Abstract: A method of performing an operation in a neural network apparatus may include acquiring operation groups each comprising at least one input feature map and at least one kernel, and tag information corresponding to each of the operation groups, determining an operating unit in an idle state from among operating units, performing, at the operating unit in the idle state, a convolution operation between an input feature map and a kernel included in a operation group from among the operation groups to create an intermediate feature map, determining, based on tag information corresponding to the operation group, a post-processing unit from among post-processing units, and creating, at the post-processing unit, an output feature map using the intermediate feature map.

Abstract: Systems and methods are provided for offloading a task from a central processor in a radio access network (RAN) server to one or more heterogeneous accelerators. For example, a task associated with one or more operational partitions (or a service application) associated with processing data traffic in the RAN is dynamically allocated for offloading from the central processor based on workload status information. One or more accelerators are dynamically allocated for executing the task, where the accelerators may be heterogeneous and my not comprise pre-programming for executing the task. The disclosed technology further enables generating specific application programs for execution on the respective heterogeneous accelerators based on a single set of program instructions.

Abstract: A request to execute a recurring job is received by a cloud computing environment. Thereafter, available computing resources available to execute the job are determined based on historical resource consumption trends. A resource prediction for the job is then generated using an ensemble model ((S)ARIMA model) that combines an autoregressive moving average (ARMA) model and an autoregressive moving average (ARIMA) prediction models. The resource prediction characterizes resources to be consumed for successfully executing the job. Execution of the job can then be scheduled by the cloud computing environment based on the resource prediction and the available computing resources. Related apparatus, systems, techniques and articles are also described.

Abstract: A method, system, and article of manufacture provide for multi-user collaboration on a three-dimensional (3D) design. The 3D design is acquired in a computer-aided design (CAD) application. A commenting process for a comment to be associated with a selected part of the 3D design is activated. Textual user input for the comment is dynamically processed as the comment is received. The processing recognizes that the text relates to creating or modifying the selected part, retrieves a list of alternative parts (based on similarities between the alternative parts and the selected part), and displays a graphic representation of an alternative part. An alternative part is selected and inserted in the comment as a proposed replacement part. The comment including the proposed replacement part is provided to another user.

Abstract: Implementations of the present specification provide an inter-thread interrupt signal sending method and apparatus. In the inter-thread interrupt signal sending method, a processor in which a first thread is located sends a notification message to a PCI device via a PCI bus by using an MMIO write operation. The MMIO write operation is implemented based on a virtual space address of the first thread to which a memory address of an MMIO memory of the PCI device is mapped. The PCI device generates an interrupt signal for a second thread in response to receiving the notification message, and sends the interrupt signal to a processor in which the second thread is located based on an interrupt signal sending manner configured in interrupt configuration information of the PCI device.

Abstract: In a case where an active system virtual server detects stopping of a heartbeat received from a standby system virtual server, the active system virtual server notifies a coordination apparatus of the detection of the stopping of a heartbeat. In addition, the active system virtual server reports whether or not a service is operating to the coordination apparatus. The coordination apparatus instructs the standby system virtual server to restart the system in a case where the stopping of a heartbeat and the operation of a service are received. The standby system virtual server restarts the service, restarts an operating system, or restarts the operating system in a case where the restart of the service is not successful, in response to an instruction for restarting the system.

Abstract: A system according to an embodiment of the present invention includes at least two virtual machines running on a hardware platform using either a hosted or a bare metal hypervisor. The virtual machines may communicate with an agent-server resident in the host operating system or in one of the virtual machines to switch control of the hardware component, such as graphics hardware, from one virtual machine to another.

Abstract: A method for quantifying resource usage may include measuring a quantity of a plurality of physical computing resources being used over a period of time. The method may also include normalizing each measured quantity of each physical computing resource being used. The method may also include summing the normalized measured quantities of the physical computing resources being used to generate a single usage value representative of the physical computing resources being used over the period of time.

Abstract: An example method of secure attestation of a workload deployed in a virtualized computing system is described. The virtualized computing system includes a host cluster and a virtualization management server, the host cluster having hosts and a virtualization layer executing on hardware platforms of the hosts. The method includes storing, in a trust authority, a pre-defined attestation report for a workload executing in a virtual machine (VM) managed by the virtualization layer, the pre-defined attestation report including a hash of at least a portion of an image of the VM; receiving, at the trust authority from a security module of a host in which the VM executes, an attestation report generated by measuring memory of the VM; comparing the attestation report with the pre-defined attestation report; and generating an indication of validity for the workload based on a result of the comparison.

Abstract: Methods, apparatus, systems and articles of manufacture for load shedding are disclosed. An example apparatus includes memory, and at least one processor to execute machine readable instructions to cause the at least one processor to at least access a request to execute a process using a second computing environment, the second computing environment being a replica of a first computing environment, determine an expected execution latency associated with execution of the process using the second computing environment, and reject the request in response to the expected execution latency exceeding a threshold.

