Patents
Literature
Patsnap Eureka AI that helps you search prior art, draft patents, and assess FTO risks, powered by patent and scientific literature data.

306 results about "CUDA" patented technology

CUDA (Compute Unified Device Architecture) is a parallel computing platform and application programming interface (API) model created by Nvidia. It allows software developers and software engineers to use a CUDA-enabled graphics processing unit (GPU) for general purpose processing — an approach termed GPGPU (General-Purpose computing on Graphics Processing Units). The CUDA platform is a software layer that gives direct access to the GPU's virtual instruction set and parallel computational elements, for the execution of compute kernels.

GPU computing power scheduling method based on one-cloud multi-core heterogeneous computing power platform

The invention provides a GPU computing power scheduling method based on a one-cloud multi-core heterogeneous computing power platform, and the method comprises the following steps: S1, carrying out heterogeneous resource registration and modeling, accessing hardware equipment containing multiple types of GPUs through a resource registration module, collecting the equipment model, the video memory capacity and performance index data, and carrying out heterogeneous resource modeling; constructing a resource feature database containing a topological relation, and supporting hybrid access of chips; s2, virtualized resource reconstruction: pooling a physical GPU into virtual GPU resources by adopting a hardware abstraction layer technology, realizing video memory isolation and calculation unit division through a containerization technology, and configuring each virtual GPU instance with an independent drive stack and a security sandbox; and S3, submitting a multi-modal task, receiving a CUDA / OpenCL calculation task submitted by a user, analyzing task demand parameters including a calculation core number, a video memory occupation amount and a data throughput threshold, and generating a task descriptor containing a priority label.
Owner:SAISI TECH (XIAN) CO LTD

Method and system for realizing CUDA call tracking based on eBPF

InactiveCN120723587AHardware monitoringCall tracingSoftware engineering
The invention provides a method and system for achieving CUDA call tracking based on an eBPF, and relates to the field of GPU call tracking of the eBPF. The method comprises the following steps: dynamically mounting an eBPF program at an API function entry and an API function exit of a user mode CUDA runtime library through an uprobe technology of the eBPF; the eBPF program captures metadata of the CUDA API calling event in a kernel mode and packages the metadata into a predefined structural body; asynchronously transmitting the structural body data to a user mode through an annular buffer area; and calling a standard function library by a user mode program to load an eBPF program, reading data in the annular buffer area in real time, and completing analysis and visual display of a CUDA calling event. According to the method, the calling parameter, the return value and the nanosecond timestamp are effectively captured, and the observability and the debugging efficiency of the GPU application program are improved.
Owner:JINAN INSPUR DATA TECH CO LTD

Image-fused end-side cloud collaborative intelligent fire-fighting fire monitoring system

The invention discloses an end-side cloud collaborative intelligent fire-fighting fire monitoring system based on image fusion, and relates to the technical field of intelligent fire-fighting, the system is composed of a plurality of functional modules, and the system comprises a multi-modal image fusion module which generates a dynamic scanning priority map based on prior data, distinguishes a natural heat source from an abnormal fire by using a dual-light fusion algorithm, and sends an image fusion result to a cloud server; a scanning area is divided according to the thermal risk grade, and the thermal imaging resolution is dynamically adjusted; the distributed edge computing module is used for carrying out space-time synchronization on cross-modal data through a multi-modal feature alignment network, and carrying out dynamic allocation on a CUDA core and CPU resources through adaptive computing scheduling; an improved artificial bee colony algorithm is adopted, the bandwidth of the multi-sensor data flow is dynamically allocated through a time-sharing multiplexing protocol, and three-dimensional path planning is carried out; and the end-side cloud collaborative decision module constructs a federated learning driven model sharing network, and each edge node trains a lightweight YOLOv5s pruning model based on local data.
Owner:HANGZHOU ZIPENG TECH CO LTD

Method for improving reasoning solution and training speed of deep operator network model

