18 results about "Term memory" patented technology

A memory alignment system for efficient data transfer between a host system and a remote system comprises a data communications controller configured to align transmitted and received data based on formatting information received from the host system. When transmitting data from local system memory, for example over an Ethernet connection, communications control driver software may first write formatting information corresponding to the data into a configurable transmit data buffer. The data communications controller is operable to align the data based on the formatting information as the driver software moves the data into the transmit data buffer. Similarly, the driver software may write formatting information for receive data into a receive-format configuration buffer. The data communications controller may align the receive data based on the receive-formatting information as the receive data is being written into a configurable receive buffer. As the data communications controller performs all the required data alignment, no data alignment by the host processor is required.

Persistent Memory, byte-addressable non-volatile memory technologies, offer performance advantages and access similar to Dynamic Random Access Memory while having the persistence of disk. Hardware Transactional Memory support, originally designed for DRAM concurrency control, can corrupt persistent memory transactions due to cache evictions before system failure. Unifying storage and memory on the main-memory bus and accessed directly while using HTM for concurrency control has previously required the additional burden of changes to processors to prevent possible data corruption.

The present invention provides a solution for the durability of transactions to persistent memory while using HTM as a concurrency control mechanism, without any changes to processors or cache-coherency mechanisms. The invention includes a software only method and system that provides durability and ordering of HTM transactions to persistent memory. The invention also discloses a back-end memory controller that supports HTM transactions for durability to persistent memory without up-front processor changes.

The invention relates to a high-performance mobile terminal application software log collection system, which collects large-scale application software logs and executes the following steps that: S1,a mobile terminal obtains a collection node with the shortest communication delay and uploads logs to the collection node in the subsequent operation process; S2, the mobile terminal generates logs ina standard format; S3, the mobile terminal compresses the logs and then caches the logs in a local file; S4, the mobile terminal uploads the logs according to an uploading strategy; S5, a collectioncenter receives the logs uploaded by the mobile terminal and splices the received compressed logs to a memory cache block; S6, the collection center sends cache block size log data to a storage node according to a cache mechanism; S7, the storage center receives the log data sent by the collection center and encapsulates the log data into a log data file; and S8, the log data files between the storage centers are synchronized at regular time. The high-performance mobile terminal application software log collection system has the beneficial effects of low software and hardware cost, high resource utilization efficiency and simple management, operation and maintenance.

The invention provides an MPK technology-based inter-microkernel-module communication method and system, and a medium, and the method comprises the steps: an execution domain step based on MPK technology: controlling the read-write authority of a memory domain through the MPK technology, and separating the resources and states of execution domains from each other; a checkpoint function step of communication between the kernel modules, wherein a checkpoint function modifies the read-write permission of a memory domain at the current moment through a WRPKRU instruction, and switching of an execution domain is carried out; and in the execution domain to which the kernel module of the requested service belongs, requesting the service of the kernel module in the checkpoint function in the formof function call. According to the invention, the IPC performance overhead of a microkernel is effectively reduced, and the microkernel applying the invention has the same safety guarantee as the traditional microkernel through the MPK technology.

The invention discloses a method for skipping the functional software branches of a vehicle-mounted controller. The method comprises the steps of respectively placing a hardware test software and a control software in the different sections of a flash, so as to share a same RAM (Random Access Memory) memory unit in a time-sharing way under the premise that addresses are not conflicted; setting a group of skip combined signal input values as a condition for switching different function softwares, after the vehicle-mounted controller is electrified, reading the skip combined signal input values in a skip procedure, and then skipping to either the hardware test software or the control software according to the skip combined signal input values; and synthesizing the output files of the skipping procedure, the hardware test hardware and the control software to a target production file, and burning the target production file into the vehicle-mounted controller in the process of manufacturing. According to the method, the skipping of the function software branches (the hardware test software and the control software) can be realized in the inspection and evaluation of the vehicle-mounted controller hardware and the functional verification of the control software, so that the additional work caused by frequently burning the softwares can be reduced, and the production efficiency can be improved.

Embodiments of the invention relate to the technical field of data transmission, and provide a USB camera data acquisition method and device, electronic equipment and a storage medium. The method comprises the steps of monitoring a device state of a USB camera; if the USB camera is in an initialized state, creating a shared memory based on the memory writing service process; and based on the memory writing service process, reading image data acquired by the USB camera through a port of the USB camera, and storing the image data into the shared memory, so that each client process reads the image data from the shared memory. According to the USB camera data acquisition method and device, the electronic equipment and the storage medium, data distribution of the USB camera is achieved, data distribution can be achieved only through one-time data encoding and decoding in the single data acquisition process, the data acquisition efficiency is improved, and the performance loss of data acquisition is reduced.

