717 results about "Graphical processing unit" patented technology

An aspect relates to determining if a request received from a user device is sufficiently similar to a cached request, wherein the request requesting an identification of streamable content meeting one or more criteria. The determination comprises comparing a vector corresponding to the received request with vectors of previously received, cached requests. If the vector corresponding to the received request is sufficiently similar to a first vector of a first previously received, cached request, then a previously generated response corresponding to the first previously received, cached request may be accessed and transmitted to the user device. If a sufficiently similar vector is not identified, then at least a portion of the received request may be transmitted, in association with an identification of content items in a library, to an artificial intelligence learning engine. The response from the artificial intelligence learning engine may be transmitted to the user device.

An example operation may include at least one of receiving, via a user interface of a device, an activation input from a user to initiate a session, capturing, by a camera of the device, a scan of a body of the user, wherein the capturing comprises recording at least one image and / or at least one video of the user, processing the at least one image and / or video to generate a three- dimensional model of the user comprising measurements and contours of the body, retrieving, from a database, at least one clothing item associated with the user, the at least one clothing item comprising dimensional attributes and texture attributes, rendering, by a graphics processing unit, the at least one clothing item onto the three-dimensional model to generate a visual representation, wherein the rendering simulates draping behavior, movement, and light interaction of the at least one clothing item relative to the three-dimensional model, and displaying, on the user interface, an interactive visualization comprising the visual representation of the three-dimensional model with the at least one clothing item from multiple viewing angles.

The invention discloses an inference method and system, computer equipment and a storage medium. The method is applied to a target system comprising a central processing unit (CPU), a graphic processing unit (GPU) and a field programmable gate array (FPGA), and comprises the following steps: in response to a request of a target reasoning service, the CPU transmits a target input of the target reasoning service to the GPU; the GPU performs calculation of a pre-filling stage on target input based on a network structure and weight parameters of the large language model to obtain first data including an intermediate calculation result and a final calculation result of the pre-filling stage, and transmits the first data to the CPU; the CPU transmits the first data to the FPGA; the FPGA performs calculation in a decoding stage based on a network structure of a large language model, a weight parameter and the first data to obtain second data and transmits the second data to the CPU; and the CPU takes the second data as the output of the target reasoning service. According to the method, the target reasoning service can be deployed to the GPU and the FPGA in stages, the target reasoning service is completed in a cooperative operation mode, and hardware resources of the GPU and the FPGA are fully utilized.

Apparatuses, systems, and techniques to train a neural network on multiple graphics processing units (GPUs). In at least one embodiment, the neural network may be trained in parallel based, at least in part, on two or more randomly selected, similarly-sized portions of one or more datasets.

A system and method for real-time monitoring of personal protective equipment (PPE) compliance at a worksite. The system includes a processor, a host memory, a graphical processing unit (GPU), a GPU memory, a storage device, an input device, a plurality of cameras, and a display device. The system creates multiple inference engines and loads input frames from the storage device for pre-processing. The pre-processed frames are processed to obtain annotations and generate an output file. The system also includes a preparatory area and an operations area with cameras configured to capture and transmit video frames. The system can process input frames at a rate of at least 15 frames per second, providing real-time PPE compliance information.

An integrated system for detecting dam breaches, as an early warning system and for evacuation assistance, consisting of: A geoinformatics analysis module comprising a modeling processor with a graphics processing unit, configured to: analyze spatial, hydrological, and topographic data of a dam and its downstream area; generate hydrological models using probable maximum rainfall (PMP) data to determine probable maximum flood discharge (PMF); generate hydraulic models to simulate flood propagation and determine flood depth, flood extent, flow velocity, and flood arrival time; create flood vulnerability models, vulnerability models, and flood hotspot models using a multi-criteria decision-making (MCDM) framework; and generate evacuation plans for high-risk zones based on the models. A sensor monitoring subsystem includes: a distributed fiber optic sensor (DFOS) configured to monitor the structural load and deformation of a dam, and a radar level gauge configured to measure the water level in the reservoir; a microcontroller unit configured to receive and process sensor data from the sensor monitoring subsystem, to detect dam breaches based on abnormal structural movements detected by the DFOS or sudden drops in water level below predefined thresholds detected by the radar level gauge; a communication module configured to transmit warning messages when dam breaches are detected; a warning system configured to: send SMS notifications to residents in downstream high-risk zones via a GSM communication module, generate real-time alerts via an IoT platform on user devices, and activate siren systems in villages to issue loud alarms and voice announcements; and an evacuation assistance platform configured to provide real-time evacuation assistance in emergencies due to dam breaks, displaying site maps with nearby emergency shelters, evacuation routes, hospitals, government buildings and aid centers, showing hydrological layers including flood extent, water depth, flow velocity and arrival time of the water, and enabling real-time location tracking in relation to safety zones and flood-prone areas.