The invention belongs to the technical field of artificial intelligence and high-performance computing, and particularly relates to a method for improving the reasoning solution and training speed of a deep operator network model. According to the method, multi-level optimization is carried out on the PI-DeepONet model, and the reasoning solving speed and the training speed of the PI-DeepONet model are improved. According to the method, the SGEMM operator is customized, and compared with the best single-precision matrix multiplication in a basic linear algebraic library of a CUDA (Compute Unified Device Architecture) computing platform, the SGEMM operator can obtain a speed-up ratio of 1.5 times at most in a common model size range of PI-DeepONet. A fusion operator is designed according to the special structure of the modified MLP used in the model, the speed-up ratio of the end-to-end reasoning speed of the model is more than two times, and the calculation time of a graphics processor can be shortened by 5.7 times at most. According to the method, data parallel is used for carrying out distributed training on the model, a communication concealment strategy is used, and the expandability is kept at 83%-85%.
Owner:CHINA UNIV OF PETROLEUM (EAST CHINA)

Inference method and device, equipment and medium

The invention provides a reasoning method and device, equipment and a medium. The method comprises the steps that when a reasoning request is received, model parameters of a target model deployed on each GPU are obtained, the reasoning request comprises information of a reasoning task, and the information of the reasoning task comprises input data; then, according to the structure information of the target model, segmenting the reasoning task into a plurality of groups of subtasks corresponding to the GPU according to a combination mode of a calculation task and a communication task; segmenting the input data into input sub-data corresponding to each group of sub-tasks; afterwards, generating a CUDA flow corresponding to the GPU according to a sequence dependency relationship between the calculation task and the communication task in each group of subtasks, the CUDA flow being used for realizing parallel execution of multiple groups of subtasks; then, a CUDA kernel on the GPU is scheduled according to the CUDA flow, so that the GPU executes multiple groups of subtasks in parallel according to model parameters and input subdata, and a reasoning result corresponding to the reasoning request is obtained. According to the invention, the reasoning time delay can be reduced.
Owner:ZHONG KE JIA HE (BEI JING) KE JI YOU XIAN GONG SI

GPU program optimization method based on CUDA parallel environment

The invention relates to the technical field of GPU (Graphics Processing Unit), in particular to a GPU program optimization method based on a CUDA (Compute Unified Device Architecture) parallel environment, which comprises the following steps: acquiring a task request sent by a target GPU program, and carrying out priority queue classification on the task request based on a priority coefficient of the task request so as to divide the task request into different priority queues; the activity coefficient of each priority queue is calculated based on the delays of the different priority queues, resource scheduling is carried out on each priority queue through each calculation node in the calculation cluster based on the activity coefficient of each priority queue, and each calculation node is composed of a CPU and a GPU. According to the method, the task requests are subjected to priority queue classification, so that the tasks with high priorities can be preferentially processed, the response speed and efficiency of the whole system are improved, the tasks with different priorities can be effectively managed through task priority queue classification, resource conflicts are avoided, and the processing capacity of key tasks is improved.
Owner:无锡九方科技有限公司

Method for converting C + + code into CUDA-C code

The invention relates to the technical field of computers, and discloses a method for converting a C + + code into a CUDA-C code, and the method comprises the steps: receiving a C + + source code, carrying out lexical analysis and grammatical analysis, generating an abstract syntax tree, and carrying out semantic analysis to ensure that the source code is correct semantically; analyzing the abstract syntax tree, identifying a data dependency relationship among the calculation tasks, and determining loop statements and conditional statements in the source code; and converting a loop statement and a conditional statement in the source code into a kernel function, a thread block and a thread management code corresponding to the CUDA, and further generating a corresponding compilable source file. According to the method, the abstract syntax tree of the C + + code is analyzed, the parallelism in the abstract syntax tree is analyzed, the code is optimized, the CUDA-C efficient code is generated, then efficient automatic conversion of the code is achieved, and the method has great significance in accelerating program performance and shortening the development period.
Owner:SHANDONG INSPUR SCI RES INST CO LTD

Method and system for generating and optimizing CUDA (Compute Unified Device Architecture) code based on multi-dimensional feature search and enhancement

The invention provides a CUDA code generation and optimization method and system for multi-dimensional feature search and enhancement, and the method comprises the steps: generating an initial CUDA code through a large language model based on task description and GPU hardware parameters; executing triple verification and a feedback mechanism on the initial CUDA code; according to feedback information of triple verification, optimizing CUDA codes through multi-dimensional feature search; based on error information, adjusting large language model input to regenerate codes; dynamically selecting an optimization strategy based on a performance index, wherein the optimization strategy comprises a thread block size and a memory access mode; and iteratively executing the steps until the CUDA code which passes triple verification and meets the target performance is generated. According to the method, the code features of the CUDA code are analyzed, and a triple verification and feedback mechanism including compilation feasibility, logic correctness and performance during execution is formed, so that an automatic evaluation process of the CUDA code is realized, and the problems that the CUDA code optimization process is difficult to be automated and an optimization target is lacked in the CUDA code optimization process are solved.
Owner:SHANGHAI JIAOTONG UNIV

