8 results about "Kernel system" patented technology

Systems and methods for processing a plurality of images include obtaining input data including the plurality of images; providing the input data to a first machine learning model; providing an output of the first machine learning model to a second machine learning model and a third machine learning model; generating a first feature map corresponding to a plurality of kernels based on an output of the second machine learning model; generating a second feature map corresponding to a plurality of weights based on an output of the third machine learning model; generating a predicted kernel based on a weighted sum of the plurality of kernels; and generating output data based on the input data and the predicted kernel.

An information handling system comprising a hardware processor to execute computer-readable program code instructions of a first unprivileged artificial intelligence (AI) productivity-tool software process for operations of an AI productivity tool software module to detect receipt of audio / video user query input data and generate a first encryption key for transmission to an audio / video encryption driver and a second unprivileged AI productivity-tool software process. The hardware processor to execute computer-readable program code of the audio / video encryption filter driver to encrypt the audio / video user query input data with the first encryption key for storage in an allocated encrypted storage buffer portion of the kernel system memory and the for second unprivileged AI productivity-tool software process to access the audio / video user query input data from the allocated encrypted storage buffer portion of the kernel system memory with the first encryption key to determine a capability intent action responsive to a user query input.

The invention provides a safe starting method and upgrading method of an MCU (Microprogrammed Control Unit) and electronic equipment. A first kernel and a second kernel run in the MCU, the first kernel is used for running a hardware security module, the second kernel is used for running an application program, and the safe starting method of the MCU comprises the following steps: responding to an MCU reset signal, starting the first kernel, and performing primary verification on the application program by adopting a first verification algorithm; when the first verification is passed, a second kernel is started to execute a key task in the application program, and meanwhile, a second verification algorithm is adopted to conduct second verification on the application program; and when the secondary verification is passed, executing the target task in the application program, and judging that the MCU application program is successfully started. According to the method, the HSM kernel and the first verification algorithm are adopted to guarantee basic safety and quick start, so that the application kernel system can quickly process key tasks; and after starting, resources are utilized to complete full-quantity verification in parallel, so that the final safety is ensured while the speed is considered.

The application provides an NPU instruction optimization system and method for OpenCL heterogeneous computing. An OpenCL compiling and runtime layer is inserted between a host end and an NPU device. Standard OpenCL kernel code is deeply analyzed, and a neural network operator that can be optimized is identified and converted into a special intermediate representation designed by the NPU. The special intermediate representation is mapped to a special hardware unit and a kernel system of the NPU device to generate an instruction stream containing NPU special instructions. The application can efficiently map high-level operators and data patterns related to neural network computing that are implied in the standard OpenCL kernel code to the special hardware unit and the memory system of the self-developed NPU through identification, conversion and fusion technologies while supporting the OpenCL standard, so that the performance level of the NPU native private software stack is approached or even reached while the programming convenience of OpenCL is maintained.

A method is disclosed and includes determining a test plan to test a kernel on a mobile device, and determining an interaction input message according to the test plan, the interaction input message comprising first data. The method also includes transmitting the interaction input message comprising the first data to the mobile device over a network-based communication channel. The kernel in the mobile device generates the interaction output message in response to receiving the interaction input message. The method also includes receiving, from the mobile device, the interaction output message comprising second data from the mobile device over the network-based communication channel, and determining if the interaction output message is consistent with the test plan.

An information handling system comprising a hardware processor to execute computer-readable program code instructions of a first unprivileged artificial intelligence (AI) productivity-tool software process for operations of an AI productivity tool software module to detect receipt of audio / video user query input data and generate a first encryption key for transmission to an audio / video encryption driver and a second unprivileged AI productivity-tool software process. The hardware processor to execute computer-readable program code of the audio / video encryption filter driver to encrypt the audio / video user query input data with the first encryption key for storage in an allocated encrypted storage buffer portion of the kernel system memory and the for second unprivileged AI productivity-tool software process to access the audio / video user query input data from the allocated encrypted storage buffer portion of the kernel system memory with the first encryption key to determine a capability intent action responsive to a user query input.

The application discloses a kind of monitoring Linux full quantity directory file landing method, system, equipment and storage medium, the method includes: initialization fanotify;File is monitored based on fanotify in the file system in system;When first write event occurs, fanotify reports file modification event;The path of the file modification event is recorded to buffer area, and timeout time is recorded.Through the processing scheme of the present disclosure, in the case where kernel system is not intruded, full quantity directory high performance, high real-time monitoring effect can be achieved, and maintenance cost is low.

The invention provides an NPU instruction optimization system and optimization method for OpenCL heterogeneous computing, and the method comprises the steps: inserting an OpenCL compiling and runtime layer between a host end and NPU equipment, carrying out the deep analysis of a standard OpenCL kernel code, recognizing an optimizable neural network operator, and converting the optimized neural network operator into a special intermediate representation of NPU design; and mapping the special intermediate representation to a special hardware unit and a kernel system of the NPU equipment, and generating an instruction stream containing the NPU special instruction. On the premise of supporting the OpenCL standard, advanced operators and data modes which are implied in a standard OpenCL kernel code and are related to neural network calculation are efficiently mapped to a special hardware unit and a memory system of a self-developed NPU (Network Processing Unit) through identification, conversion and fusion technologies, so that the programming convenience of the OpenCL is kept, and the programming efficiency of the OpenCL is improved. And the performance level of the NPU native private software stack is approached or even reached.