A method for training or performing inference using a neural network involves performing per-layer decompression and compression of neural network weights. More particularly, compressed weights are retrieved for a particular layer of the neural network. The weights correspond to neurons in the layer. The compressed weights are decompressed, and input data for that layer is subsequently processed using the decompressed weights. This dynamic decompression and recompression of weights allows memory, and in particular random access memory of graphical processing units, to be efficiently used.

Technology advancements in computer processors, including in particular graphics processing units (GPUs), have included the introduction of parallel processing capabilities which offer high compute throughput and memory bandwidth. Currently, programmers spend considerable time optimizing their code to best exploit the available parallel processor resources. However, some applications are still unable to consistently attain high compute throughput or memory bandwidth despite the presence of abundant parallelism. The present disclosure provides both hardware and software support for accelerating parallel processing pipelines.

Discussed herein is a routing mechanism for graphical processing units (GPUs) that are hosted on several host machines in a cloud environment. For a packet transmitted by a GPU of a host machine and received by a network device, the network device determines an incoming port-link of the network device on which the packet was received. The network device obtains a flow information associated with the packet, and computes, based on the flow information, an outgoing port-link of the network device in accordance with a hashing algorithm. The hashing algorithm is configured to hash packets received on a particular incoming port-link of the network device to be transmitted on a same outgoing port-link of the network device. The network device forwards the packet on the outgoing port-link of the network device.

A personalized diagnostic system for healthcare, comprehensive: a deep learning framework trained on a diverse dataset of medical images to analyze and interpret medical images; a data fusion module that integrates imaging data with non-imaging patient data such as electronic health records (EHRs), genetic information, and biomarker profiles; a high-performance computing system that uses GPUs (Graphics Processing Units) and TPUs (Tensor Processing Units) for real-time and batch processing of medical data; an intuitive user interface that displays diagnostic results, visualizations of medical images, and interactive functions for adjusting diagnostic parameters; a continuous learning mechanism that updates the deep learning models based on new data and expert feedback; a cloud-based platform for managing the continuous learning process, including the storage of anonymized patient data and diagnostic results; an explanatory module that provides detailed explanations of the diagnostic process using techniques such as Grad-CAM (Gradient-weighted Class Activation Mapping) and SHAP (SHapley Additive exPlanations); and a role-based access control (RBAC) mechanism to secure patient data.

A method for dynamically optimizing and modifying allocation of at least one disaggregated graphical processing unit (GPU) to at least one process via at least one virtual graphical processing units (vGPU) includes receiving, by a resource manager, from a runtime component executing on a first computing device, a request for allocation, to a process executing on the first computing device, of access to a GPU in the first computing device. The method includes allocating, by the resource manager, to the process, access to at least one vGPU associated with a second GPU on a second computing device. The method includes transmitting, by the resource manager, to the runtime component, an identification of the at least one vGPU identifying the at least one vGPU as located on the first computing device. The method includes instantiating, by the runtime component, the process with access to the at least one vGPU.

Systems, devices, and methods for disaggregating networking components are provided. An example networking chassis includes a first disaggregated server device supported by the networking chassis that includes a first central processing unit (CPU) and a first graphics processing unit (GPU) coupled with the first CPU. The networking chassis further includes a first insertable switch module communicably coupled with the first disaggregated server device that includes first switching chipsets and a first fabric management controller coupled with the first switching chipsets. The first insertable switch module at least partially controls data transmission associated with the first disaggregated server device. The first GPU of the first disaggregated server device is isolated on the first disaggregated server device, supported on the first disaggregated server device in the absence of other GPUs, or is otherwise the only GPU on the first disaggregated server device so as to provide modularity in networking applications.