High-fidelity lightweight world model construction method for end-to-end automatic driving test

The invention relates to a world model construction method, in particular to a high-fidelity lightweight world model construction method for an end-to-end automatic driving test, which is used for constructing a high-fidelity world model and performing knowledge distillation on the world model to solve the problems of huge parameters, low reasoning efficiency and the like of the world model. Model parameters are reduced on the basis of reserving the world model generation capability, and the reasoning efficiency is improved; a CUDA operator is developed in a user-defined mode for the bottleneck part of world model calculation, video memory allocation is optimized, and the reasoning efficiency of a high-fidelity world model is improved based on a single-device multi-thread scheduling and multi-device cooperative calculation method. According to the method, the high-fidelity lightweight world model for the end-to-end automatic driving test can be constructed, the problems that an existing world model is low in multi-modal information alignment precision, poor in cross-view and cross-frame consistency, low in reasoning efficiency and the like are effectively solved, the confidence coefficient of the end-to-end automatic driving system test process is improved, and the test efficiency of the end-to-end automatic driving system is improved. And the testing efficiency of the end-to-end automatic driving system is greatly improved.
Owner:JILIN UNIVERSITY

Data transmission efficiency improvement method and device, equipment and storage medium

The invention provides a data transmission efficiency improving method and device, equipment and a storage medium, and the method comprises the steps: carrying out the equipment recognition and address space mapping of expansion card equipment connected with a PCIe interface on a GPU, and obtaining GPU data transmission configuration information; according to the configuration information, carrying out storage space division and transmission channel establishment on the expansion card equipment to obtain a GPU data transmission channel configuration result; scheduling and distributing the GPU data to be transmitted based on the configuration result and a preset CUDA (Compute Unified Device Architecture) shared storage mechanism to obtain a data transmission execution scheme; and starting, controlling, monitoring in real time and dynamically adjusting the transmission process of the data to be transmitted according to the execution scheme and a preset RAID (Redundant Array of Independent Disks) disk array. According to the method, the expansion card equipment and the RAID disk array are introduced, and a CUDA shared storage mechanism is combined, so that the available storage space of the GPU is effectively expanded, the data transmission process is optimized, the data transmission efficiency when the GPU processes a large-scale data set is remarkably improved, and meanwhile, relatively low hardware cost is kept.
Owner:SHENZHEN QUANXING TECH CO LTD

Underwater target detection method and system based on edge calculation, and related equipment

The invention relates to the technical field of ocean intelligent analysis, and particularly discloses an underwater target detection method and system based on edge computing and related equipment.The method comprises the following steps that edge computing equipment is configured, and the edge computing equipment comprises an embedded AI development board and a binocular camera; an underwater target image is extracted based on a binocular camera, and the underwater target image is processed based on an embedded AI development board; obtaining an image processing model adaptive to the real-time change of the underwater target image based on model training and optimization; a YOLOv5 model is deployed on an NVIDIA Jetson Nano development board, an NVIDIA CUDA model is utilized to accelerate the reasoning process, an underwater target image is analyzed in real time, and preprocessing parameters are adjusted to adapt to different underwater environment conditions; binocular ranging is carried out on the underwater target image based on a binocular ranging algorithm, and the target position of the underwater target is obtained; visual display is carried out through a visual user interface.
Owner:SHANGHAI OCEAN UNIV