The invention discloses a cross-partition data processing method and device. The method comprises the steps of: when a task in a first partition generates a cross-partition function call request, switching the memory access permission of the first partition to the initial memory access permission of the second partition, calling the target function, and recovering the memory access permission of the first partition to the initial memory access permission after calling is finished. According to the initial memory access permission, the task in each partition can only perform write operation onthe memory space corresponding to the partition, so that each partition only needs to occupy one register resource in the MPU, and the consumption of the register resources can be reduced; when the target function is called, the memory access permission of the first partition is switched into the initial memory access permission of the second partition, and after calling is completed, the memory access permission of the first partition is recovered to the initial memory access permission of the first partition, so that the target function can be prevented from unexpectedly accessing the memoryspace corresponding to the first partition, and the possibility of data security problems is reduced.

Providing memory management unit (MMU)-assisted address sanitizing in processor-based devices is disclosed. In one aspect, a processor-based device provides an MMU that includes a last-level page table that is configured to store page table entry (PTE) tokens for validating memory accesses, as well as fragment order indicators representing a count of page fragments for each memory page in the system memory. Upon receiving a memory access request comprising a pointer token and a virtual address of a memory fragment within a memory page of the system memory, the MMU uses the virtual address and the fragment order indicator of the PTE corresponding to the virtual address to retrieve a PTE token for the virtual address from the last-level page table, and determines whether the PTE token corresponds to the pointer token. If so, the MMU performs the memory access request using the pointer, and otherwise may raise an exception.

The invention discloses a satellite-borne DSP program external boot loading method and loading system and relates to a DSP program loading technology. According to an existing DSP program loading method, a program memory must be smaller than an internal RAM memory of a DSP; and if a hardware circuit or FPGA sequential logic related to external guidance fails, the guidance process fails, the DSP cannot work, and the like, the system is formed by sequentially connecting the DSP, the FPGA andan EEPROM. Wherein three parts of secondary bootloader programs are stored in the EEPROM; the FPGA servesas the key of DSP guiding loading two-out-of-three, an SPI communication protocol is achieved, the FPGA communicates with the DSP and the EEPROM, and two-out-of-three verification is conducted on three pieces of two-stage bootloader program data in the EEPROM according to bits. According to the guiding method, the internal guiding function and the external guiding function are achieved at the sametime, faults of external guiding circuit hardware or FPGA programs are prevented, the guiding modes of the remote control instructions can be flexibly switched through the FPGA, and normal operationof the user programs is guaranteed.

The invention discloses a ping-pong RAM-based strip array convolution module and an operation method thereof. The convolution module comprises a direct memory access module DMA, a DMA controller, a data distribution module, a ping-pong RAM, an RAM reading module, a strip array module, a block accumulation module and a register module. The DMA controller is used for controlling DMA to read data from a memory, and the data comprises a feature and a weight; the characteristic feature and the weight weight weight of the data are separted through a data distribution module, and then the data is stored into a ping-pong RAM; the RAM reading module is used for reading features and weights from the ping-pong RAM and sending the features and the weights to the strip array module for operation processing, and an operation result is output through the block accumulation module; and the register module is used for receiving a register configured by the central processing unit to control the convolution module to operate. According to the invention, the utilization rate of the multiplier can be improved, and convolution kernels of different sizes can be compatible.

The invention provides a process processing method, a process processing device and electronic equipment. The process processing method comprises the following steps: creating a sub-process based on atarget process, the parameter value of the specified parameter of the target process being a first parameter value, the specified parameter being used for indicating that: when the memory overflows,determining a process needing to stop running from the memory; setting a parameter value of a specified parameter of the sub-process as a second parameter value, the second parameter value and the first parameter value having a specified numerical relationship to enable the sub-process to be terminated to run prior to the target process when the memory overflows. In the mode, the parameter valuesof the specified parameters of the target process and the sub-process are set to have the specified numerical relationship, when the memory overflows, the numerical relationship can enable the sub-process in the service to be stopped running prior to the target process, and the memory is released by stopping the sub-process, so that the probability that the target process is stopped running can bereduced; and thus, the service operation stability is improved.

The invention discloses a data processing method and system based on efficient file storage and a terminal. The method comprises the steps that storage files are controlled to be named as dates, one directory is stored every day, one storage file is set for each device every day, and the storage file of each device is named as a device number and placed in the directory of the current date; the method comprises the following steps: presetting two temporary storage files, mapping the two temporary files into a memory through a function for storing GPS track data reported by equipment, and writing the GPS track data in the temporary files meeting a preset condition into an equipment file corresponding to a directory of a date corresponding to a directory structure according to the date and an equipment number; and when the GPS track data of the specified equipment needs to be inquired, traversing the specified GPS file by using the equipment number in units of days, taking out the data, sequencing and displaying, and reading the specified equipment file of the specified date. According to the invention, efficient storage and query of the GPS track data are realized.