Abstract: A system-on-chip can include a data input chiplet to obtain data from one or more data sources. The system-on-chip can further include one or more workload processing chiplets that access the data obtained by the data input chiplet to execute respective workloads. The system-on-chip further includes a central chiplet including a shared memory comprising a reservation table listing a plurality of workload entries. Each respective workload entry can correspond to a specified workload to be executed by the one or more workload processing chiplets. The central chiplet can input a thread number for each respective workload entry in the reservation table, where the thread number identifies a workload pipeline in which the specified workload is to be executed.

Abstract: Disclosed are a method and an electronic apparatus including an accelerator for lightweight and parallel accelerator task scheduling. The method includes pre-running a deep learning model with sample input data having a preset data form and generating a scheduling result through the pre-running.

Abstract: A pipeline task verification method and system is disclosed, and may use one or more processors. The method may comprise providing a data processing pipeline specification, wherein the data processing pipeline specification defines a plurality of data elements of a data processing pipeline. The method may further comprise identifying from the data processing pipeline specification one or more tasks defining a relationship between a first data element and a second data element. The method may further comprise receiving for a given task one or more data processing elements intended to receive the first data element and to produce the second data element. The method may further comprise verifying that the received one or more data processing elements receive the first data element and produce the second data element according to the defined relationship.

Abstract: A method for alleviating data poisoning in an edge computing resource includes receiving a numeric value from an Internet of Things (IoT) unit and associating the numeric value with a cluster selected from a plurality of clusters in accordance with a suitable clustering algorithm such as a k-means clustering algorithm. In at least some embodiments, the numeric value comprises a poisoned numeric value including an adversarial component injected by an adversary to negatively impact a trained model of a cloud-based artificial intelligence engine. Rather than permitting the injected adversarial component to corrupt the AI engine, a cluster with which the numeric value is associated is sampled in accordance with a probability distribution of the cluster to obtain a surrogate for the poisoned numeric value. The surrogate may then be provided as an input to an inference module of the AI engine to generate a prediction.

Abstract: The system of the present technology includes an embodiment that provides a host audio, video and control operating system configured to establish or interact with one or more virtual machines, each with a guest operating system.

Abstract: Systems, apparatuses and methods include technology that analyzes an input stream and an artificial intelligence (AI) model graph to generate a workload characterization. The workload characterization characterizes one or more of compute resources or memory resources, and the one or more of the compute resources or the memory resources is associated with execution of the AI model graph based on the input stream. The technology partitions the AI model graph into subgraphs based on the workload characterization. The technology selects a plurality of hardware devices to execute the subgraphs.

Abstract: Techniques are disclosed for deploying a computing resource (e.g., a service) in response to user input. A computer-implemented method can include operations of identifying a first set of computing components already deployed within the cloud-computing environment and identifying a second set of computing components available for deployment within the cloud-computing environment. A request for deployment may be subsequently received for one of the available computing components. A bootstrap request corresponding to the particular computing component requested may be transmitted to a deployment orchestrator, the deployment orchestrator being configured to deploy the particular computing component to the cloud-computing environment based at least in part on the bootstrap request. A user interface may present status indicators for each computing component (e.g., deployed, available, requested, etc.).

Abstract: A method includes, by a scheduling controller, receiving from a user a request for an application to be executed by a computing system associated with a data center, wherein the application includes a plurality of tasks, and wherein the request includes an estimated execution time corresponding to an estimated amount of real-world time that the tasks will be actively running on the computing system to fully execute the application. The method includes receiving from the user a service level objective indicating a target percentage of a total amount of real-world time that the tasks will be actively running on the computing system and generating, in response to determining that the job can be completed according to the service level objective and the estimated execution time, a notification indicating acceptance of the job.

Abstract: Techniques for managing critical workloads in container-based computing environments are disclosed. In one example, a method determines a resource trigger threshold associated with executing at least one containerized workload associated with a first service having a first criticality level, the resource trigger threshold corresponding to a resource capacity allocated to execute the first service. The method determines when the resource capacity allocated to execute the first service reaches the resource trigger threshold, and then re-allocates resource capacity allocated to execute at least one containerized workload associated with a second service having a second criticality level to the first service when the resource trigger threshold is reached. For example, the first criticality level may be higher than the second criticality level.

Abstract: Example implementations described herein include an innovative method for copying a first virtual machine accessing a primary volume in a hybrid cloud environment, the copy of the first virtual machine accessing a secondary volume in the hybrid cloud environment. The method may include creating the secondary volume. The method may further include creating an image of the first virtual machine and launching, from the image, the copy of the first virtual machine by (1) disabling a set of connection ports for connecting to the primary volume, (2) updating the set of connection ports for connecting to the secondary volume, and (3) enabling the set of connection ports.