Apparatuses, systems, and techniques to detect uplink data in a multi-user multiple input multiple output wireless communication system. In at least one embodiment, uplink data is detected using one or more graphics processing units to perform parallel computations in order to form a consensus belief about information received by one or more base stations in a wireless network.

The invention discloses a graphic processing unit cooperative processing method and system for accelerating a three-dimensional generative model. The model comprises a diffusion processing stage and a Gaussian splash rendering stage. In the diffusion processing stage, the time step number and the data precision calculated by the neural network are dynamically adjusted according to the visual angle sensitivity, and a specific graphic processing unit core is scheduled for execution. In the Gaussian splash rendering stage, pixel block tasks are predicted and intelligently scheduled to a stream processor based on historical data; and in the stream processor, the processing sequence of the Gaussian ball by the rendering unit is optimized according to the geometric distance, and the close-range unit is preferentially processed. During back propagation, the built-in logic of the stream processor preaggregates the gradient contributions of the same Gaussian ball, and then updates the gradient contributions through a single write operation. Through collaborative optimization of software and hardware, the efficiency bottleneck of the 3D generation model on a graphic processing unit is effectively solved, the calculation speed, the throughput and the hardware utilization rate are remarkably improved, and the model generation time is shortened.

The invention relates to the technical field of data storage, and discloses a CSD-based key value database storage method and system, and the method comprises the steps: responding to a user inserting a key value pair into a key value database, storing the key value pair into a memory table, and converting the memory table into an immutable memory table after a preset triggering condition is satisfied; aligning and coding the key value pairs in the immutable memory table according to a uniform length to generate key value pairs with fixed step lengths; calling a computing storage device to automatically compress the key value pair with the fixed step length and flash the key value pair into an SSTable file; receiving a merging operation triggering instruction, and controlling the computing storage device to obtain the SSTable file and decompress the SSTable file; and calling the graphic processing unit to merge the decompressed files to generate a new SSTable file, and calling the computing storage device to automatically compress the new file and flash the new file to a disk. According to the method, the read-write throughput and merging efficiency of the key value database are remarkably improved, and the problems of write blockage and write pause are effectively relieved.

The present invention is directed to an integrated computing architecture designed for distributing a layered data set according to different computation tasks and to different processing units based on the layer type descriptors to carry out different computation tasks. Computation tasks involving matrix operations are distributed to Graphic Processing Units and others involving computation tasks requiring application of specialized computation models and distributed to one or more specialized processing units. The specialized processing units can include data layers that cannot be efficiently handled by GPUs and / or data layers that require models that include non-classical models such as quantum models. The integrated computing architecture is particularly useful when at least one of the data layers requires processing by a quantum cognition model and it may be extended by models with synthetic agents such as Artificial Intelligence Agents (AI Agents).

The embodiment of the invention relates to a data processing method and device based on a key value database, equipment, a storage medium and a key value database system.The method comprises the steps that after it is determined that to-be-cached key value data exists in a video memory of a graphic processing unit, a data storage request for the to-be-cached key value data is sent to a key value database server side, the data storage request carries storage address information and storage length information of value data of to-be-cached key value data in a video memory; receiving a data reading request which is sent by a key value database server and carries storage address information, and controlling the first network card supporting the remote direct memory access protocol to read value data of the key value data to be cached from the video memory according to the storage address information, sending the value data to the key value database server side, and writing the value data through the second network card supporting the remote direct memory access protocol. Visibly, according to the embodiment of the invention, the key value data in the GPU video memory can be stored without a memory copying process, and the processing efficiency of the key value data is improved.

In general, techniques are described for providing a virtual graphics processing unit driver by receiving, in a user mode of a guest process executed by a computing system, one or more graphics commands from an application of the guest process; triggering, in the user mode of the guest process, an operating system call in response to the one or more graphics commands; translating, in a kernel mode of the guest process, the operating system call into a first virtual buffer; translating, in the user mode of a host process executed by the computing system, the first virtual buffer into the operating system call; and sending, in the user mode of the host process, the operating system call to a graphics processing unit driver in a kernel mode of the host process.

