31results about How to "Large cache capacity" patented technology

The invention discloses a method for storing log data and a log data storage system. The method comprises the steps of acquiring a log ID (Identity) and a log data structure which are distributed to a service log to be stored; enabling a log report agent, and interlinking a previously-obtained API (Application Program Interface) in a service program; calling the API in the service program and passing in parameters of the log ID and each field of the log data structure corresponding to the log ID; encapsulating the log data and outputting the encapsulated log data to a shared memory for cache by the API; firstly reading the log data in the shared memory, analyzing information such as the log ID, contained in the log data, and writing the read log data into a cache file according to the information by the log report agent; and reporting the data in the cache file to a log server to store by the log report agent through the network connection with the log server. According to the application of the method for storing the log data and the log data storage system, disclosed by the invention, a cross-business integrated log service system is created, thus the resource use efficiency and the reliability of log data storage are improved.

To avoid degradation of performance of a storage apparatus even when access biased to writing or reading of data is performed in a buffer memory that adopts a dual ring buffer system as a data structure. A sector buffer for caching an access request is provided in a DRAM of the storage apparatus. The sector buffer has a first ring buffer (BUF_W) that temporarily holds an external write request for a recording medium in advance of its process to the recording medium, and a second ring buffer (BUF_R) that retains data read from the recording medium according to an external read request for the recording medium. An MPU of the storage apparatus shifts a BPAGE of the BUF_R or BUF_W according to external access patterns to the recording medium to thereby change the buffer sizes of the BUF_R and BUF_W.

The invention discloses a three-layered cooperative type caching method of a fog wireless accessing network. The three-layered cooperative type caching method comprises the following steps: (1) dividing the fog wireless accessing network into a fog wireless accessing node layer and a cloud node layer; (2) regulating forwarding routing of a file request in the network; (3) calculating a request arriving rate of each file by a fog wireless accessing node through the received file request, and calculating the weight of each file in a knapsack algorithm according to the request arriving rate of each file; finally, solving optimal caching distribution in a fog wireless accessing point cluster by adopting the knapsack algorithm; (4) calculating the request arriving rate of each file by a clusterhead through the received file request; calculating the weight of each file in the knapsack algorithm according to the request arriving rate; finally, solving optimal inter-cluster cooperation caching distribution by adopting the knapsack algorithm; (5) calculating a file arriving rate by a cloud node through the received file request, and calculating the weight of each file in the knapsack algorithm according to the request arriving rate; finally, solving cloud node layer optimal caching distribution by adopting the knapsack algorithm. According to the three-layered cooperative type cachingmethod disclosed by the invention, a returning chain load can be effectively reduced.

The invention discloses a high speed large capacity data collection system which is based on CPLD and SDRAM. Analog signals are inputted into the signal input port of suited input network circuit(1), the signal output port of the (1) is connected to the signal input port of magnifying and buffering circuit (2), the signal output port of the (2) is connected to the analog signal input port of analog /digital transforming circuit (3), the data output port of the (3) is connected to the data input port of CPLD control unit circuit (4), the time signal input port of the (3) is connected to the time signal output port of the (4), the data output port of the (4) is connected to the data input port of the data input port of interface circuit(5), the data output port (4) is connected to the data input and output ports of (6). The highest sampling frequency is 100MHz, storing capacity is 256MBit, and it has several excellences such as big buffer capacity, expansible, high integrity, high reliability, high antijamming capability and so on, it can be used in high speed testing and controlling system, and image collecting and dealing system.

The invention relates to a switchboard based extensible DSPEED-DSP_Q6474 signal processing board, which belongs to the field of high-speed real-time signal processing. The processing board comprises a single-core DSP processing node, four three-core DSP processing nodes, an FPGA (Field Programmable Gate Array) processing node, a serial Rapid IO (Input-Output) switchboard module and a CPCI 6U standard board type, wherein the single-core DSP processing node consists of a TMS320C6455 and a set of DDR2 SDRAMs of which the volume is 512MB; the three-core DSP processing node consists of a TMS320C6474 and a set of DDR2 SDAMs of which the volume is 512MB; and the FPGA processing node consists of an XC5VSX95 or an XC5VLX110T and two sets of DDR2SDRAMs of which the volume is 512MB. The invention solves the problems of small data exchange bandwidth between the traditional high-speed real-time signal processing board processors, non-uniform interface protocols and poor intra-board extendibility.