Protocol-Aware Provisioning of Resource over CXL Fabrics

Dynamic provisioning of resources in a datacenter improves the utilization efficiency of compute, memory, storage, and network resources, while maintaining flexibility to meet changing demands. Embodiments herein disclose protocol-aware provisioning of resources over Compute Express Link (CXL) fabrics, enabling multi-protocol pooling and disaggregation of resources, including memory and workload-specific accelerators such as GPUs and DSAs. In some embodiments, a Resource Provisioning Unit (RPU) facilitates intent-based protocol translations and mappings between address spaces, potentially enabling the creation of large-scale compute-memory fabrics that may utilize both coherent and non-coherent Non-Transparent Bridging (NTB) between multiple protocols in a single system, serving as the underlying infrastructure for executing workloads such as Large-Language Models (LLMs) and Deep Learning Recommendation Models (DLRMs). Some embodiments also optimize low-latency communication between processes running on different nodes by enabling host-to-host memory provisioning for libraries such as OpenMP, Pthreads, or CUDA, which can utilize shared memory.
Owner:HYATT GAYA OPAL MS +1

Data storage position dynamic optimization method and device, equipment and storage medium

The invention provides a data storage position dynamic optimization method, device and equipment and a storage medium, and the method comprises the steps: connecting expansion card equipment with a GPU through a PCIe interface on the GPU, and carrying out BDF structure scanning and space mapping to obtain storage space configuration information; preprocessing a shared keyword code segment of the GPU kernel function code and inserting an expansion card identifier, and configuring data transmission parameters to obtain initial configuration of a CUDA shared storage mechanism; the method comprises the following steps of: partitioning a running capacity region and a high-speed cache region of a storage space of expansion card equipment, and performing thread synchronization and parallel calculation on to-be-operated data of a GPU (Graphics Processing Unit) through SLC (Single Level Cell) data block partitioning to obtain running state information of the storage space; and dynamically migrating the distribution of the to-be-operated data between the operation capacity region and the cache region according to the operation state information. By dynamically optimizing the data storage position, the utilization rate of storage resources is improved, the data access delay is reduced, the GPU computing process is accelerated, and the overall system performance is improved.
Owner:SHENZHEN QUANXING TECH CO LTD

Multi-view collaborative 3D Gaussian splash optimization method and system

The invention discloses a multi-view collaborative 3D Gaussian splash optimization method and system, and the method comprises the steps: constructing a multi-level heterogeneous video memory pool, and dynamically dividing a video memory in 3D Gaussian reconstruction into a view exclusive memory block and a global shared memory pool; a mixed rendering-gradient pipeline is designed, and hardware-level pipeline parallelism in 3D Gaussian reconstruction is realized through a double-buffer asynchronous switching mechanism based on a CUDA Warp-level parallel primitive fusion forward rendering and back propagation thread group; performing multi-view gradient joint optimization, screening an effective gradient path in 3D Gaussian reconstruction based on the visibility mask matrix, and performing projection error weighted fusion on a multi-view gradient tensor; and implementing a multi-modal densification decision, generating a 3D Gaussian candidate splitting position in 3D Gaussian reconstruction through Monte Carlo sampling, calculating a joint optimization objective function by combining a multi-view projection residual error and a gradient contribution factor, and finally realizing 3D Gaussian reconstruction. According to the invention, high-precision and low-delay large-scale scene real-time rendering and training can be realized.
Owner:ZHEJIANG UNIV

Slope instability whole process calculation method, system and equipment based on particle method

The invention discloses a slope instability whole process calculation method, system and equipment based on a particle method. The method is based on a smoothed particle hydrodynamics (SPH) particle calculation method, combines a Riemannian large deformation theory, and realizes landslide simulation under a regional scale through a C + + / CUDA writing program. According to the method, the whole process of generation, connection, starting and accumulation of the sliding surface can be truly represented, the influence mechanism of rock and earth mass properties such as an internal friction angle, cohesion, particle size, elasticity modulus and Poisson's ratio on the landslide process can be analyzed, meanwhile, through a parallel heterogeneous CPU / GPU programming architecture, the calculation efficiency is improved by nearly a hundreds of times compared with the traditional CPU programming efficiency, and the calculation cost is greatly reduced. The watershed scale landslide flow velocity and accumulation range prediction is realized, the adaptability is very high, the defect that discrete element and finite element methods are limited by simulation scale and calculation efficiency is overcome, and a new thought is provided for large-scale landslide disaster prediction.
Owner:WUHAN UNIV

