method, applied to a pooling processing system of a convolutional neural network. The method includes receiving a feature map comprising a width dimension and a height dimension; establishing a data region comprising k+m columns and n rows, k, m, and n being integers; transferring data blocks each of size k in width and size n in height from the feature map in a width-height sequential scanning order into columns m+1 to k+m and rows 1 to n of the data region in pooling cycles; and in each pooling cycle corresponding to a data block transferred from the feature map to the data region: retaining columns k+1 to k+m of the data region of a previous pooling cycle to columns 1 to m of the data region; performing a first level pooling within each column of the data region to obtain k+m first level pooling values; and performing P second level pooling values by pooling P sets of first level pooling values selected according to a position of the data block relative to a width boundary in the feature map, P being an integer.

Pooling processing method and system applied to convolutional neural network

This application discloses a pooling processing method, applied to a pooling processing system of a convolutional neural network. The pooling processing system includes a first storage device, a data region, a pooling computation kernel, and a pooling controller. The method includes: reading, by the pooling controller, k pieces of feature data from the first storage device in each reading cycle, the k pieces of feature data being components in a feature map generated by a convolution operation of the convolutional neural network, and k being an integer greater than 1; writing, by the pooling controller, the k pieces of feature data read from the first storage device into the data region, wherein the k pieces of feature data form one group among n groups of k pieces of data with each group arranged in a first dimension and the n groups arranged in a second dimension, wherein the n groups of k pieces of data are written into the data region in an updating cycle, wherein a duration of the updating cycle is n times a duration of the reading cycle, and wherein n cis an integer greater than 1; and transmitting, after the updating cycle is ended, data in the data region to the pooling computation kernel to perform a pooling operation, wherein the data in the data region comprises the n groups of k pieces of data and last m groups of data from a previous updating cycle with each group along the second dimension, wherein the last m groups of data are temporarily stored in the data region for use in pooling calculation by the pooling computation kernel in a next updating cycle. The technical solution in this application reduces the number of storage, numbers of reading and writing due to data reuses, and improves the efficiency of pooling processing.

Embodiments of the present invention disclose a result ranking method and device. The method includes: acquiring search key-information by using an interaction application, and searching a prestored interaction data set for at least one search result associated with the search key-information; ranking the at least one search result according to a quality assessment score corresponding to each of the at least one search result; and outputting the ranked at least one search result. The quality assessment score corresponding to each search result is a value generated according to a number of historical operations performed on each search result and an interactive influence score of an application identifier performing a historical operation on each search result.

Search result ranking method and device

The present disclosure describes methods, devices, and storage mediums for adjusting computing resource. The method includes obtaining an expected pooling time of a target pooling layer and a to-be-processed data volume of the target pooling layer; obtaining a current clock frequency corresponding to at least one computing resource unit used for pooling; determining a target clock frequency according to the expected pooling time of the target pooling layer and the to-be-processed data volume of the target pooling layer; and in response to that the convolution layer associated with the target pooling layer completes convolution and the current clock frequency is different from the target clock frequency, switching the current clock frequency of the at least one computing resource unit to the target clock frequency, and performing pooling in the target pooling layer based on the at least one computing resource unit having the target clock frequency.

Neural network operational method and apparatus, and related device

Embodiments of this application relate to a convolutional computing accelerator, a convolutional computing method, and a convolutional computing device, which belong to the technical field of electronic circuits. The convolutional computing accelerator includes: a controller, a computing matrix, and a first cache. The computing matrix comprising at least one row of computing units, each row of computing units comprising at least two adjacent connected computing units. The controller is configured to control input data of each row of computing units to be loaded into the first cache, and to control the input data loaded into the first cache to be inputted into the two adjacent computing units in a corresponding row. Each of the computing units in the corresponding row is configured to perform, in a first clock cycle, a convolutional computation based on received input data and a pre-stored convolutional kernel.

Convolutional computing accelerator, convolutional computing method, and computer-readable storage medium

Embodiments of this application disclose an instruction set processing method based on a chip architecture and apparatus, and a computer-readable storage medium. The method includes compiling a deep learning model based on the architecture of the chip, to obtain a deep learning instruction set corresponding to the chip; compressing the deep learning instruction set, to obtain a compressed instruction set; and storing the compressed instruction set in an instruction set buffer of the chip by writing in a register, the compressed instructions executing a task.