Abstract: Techniques for capacity management in computing systems are disclosed herein. In one embodiment, a method includes analyzing data representing a number of enabled users or a number of provisioned users to determine whether the analyzed data represents an anomaly based on historical data. The method can also include upon determining that the data represents an anomaly, determining a conversion rate between a change in the number of enabled users or the number of provisioned users and a change in a number of active users of the computing service and deriving a future value of the number of active users of the computing service based on both the detected anomaly and the determined conversion rate. The method can further include allocating and provisioning an amount of the computing resource in the distributed computing system in accordance with the determined future value of the active users of the computing resource.

Abstract: In a synchronization method, a first processor creates a first synchronization object for a first synchronization event. The first synchronization object includes an identifier of a first synchronization register. A value of the first synchronization register includes a first value or a second value. The first value is used to indicate that the first synchronization event does not occur, and the second value is used to indicate that the first synchronization event occurs. The first processor includes a first CPU. The second processor determines, based on the value of the first synchronization register, whether the first synchronization event occurs. The second processor includes a first NPU.

Abstract: In one example, a system for executing applications can include a main processor to initialize a virtual machine to execute an application. The main processor can also determine a main utilization indicator of the main processor is above a threshold and an auxiliary utilization indicator of an auxiliary processor is below a threshold, wherein the auxiliary processor is based on an auxiliary instruction set architecture. Additionally, the main processor can transmit an instruction from the application to the auxiliary processor for execution and update context data for the application in response to receiving an execution result from the auxiliary processor.

Abstract: An Internet of Things system comprises an IoT hub and a local subsystem with a plurality of subsystem devices. These subsystem devices include an edge hub communicatively coupled to the IoT hub and to each other subsystem device; a requestor module configured to perform a task according to a requestor module schedule; and a scheduler module with a persistent time loop. The scheduler module receives a scheduler request from the requestor module via the edge hub, and based on this scheduler request generates a subsystem schedule that includes the requestor module schedule. The scheduler module transmits at least a part of this subsystem schedule to a persistence layer outside of the local IoT subsystem, via the IoT hub. The scheduler module flags scheduled event occurrences via the time loop and the subsystem schedule, and transmits task-specific triggered messages to the requestor module in response to these event occurrences.

Abstract: A method, an apparatus and a medium for optimizing allocation of switching resources in the polymorphic network. The method selects the ASIC switching chip, FPGA and PPK software switching on the polymorphic network element based on machine learning, and specifically comprises the following steps: manually pre-configuring, formulating basic rules for polymorphic software and hardware co-processing; offline learning, designing training configuration in the offline learning stage to capture different switching resource usage variables, running experiments to generate the original data of a training classifier, and using the generated performance indices to train the model offline; and online reasoning, obtaining switching resource allocation advises, and updating modality codes according to the switching resource allocation advises.

Abstract: Systems and methods for managing resource availability are disclosed. In one embodiment, a method for managing resource availability may include: (1) receiving, by a deployment computer program, an identification of a rack to build; (2) retrieving, by the deployment computer program, a plurality of application placement profiles for a plurality of applications; (3) selecting, by the deployment computer program, a subset of the applications to deploy to the rack, wherein the deployment computer program optimizes the selection based on metrics in the application placement profiles and a capacity of the rack; and (4) deploying, by the deployment computer program, the subset of applications to the rack.

Abstract: Techniques for in-band modification of event notification preferences for server events are provided. One method comprises obtaining an event notification; providing the event notification to a target device based on rule-based preferences of a user associated with the target device; obtaining a reply to the event notification from the target device, wherein the reply comprises event preferences of the user; and updating the rule-based preferences of the user based on the event preferences of the user. The updating of the rule-based preferences of the user may comprise creating, modifying and/or canceling at least one event preference rule of the user. A plurality of the event preference rules matching the event notification may be resolved in an order determined by one or more event preference rule resolution criteria.

Abstract: A system for coordinating data offloading includes a controller. The controller is configured to at least receive a set of data metrics corresponding to a set of vehicles, generate a set of vehicle clusters based on the set of data metrics, determine a thread distribution based on the set of vehicle clusters, and transmit a data offloading control signal to each of the set of vehicles according to the thread distribution.

Abstract: A method for graphics processing, wherein a graphics processing unit (GPU) resource is allocated among applications, such that each application is allocated a set of time slices. Commands of draw calls are loaded to rendering command buffers in order to render an image frame for a first application. The commands are processed by the GPU resource within a first time slice allocated to the first application. The method including determining at least one command has not been executed at an end of the first time slice. The method including halting execution of commands, wherein remaining one or more commands are not processed in the first time slice. A GPU configuration is preserved for the commands after processing a last executed command, the GPU configuration used when processing in a second time slice the remaining commands.

Abstract: Concepts and technologies are disclosed herein for service creation and management. A processor can detect an event relating to a service, and access a service creation database to identify a recipe associated with the service. The recipe can define a resource to be used to provide the service. The processor can access an inventory to determine if the resource is available, identify a service control to control the service, instruct an infrastructure control to allocate virtual machines to host components of the service, and issue instructions to the service control. The instructions can instruct the service control to load service functions to the virtual machines.