CUDA kernel generation method and device based on reinforcement learning

The invention provides a CUDA kernel generation method and device based on reinforcement learning, and a CUDA kernel generation method and device based on reinforcement learning, which utilize a CUDA kernel generation and optimization model of reinforcement learning training and adopt a multi-round reinforcement learning training scheme. According to the method, the continuous optimization and iteration process of code trigger generation, execution and feedback followed by an engineer during development of the CUDA kernel is simulated, and performance feedback in a real execution environment can be fused into a model training process, so that the model autonomously explores an optimization path under a reinforcement learning mechanism, and automatic and intelligent generation of the high-performance CUDA kernel is realized.
Owner:YUANQIXIN (SHANDONG) SEMICONDUCTOR TECHNOLOGY CO LTD

Space-space-based infrared imaging simulation platform based on CUDA (Compute Unified Device Architecture)

The invention relates to an air-space-based infrared imaging simulation platform based on CUDA (Compute Unified Device Architecture). The air-space-based infrared imaging simulation platform comprises a global calculation class, a three-dimensional position calculation module, a camera generation module, a sun calculation module, a background data extraction module, a target calculation module and an imaging effect calculation module. The platform meets the requirement of high-precision signal-level space-space-based infrared detection imaging simulation, the simulation image data generation efficiency is greatly improved under the condition that the signal-level imaging simulation precision in a space scene is guaranteed, and rapid generation of imaging simulation data of a space platform joint observation target in a multi-observation track scene is achieved. And the universality of the platform is improved by using modular data and development.
Owner:XIDIAN UNIV

Strong real-time signal processing method based on multi-stream processing

The invention discloses a strong real-time signal processing method based on multi-stream processing, and the method comprises the steps: carrying out the remainder of a preset stream number of received coherent processing interval data through a CPU according to a CPI number, obtaining a stream number, distributing the coherent processing interval data to each stream distributed calculation program corresponding to the stream number, each stream distributed computing program comprises a CUDA stream and a CPU processing thread which are executed in sequence; the GPU and the CPU execute each stream distributed calculation program to overlap the calculation time and the data copying time of the CUDA stream and the CPU processing thread to obtain each trace point condensation data with different completion time; the CPU obtains the corresponding report to be sent according to the trace point condensation data, and orderly sends the report to be sent to the target program based on a report waiting synchronization mechanism, and the application can complete front-end downloading data receiving, detection processing and report sending in strong real time, thereby improving the development efficiency, shortening the product development period, and reducing the project cost.
Owner:CNGC INST NO 206 OF CHINA ARMS IND GRP

Method and system for selecting optimal network topology structures of different types of large models

The invention discloses an optimal network topology structure selection method and system for different types of large models, and belongs to the technical field of large models. The method comprises the following steps: expanding a search object into a large model, selecting the large model, expanding an Embedding operator for each selected large model, and adding a data processing dimension; determining several kinds of candidate cluster topological structures, and performing training time simulation on each candidate cluster topological structure on a coarse granularity level; performing time measurement on the key operator by using CUDA (Compute Unified Device Architecture), wherein the time measurement is used as an input parameter in a simulation process; the simulation training time under different cluster topological structures is compared, and whether an obvious topology selection tendency exists or not is determined according to a simulation result; and selecting the cluster topological structure with the shortest training time as an optimal solution. The topological structure can provide the fastest training speed under a given large model and a data processing task; and the efficiency of the selection process is greatly improved.
Owner:SHENZHEN INST OF ADVANCED TECH

Homomorphic encryption acceleration method based on polynomial multiplication optimization and GPU multi-thread mapping

The invention provides a homomorphic encryption acceleration method based on polynomial multiplication optimization and GPU multi-thread mapping, and relates to the field of computer science and technology. The method comprises the following steps: on the basis of a CUDA-GPU structure model, according to parallel computing characteristics of a GPU, adopting a nested loop optimization method to perform loop optimization on an algorithm structure based on NTT transformation and an algorithm structure based on INTT transformation respectively to obtain a loop optimization algorithm based on NTT transformation and a loop optimization algorithm based on INTT transformation, so as to obtain a loop optimization algorithm based on NTT transformation and a loop optimization algorithm based on INTT transformation; through a BFV primitive root pre-calculation and pre-storage joint optimization algorithm, homomorphic multiplication in GPU parallel calculation environment homomorphic encryption operation is improved, and improved homomorphic multiplication is obtained; inputting two pieces of ciphertext data to be processed; and processing the two pieces of to-be-processed ciphertext data by adopting improved homomorphic multiplication to obtain a processed ciphertext result. By adopting the method, the polynomial multiplication execution efficiency of the GPU parallel computing environment can be improved.
Owner:UNIV OF SCI & TECH BEIJING +1