Apparatuses, systems, and frameworks for provisioning of efficient pipelines capable of multi-model inference and data processing using multiple processing units, including streaming data applications. The disclosed techniques include, during an initialization stage, assigning a plurality of machine learning models (MLMs) for execution on graphics processing units (GPUs), allocating memory space, on a hub GPU, to the plurality of MLMs, storing input data on the hub GPU before transferring the input data to other GPUs for execution. During an execution stage, output data is initially stored on GPUs that generated the output data before transferring the output data to the hub GPU.

The invention relates to the technical field of fuel cell monitoring and health management, and discloses an intelligent cooperative high-power proton exchange membrane fuel cell life prediction and fault diagnosis method and system, and the method comprises the following steps: S1, collecting multi-dimensional data of a stack key region in real time through a distributed sensing array; s2, removing noise and abnormal values of the collected data, and carrying out feature extraction; s3, constructing a dual-model collaborative prediction framework; s4, inputting the real-time features into the double models; s5, generating a maintenance strategy and early warning; according to the invention, special hardware accelerators such as a graphic processing unit or a field-programmable gate array are integrated on the edge computing module, so that the processing speed and efficiency of high-frequency and multi-source sensor data at a local end are remarkably improved; a distributed computing framework is adopted in a cloud analysis module to process mass feature data transmitted by an edge end, so that the capability and scale of the system for performing deep analysis on the health state of the fuel cell, accurately predicting the remaining service life and quickly tracing the fault are enhanced.

A technique manages computing resources on a container orchestration platform. Such a technique involves establishing a pool of computing resources on the container orchestration platform. Such a technique further involves, after the pool of computing resources is established, receiving graphics processing unit (GPU) provisioning requests (GPRs) which identify workspaces. Such a technique further involves allocating computing resources from the pool to the workspaces identified by the GPRs based on a set of GPR prioritization policies.

The invention provides a remote desktop resolution optimization method and system in a wayland environment, and the method comprises the steps: S1, obtaining a wayland protocol of screen display data, and adding a mode event in the wayland protocol, so as to achieve the communication between a remote desktop client and a local synthesizer; s2, the synthesizer calculates to obtain the optimal resolution, and sends the optimal resolution to a server side of the remote desktop through a mode event; and S3, after receiving the optimal resolution, the server issues the optimal resolution to the graphic processing unit and updates the modified display effect in the next frame of picture, and the step S2 is executed repeatedly until a stable state is reached. The method has the advantages of high flexibility, high efficiency and the like, and can allow the user to adjust the resolution of the remote desktop in real time under the condition of keeping the remote session uninterrupted.

The invention discloses a fault detection method and device, electronic equipment and a storage medium, and relates to the technical field of computers.The method comprises the steps that enhanced error report register data and machine inspection architecture register data from a server are received, the enhanced error report register data and the machine inspection architecture register data are sent when the server detects that the graphics processing unit has uncorrectable errors and the number of times of correctable errors reaches a preset threshold value; under the condition that the enhanced error report register data and the machine check architecture register data are valid, determining whether the graphics processing unit is in a healthy state according to the machine check architecture register data; if not, the fault type and the fault position of the graphic processing unit are determined according to at least one of the processing core information and the memory information included in the enhanced error report register data, so that specific internal equipment with faults of the graphic processing unit can be accurately positioned.

The invention discloses a distributed screen projection method and system based on GPU rendering, and the method comprises the steps: calling a window management interface and a graphic interface, and creating a virtual screen; setting one or more surface layers on the virtual screen, and drawing image data on the surface layers; and completing the synthesis of the image data on the surface layer by using a graphics processing unit (GPU), encoding the synthesized image data through a multimedia encoding and decoding service, and sending the encoded image data to a receiving end in a distributed manner for screen projection display. According to the method, the virtual screen is created by calling the window management interface and the graphic interface, image data synthesis is completed by utilizing the GPU, efficient coding is performed through the multimedia coding and decoding service, and finally the image data are sent to the receiving end in a distributed manner for screen projection display, so that low-delay and high-resolution compatible distributed screen projection is realized.

