4results about How to "Avoid truncation" patented technology

The application provides a kind of transmission layer data packet management method based on fragmentation and reorganization, belong to computer network technical field.Each device includes a server and multiple clients;The process of the first device transmitting messages to the second device is as follows: multiple clients of the first device transmit messages to a sending queue;The server of the first device monitors the sending queue, fragments the message data packet to obtain multiple sub-packets;The server of the second device receives the sub-packets sent by the server of the first device, reassembles the sub-packets, and sends them to a receiving queue;The clients of the second device block and wait for messages from the receiving queue, filter messages by analyzing the identity field in the message header, and process data packets that match their own identity.This method dynamically adjusts the packet size based on the network environment state, avoiding the packet loss or truncation problem caused by MTU limitation on high latency links in the traditional fixed packet mode.

The application discloses a kind of spherical granular material slice image generation and binarization method, belong to granular material structure analysis and image processing technical field, including: first read the three-dimensional coordinates of sphere center and radius data, scale uniformity is realized by coordinate mapping standardization and radius mapping, construct three-dimensional voxel grid and generate particle distribution field;Again, extract slice along coordinate axis, complete binarization using three-dimensional interpolation and mean threshold, finally output TIFF image and DAT numerical file.The application adopts the above-mentioned spherical granular material slice image generation and binarization method, without physical sample and CT scanning, not limited by resolution, without partial volume effect, particle boundary is clear, can adaptively process arbitrary coordinate distribution data, slice is smooth, high precision, can provide efficient and accurate basic data for granular material pore structure analysis, seepage and mechanics numerical simulation, applicable to material science and geotechnical engineering and other fields.

The application relates to the technical field of power equipment monitoring, and discloses a power equipment multi-source monitoring data fusion processing method and system, which comprises the following steps: receiving and sending data streams and performing time axis alignment, separating main features, secondary features and basic environment data streams; monitoring the main feature data stream to generate event anchor points, and combining spatial conduction parameters to generate a reference physical time delay; correcting the reference physical time delay by using the basic environment data stream to generate a medium state time delay drift window; mapping the medium state time delay drift window into a target logical block interval set, encapsulating the logical moving pointer and binding it to the event anchor point landing disk; searching and extracting the underlying data based on the interval set, and performing time axis cutting on the decompressed data by using the medium state time delay drift window. The application corrects the physical wave speed offset in combination with the running working condition, and decouples the time addressing from the underlying physical storage, thereby reducing the index maintenance cost caused by data reorganization and improving the precision of multi-source data fusion alignment.

This invention discloses a GPU code policy-driven generation method and system based on parallelism inference constraints. It collects cross-iteration memory access address features of variables using GDB, constructs a spatiotemporal program graph, and applies a spatiotemporal graph convolutional network for modeling. This yields prediction results for loop parallelism types and a five-dimensional optimization policy prediction vector. For serial loops, it directly returns the original C code. For parallelizable loops, the prediction results are converted into natural language parallelism labels and optimization instruction sets, which are then concatenated in a fixed order to construct structured hints. This drives a large language model as a syntax constraint executor to generate target CUDA code. This invention explicitly decouples parallelism inference from code generation, eliminating erroneous parallelization of the large language model and misjudgment of optimization policies, significantly improving the accuracy and execution performance of the generated code. It can be widely applied in the GPU migration of high-performance computing programs and heterogeneous computing fields.