Image connected domain marking method, electronic equipment and storage medium

The invention discloses an image connected domain marking method, electronic equipment and a storage medium. The method comprises the following steps: performing parallel processing on pixel points in a to-be-processed image based on a compute unified device architecture CUDA; specifically, foreground pixel points and background pixel points in the image are determined; aiming at each row of pixel points, determining each group of continuous foreground pixel points in the row of pixel points, and respectively taking each group of continuous foreground pixel points as a sub-connected domain; for any first connected sub-domain and any second connected sub-domain in the image, if any first pixel point in the first connected sub-domain is longitudinally adjacent to any second pixel point in the second connected sub-domain, combining the first connected sub-domain and the second connected sub-domain to obtain each connected domain; and determining a target mark value of each connected domain. And the efficiency of marking the connected domain is improved.
Owner:ZHEJIANG DAHUA TECH CO LTD

Discrete Fourier detection implementation method and system based on CUDA (Compute Unified Device Architecture) tensor core

The invention belongs to the field of randomness detection, and relates to a discrete Fourier detection implementation method and system based on a CUDA tensor core. The method comprises the following steps: realizing discrete Fourier transform (DFT) with a cardinal number of 10 on a CUDA (Compute Unified Device Architecture) by utilizing the characteristics of a tensor core of an NVIDIA GPU (Graphics Processing Unit); kernel fusion is carried out by using discrete Fourier transform with a cardinal number of 10, a plurality of small kernels are fused into a large kernel, and large-size input data are decomposed and recombined, so that the execution of fast Fourier transform (FFT) operation is optimized. In addition, the execution process is optimized by using the CUDA flow, the advantages of the CUDA architecture can be fully played, and the data processing efficiency and the algorithm execution speed are improved. Compared with the prior art, higher performance can be provided when a large-scale data set is processed.
Owner:INST OF SOFTWARE - CHINESE ACAD OF SCI

GPU-CUDA-based SPH pollutant transport simulation parallel computing method

The invention provides an SPH pollutant transport simulation parallel computing method based on GPU-CUDA. The SPH pollutant transport simulation parallel computing method based on GPU-CUDA comprises the following steps that firstly, parallel computing parameters are initialized, and particle information is initialized; 2, copying necessary data from the CPU end to the GPU end; step 3, carrying out parallel computing of the SPH method, wherein the parallel computing specifically comprises the following sub-steps: dividing background grids; particle sorting: calculating a grid where the particles are located according to the position information of the particles, sorting all the particles according to spatial positions, and ensuring that adjacent particles are continuously stored in a memory; a multi-thread particle search strategy: particle attribute updating: updating the position and speed of the particle, and completing simulation of the current time step; processing boundary conditions; iteration is repeated until the simulation end time is reached; 4, after simulation is finished, result data are copied back to the CPU end from the GPU end.
Owner:TIANJIN UNIV

Radio astronomy real-time data stream processing method and system based on distributed acquisition component

The invention discloses a radio astronomical real-time data stream processing method and system based on a distributed acquisition component, and the method comprises the steps: S1, data transmission: a server receives a user datagram protocol UDP data packet transmitted by a radio astronomical telescope through a network card, and carries out the data transmission through an open-source pulsar distributed acquisition component and a data analysis system PSRDADA; creating and managing annular buffers, and transmitting data streams among the annular buffers; s2, data processing: using a CUDA (Unified Computing Device Architecture) parallel computing platform and a programming model of a GPU (Graphics Processing Unit) to write a computing and unpacking program so as to realize parallel processing and data extraction of large-scale data; and S3, writing data into a disk, writing the data blocks into the disk by using a data storage client of the PSRDADA, and dividing the data into files with any length. According to the invention, high-speed radio data streams can be received efficiently and stably, real-time processing of data is realized, and a reliable solution is provided for high-bandwidth signal transmission and processing in radio astronomical observation.
Owner:NAT ASTRONOMICAL OBSERVATORIES CHINESE ACAD OF SCI