Various example embodiments of a processing unit fault detection capability are presented. The processing unit fault detection capability may be configured to support detection of faults in processor cores of a processing unit based on arrangement of the processor cores to form a data processing pipeline and monitoring of the processor cores of the data processing pipeline based on monitoring of the data processing pipeline (e.g., based on propagation of heartbeat messages via the data processing pipeline). The processing unit fault detection capability may be configured to support detection of faults in processor cores of various types of processing units, such as central processing units (CPUs), graphics processing units (GPUs), network processing units (NPUs), or the like.

A hardware apparatus for isolated virtual environments includes graphics processing unit comprising a first dedicated memory and a first plurality of processing cores, a central processing unit comprising a second dedicated memory and a second plurality of processing cores, a field programmable gate array comprising a third dedicated memory, a control and data bus assembly connecting the field programmable gate array, the central processing unit, and the graphics processing unit, and a hypervisor located on a non-volatile memory of the hardware apparatus, the hypervisor configured to create one or more virtual machines by isolating the graphics processing unit, the central processing unit and the field programmable gate array.

Systems and methods are disclosed for efficient memory allocation for processing large language model encoders and decoders based on an attention model. The system can utilize a plurality of two types of processors suitable for different types of LLM processing. These include Neural Processor Units and Graphic Processing Units. Each NPU processor has dedicated DDR memory coupled to each NPU. The DDR memory caches the neural network weights used in the generation of neural network activations. A plurality of GPUs provides KVQ token processing. LLM tokens processing can be performed in parallel batches or sub-batches to utilize idle NPU processors within the neural network. In some embodiments, the NPUs are structured in a matrix with a bus between adjacent processors. In another embodiment, NPUs provide both KVQ processing and neural network processing. The system can be integrated on a silicon substrate using chiplets in a 2.5 or 3-D architecture.

The invention discloses an underground space layout scheme VR visual interaction device and interaction method. The interaction device comprises a three-dimensional graphic processing unit, a virtual reality interaction terminal, a design software synchronization interface and a collaborative design communication module, wherein the virtual reality interaction terminal, the design software synchronization interface and the collaborative design communication module interact with the three-dimensional graphic processing unit. The three-dimensional graphic processing unit adopts a high-performance graphic processor and a special graphic acceleration chip, converts two-dimensional plane layout data into a three-dimensional space model and performs real-time rendering; the virtual reality interaction terminal serves as a direct interface for user experience and is integrated with the three-dimensional graphic processing unit through a low-delay wireless transmission module. The design software synchronization interface realizes data exchange with external design software and synchronizes modification to the three-dimensional graphic processing unit; and the collaborative design communication module is used for supporting real-time collaboration and communication of multiple users in the virtual environment. According to the method, high-performance hardware and special software are combined, so that rapid visualization and real-time interaction of underground space layout are realized.

The invention provides a multi-speech synthesis model bearing method and device based on a virtual GPU, and relates to the technical field of graphics processing units, and the method comprises the steps: carrying out the virtualization processing of a physical graphics processing unit, dividing the physical graphics processing unit into a plurality of virtual processing units with independent video memories and calculation quotas, and combining a resource scheduling mechanism, and deploying the speech synthesis language model instances in a plurality of service containers, and constructing a plurality of speech synthesis model bearing units. After a voice synthesis request is accessed, the scheduling module carries out load balancing according to the request connection number of each bearing unit, the request is distributed to a target bearing unit with the minimum connection number, and a voice generation task is completed by a virtual processing unit bound with the target bearing unit. According to the invention, resource division can be carried out on the physical graphic processing unit, and efficient operation of the multi-speech synthesis model is realized.

Techniques for executing multiple large language models (LLMs) on a single physical computing device, including receiving a task to be performed by at least a first LLM and a second LLM, creating a plurality of virtual compute nodes on the single physical computing device for processing at least some of first set of sub-tasks to be performed by the first LLM and second set of sub-tasks to be performed by the second LLM, wherein each one of the plurality of virtual compute nodes includes one or more graphics processing unit (GPU) cores allocated for performing a sub-task of the first set of sub-tasks or the second set of sub-tasks, and executing at least some of the first set of sub-tasks and the second set of sub-tasks substantially in parallel across the plurality of virtual compute nodes using shared memory. Inferences between the first and second LLMs may be shared.