The invention discloses a dual-layer distributed caching method of a fog wireless access network. The method orderly comprises the following steps: dividing a fog network into a F-AP (fog wireless access node) layer and a UE (user equipment) layer, wherein each of the F-AP and UE collects file application information from the around UE, and computes the field popularity according to the information; solving the optimal caching distribution of the UE by using a knapsack algorithm; and solving the optimal caching distribution of the F-AP layer by using a BP algorithm, wherein the UE decides theUE in D2D connection with the same by querying whether the around UE is receiving the file. By importing the D2D connection in the fog network, the dual-layer caching is realized, and the caching decision can be made by adopting the distribution method as long as each of the F-AP and the UE collects the local information; compared with the traditional caching algorithm, the download delay is obviously reduced, and the convergence can be finished within ten times of iteration, the complexity is low, and meanwhile, the higher robustness is provided.

The present invention provides a disk protection system based on write filtering technology, which includes an application layer, a support library, a driver layer, and an overlay layer sequentially connected from top to bottom, and the driver layer is respectively connected to the protected disk by communication. Partition and configuration information library, the application layer is the interface for the user to operate and manage the protection system, the support library is the encapsulation of the application layer, and the driver layer is used to map the write filter data to the disk To the overlay layer and for the selection and maintenance of the overlay layer, the overlay layer is used to store write filtering data, the driver layer uses the driver device of Windows Disk Filtering, and a device is created in the driver layer " Iped Device". The invention can realize the disk partition protection function on the Windows XP and Windows 7 operating systems, and at the same time solve the problem of being limited by the size of the physical memory.

The invention provides a radar intermediate-frequency receiving and signal processing board based on TM320C6678. The board comprises a power supply module, two DSP (Digital Signal Processor) processing nodes, an FPGA (Field Programmable Gate Array) processing node, three high-speed ADC (Analog to Digital Converter) modules, a high-speed DAC (Digital to Analog Converter) module, an LVDS (Low Voltage Differential Signaling) data acquisition output module, three low-speed ADC modules, two low-speed DAC modules, two isolated synchronous serial port modules and two FLASH modules. The processing board adopts a high-performance eight-core DSP and Xilinx high-performance FPGA, so that the board has the advantages that the processing capacity is high and the integration level of various functional chips is high.

The invention provides a message buffer method and device of Ethernet equipment. The method comprises that a first device as a buffer server in an Ethernet transmits a Buffer Client querying message via enabling ports of the first device itself, receives a Buffer Client registering message replied by any other device via any enabling port of the first device itself, and adds the other device, as a Buffer Client, into a distributed buffer group of reception ports of the registering message; and when the first device detects that the amount of messages received by any port with the distributed buffer function in unit time reaches a preset shunting threshold, the first device forwards a message entering the port to a buffer client selected from the distributed buffer group of the port. Thus, packet loss is prevented from Ethernet equipment.

The invention discloses an integrative radio wave perspective exploration instrument, and belongs to a geophysical radio wave perspective exploration instrument. The integrative radio wave perspective exploration instrument comprises an emitting module, a receiving module, a man-machine exchange module and a communication module, wherein the emitting module is used for emitting a radio wave signal; the receiving module is used for receiving the radio wave signal; the man-machine exchange module is used for setting the system parameter, displaying and storing data in real time, and transforming the collected signal; the communication module is used for undertaking the task of the communication between the radio wave perspetograph and a microcomputer. By mutually sending signal, the integrative radio wave perspective exploration instrument can realize the underground communication so as to determine whether the measurement work is started. After the measurement is finished, the receiving end becomes the emitting end, the emitting end becomes the receiving end; then the measurement process can be conducted again, the detection is conducted from both parts of a lane, the result is comprehensively analyzed; therefore, the shape of abnormal body can be detected in all rounds. The integrative radio wave perspective exploration instrument is large in buffer memory size, high in integration level, strong in reliability, strong in anti-jamming ability, rapid and exact in collecting process.

The embodiment of the invention provides a machine learning reasoning coprocessor. The machine learning reasoning coprocessor comprises a reasoning unit and a control unit. The reasoning unit is usedfor reading and calculating a task instruction, data and parameters so as to realize reasoning operation corresponding to the task instruction; and the control unit comprises a main control unit and aplurality of channel control units. Wherein the main control unit is used for realizing global control on the machine learning reasoning coprocessor; and each channel control unit in the plurality ofchannel control units is used for respectively controlling a channel corresponding to the reasoning unit in the reasoning unit to realize response to a single task or a user according to the task instruction. According to the embodiment of the invention, multiple users and multiple tasks can be supported, and flexible customized design can be realized.

To provide a cache system that can dynamically change a cache capacity by memory areas divided into plural. The cache system includes a line counter that counts the number of effective lines for each memory area. The effective line is a cache line in which effective cache data is stored. Cache data to be invalidated at the time of changing the cache capacity is selected based on the number of effective lines counted by the line counter.

The embodiment of the invention provides a data read-write method, a terminal and a computer readable storage medium, and the method comprises the steps: receiving a data reading instruction which comprises the feature information of to-be-read data; determining a hit data index in a memory according to the feature information; and reading data in a cache according to the hit data index, the cachebeing a solid state disk SSD.