Unmanned aerial vehicle aerial multi-image splicing algorithm based on opencv

The invention discloses an unmanned aerial vehicle aerial multi-image splicing algorithm based on opencv, and the algorithm comprises the following steps: S1, obtaining an aerial image data set, employing an OpenCV CUDA parallel processing technology, and carrying out the preprocessing of the image information of the aerial image data set on a GPU; s2, establishing a memory management mechanism, dynamically calculating a memory size request of the image information by combining and utilizing the parallel computing capability of CUDA in the preprocessing process, pre-defining memory blocks with different sizes in the memory management mechanism, and optimizing memory allocation; s3, grouping the preprocessed image information according to the feature information of the image information, and carrying out parallelization processing on image splicing tasks in each group by using CUDA (Compute Unified Device Architecture); by adopting the OpenCV CUDA parallel processing technology, the aerial image is preprocessed on the GPU, the image processing speed and efficiency are remarkably improved, the memory size request of the image information is dynamically calculated by utilizing the parallel computing capability of the CUDA, and the memory allocation is optimized, so that the overall performance is further improved.
Owner:GUANGXI NORMAL UNIV

Lightweight 3D medical image real-time reasoning method and system based on edge calculation

The invention relates to the technical field of image data processing, and particularly discloses a lightweight 3D medical image real-time reasoning method and system based on edge computing, and the method comprises the steps: building a C + + frame model, carrying out anisotropy detection and processing, separating low-resolution axis resampling, carrying out 3D resampling, determining the step length of a sliding window, and generating a weight map. And performing prediction, sliding window reasoning and image post-processing by using a lightweight student model, performing processing including size adjustment, voxel communication and integrity filling on a predicted segmentation map obtained through reasoning, and outputting a final global segmentation map. According to the method, the lightweight student model is adopted to replace the traditional nnUNet model reasoning which needs five times, the reasoning frequency is reduced by 80%, and the single reasoning time is remarkably shortened; meanwhile, CUDA parallelization is achieved through C + + language reconstruction, and through data block extraction, mirror image transformation and GPU acceleration of a weighted aggregation operator, the whole process performance is improved by more than 10 times compared with Python implementation.
Owner:XIAOZHI FUTURE (CHENGDU) TECH CO LTD

Risk content identification and abstract generation method of multi-modal large model

A method for risk content identification and abstract generation by using a multi-modal large model comprises the following steps: collecting risk content data and carrying out data preprocessing, and dividing based on the preprocessed risk content data to obtain a training set and a test set; performing model fine tuning on Chameleon-7B by using the training set to obtain a risk identification large model RiskControl-Valuer; inputting the test set data into a RiskContent-Valuer for abstract feature extraction to obtain a risk content identification and abstract generation result; wherein before fine tuning of the model, a BitsAndBytesConfig interface is utilized to quantify a basic model Chameleon-7B, the weight of the model is converted into int4 through a quantization layer, the int4 is put into a GPU, and a core calculation process is completed on a CUDA (Compute Unified Device Architecture); and performing model fine tuning by adopting supervised learning and LoRA methods. According to the method, the consumption of computing resources is greatly reduced by utilizing the LoRA, the powerful language understanding ability crossing different modalities and styles is inherited, accurate risk content assessment and total results are realized, and meanwhile, risk point abstracts are provided.
Owner:NORTHWESTERN POLYTECHNICAL UNIV

Intelligent computing center computing power asymmetric collaborative scheduling method oriented to common computing power

The invention provides an asymmetric collaborative scheduling method for computing power of an intelligent computing center oriented to common computing power, and relates to the technical field of intelligent computing centers, intelligent computing centers and computing power infrastructure, the method comprises the following steps: S1, obtaining operation parameters corresponding to each computing power operation task in a plurality of computing power operation tasks in the intelligent computing center, the operation parameters comprise a streaming processor (SM) utilization rate and other parameters, and the other parameters comprise at least one of a video memory utilization rate, a unified computing device architecture (CUDA) core activeness, a memory occupancy rate and a bandwidth occupancy rate; s2, dividing the plurality of computing power operation tasks based on the operation parameters to obtain at least one group; and S3, scheduling the computing power operation tasks included in different groups to different acceleration cards of the intelligent computing center, wherein each acceleration card of the intelligent computing center is used for executing the computing power operation tasks in one group. The resource utilization rate can be greatly improved, the computing power leasing cost is greatly reduced, and wide application of the general computing power is achieved.
Owner:DATACANVAS LTD