Instruction set processing method based on a chip architecture and apparatus, and storage medium

A computer includes a processor and multiple components directly or indirectly connected to the processor. Each component is coupled to a control device, and is configured to provide resources for one or more application programs running in the computer. The processor, when executing an application program, attaches a tag to an application request, and sends the tag-attached application request to a control device coupled with a component to which the tag-attached application request is directed. The control device receives the tag-attached application request, determines, based on the tag and a resource allocation policy, an amount of resources in the component and instructs the component to allocate the amount of resources and process the application request in accordance with the amount of resources allocated for the application request.

Computer system, and method for processing multiple application programs

A server connection method and system, which relates to the field of communications technologies, such that servers of a same specification are used to implement an optimal network, thereby reducing complexity of routing implementation which includes, providing ten servers, where each server includes five nodes, and the five nodes of each server are connected head-to-tail in series in a same connection manner, and connecting five nodes of any server in five of the ten servers to five nodes of each of the remaining five servers in a one-to-one correspondence manner, in order to form an optimal network of a Hoffman-Singleton graph.

Server connection method and system of a network

A method and an apparatus for determining a physical address are disclosed. According to the present disclosure, a page size is obtained according to the higher-order N bits of a linear address, where N is greater than 0 and less than a quantity of bits of the linear address; an index number of a translation lookaside buffer TLB is obtained according to the page size; a mask is obtained according to the page size and a supported minimum page size; a label of the TLB is obtained according to the mask; the higher-order MAC1 bits of a physical address corresponding to the linear address are obtained by searching the TLB according to the index number and the label; and the physical address is obtained according to the mask, the supported minimum page, and the higher-order MAC1 bits of the physical address.

Method and apparatus for determining physical address

A data caching method and a computer system are provided. In the method, when a miss of an access request occurs and a cache needs to determine a to-be-replaced cache line, not only a historical access frequency of the cache line but also a type of a memory corresponding to the cache line needs to be considered. A cache line corresponding to a DRAM type may be preferably replaced, which reduces a caching amount in the cache for data stored in a DRAM and relatively increase a caching amount for data stored in an NVM. For an access request for accessing the data stored in the NVM, corresponding data can be found in the cache whenever possible, thereby reducing cases of reading data from the NVM. Thus, a delay in reading data from the NVM is reduced, and access efficiency is effectively improved.

Data caching method and computer system

A memory management method implemented by a requesting node includes sending first indication information used for indicating a length of memory required by the requesting node and receiving second indication information used for indicating first remote memory provided to the requesting node by a target contributing node in at least one contributing node that can provide remote memory. The method also includes determining, from available virtual addresses, a first virtual address corresponding to the first remote memory, and sending a first data read/write instruction for the first data when first data whose pointer is within a range of the first virtual address needs to be read/written, where the first data read/write instruction includes third indication information, and the third indication information is used for indicating storage space, for storing the first data, in the first remote memory.

Memory management method, apparatus, and system

A method and an apparatus for accessing physical resources, is used to restrict access to physical resources of other light system kernel Light OSs by a first Light OS in a multi-kernel operating system and ensure security of accessing physical resources among the Light OSs. A method, executed by secure firmware, includes: receiving a physical address corresponding to a physical resource to be accessed by the first Light OS; determining whether the physical address corresponding to the physical resource is out of bounds; and if the physical address corresponding to the physical resource is within bounds, sending an access continuity signal to the first Light OS; or if the physical address corresponding to the physical resource is out of bounds, sending an access error signal to the first Light OS.

Method and apparatus for accessing physical resources

A data processing method and apparatus, where the method comprises an operating system loads a task code to a reserved hardware thread such that the reserved hardware thread executes the task code subsequently after receiving a hardware thread reservation request. Alternatively, in a process in which an operating system loads a task code to a hardware thread for execution, the hardware thread loads the small task code to a reserved hardware thread for execution without a need to create a thread for a task code corresponding to each task when the hardware thread reads a flag of a small task code.

Data processing method and apparatus for executing task code using reservation instruction and release instruction

A method is provided for treating an individual for a malarial infection, in particular drug-resistant species of, including administering to the individual in need of such treatment a composition comprising an extract containing an effective amount of one or more anti-malarial di- or tri-terpene compounds. Potent and effective extracts containing the anti-malarial compounds can be derived from Luo Han fruit andleaves, including combinations of the same, for the purpose of treating and preventing malaria in an inexpensive manner from readily available commodities. In one embodiment, the composition is administered orally in a solid or liquid ingestible form.

Compositions and methods for treating multi-drug resistant malaria

Growing community of inventors

Beijing, China

Lixin Zhang