Abstract: Some embodiments provide a method for operating a physical server in a network. The method stores multiple copies of a virtual machine (VM) image at a network-accessible storage. The method uses a first copy of the VM image as a virtual disk to execute a VM on a hypervisor of a first physical computing device. The method uses a second copy of the VM image as a virtual disk accessible via a smart network interface controller (NIC) of a second physical computing device to execute an operating system of the second physical computing device.

Abstract: A method includes: accessing a set of hardware parameters characterizing an embedded device; identifying a set of supported container functions based on the set of hardware parameters; accessing a selection of container functions; identifying a set of selected container functions based on the selection of container functions and the set of supported container functions; generating a hardware abstraction layer (HAL) including a set of libraries supporting the set of selected container functions; generating a container runtime environment (CRE) configured to execute, at the embedded device, a containerized application via the HAL, the containerized application including the set of selected container functions; installing the HAL and the CRE onto the embedded device; installing the containerized application onto the embedded device via the CRE; and at the embedded device, executing the containerized application via the CRE and the HAL.

Abstract: Techniques for reducing overhead in native function calls are disclosed. The system may receive a method invocation instruction for invoking a particular method. The method invocation instruction includes a function descriptor, a method type, and an application binary interface (ABI) descriptor. The function descriptor includes a memory layout corresponding to any data returned by the function and memory layouts corresponding to each argument for the particular method. The system can select an ABI for processing the particular method based on the received ABI descriptor. The system can further associate each argument with a corresponding particular physical register into which the argument is to be loaded. The particular register is selected based on at least the selected ABI and the function descriptor. The system can cause a virtual machine to move each argument into the corresponding associated physical register.

Abstract: An example cloud computing burst management system includes a first cloud computing resource including a first processor and a first memory, a second cloud computing resource including a second processor and a second memory, and one or more data networks connecting the first cloud computing resource and the second cloud computing resource. The first cloud computing resource is configured to perform at least one cloud computing task, and to monitor one or more leading indicator parameters associated with operation of the first cloud computing resource while performing the at least one cloud computing task. In response to the one or more leading indicator parameters satisfying a first burst criteria, the first cloud computing resource is configured to provision a task instance on the second cloud computing resource for performing at least one portion of the cloud computing task.

Abstract: Edge device task management is provided. It is determined whether a subtask cancel and migrate plan exists when a request to run a higher priority subtask of a second plurality of subtasks corresponding to a second task is received while a first task comprised of a first plurality of subtasks is running on a first cluster of edge devices. In response to determining that the subtask cancel and migrate plan does exist, a lower priority subtask of the first plurality of subtasks is canceled from a designated edge device of the first cluster of edge devices based on the subtask cancel and migrate plan. The lower priority subtask is migrated to another edge device for running based on the subtask cancel and migrate plan. The higher priority subtask of the second plurality of subtasks is sent to the designated edge device of the first cluster of edge devices for running.

Abstract: A device for managing composition of composed information handling includes persistent storage and a computer processor that obtains a composition request for a composed information handling system; makes a determination that the composition request is an intent based request; in response to the determination: matches an application specified by the composition request to computing resource requirements for the application; identifies, based on the computing resource requirements, at least one compute resource set having resources meeting the computing resources requirement; identifies, based on the computing resource requirements, at least one hardware resource set having resources meeting the computing resource requirements; sets up management services for the at least one hardware resource set using at least one control resource set to obtain logical hardware resources; and presents the logical hardware resources to the at least one compute resource set as bare metal resources.

Abstract: Apparatuses, methods and storage medium associated with embedded computing, are disclosed herein. In embodiments, an embedded computing platform includes a plurality of system-on-chips (SoCs) forming a local compute cluster; and an orchestrator disposed on one of the SoCs arranged to orchestrate fail-safe operations, in response to a reported unrecoverable failure requiring shut down or partial disabling of one of the SoCs, to consolidate execution of critical workloads on one or more of remaining fully or partially operational ones of the SoCs. Other embodiments are also described and claimed.

Abstract: A method using a memory and queue handling logic, including accessing a work control structure (WCS) configured “first-in-first-out” holding work control records (WCRs) each including a field defining work to be carried out and a completion indicator indicating whether the work has completed, and initially set to indicate that the work has not completed: upon fetching a work request (WR) for execution, pushing a WCR corresponding to the WR to the WCS, and: A) inspecting the WCR at a head of the WCS, B) when the completion indicator of the WCR at the head of the WCS indicates that the unit of work associated with the WCR at the head of the WCS has been completed, popping the WCR at the head of the WCS from the WCS, and reporting completion of the WCR at the head of the WCS to a host processor, and C) iteratively performing A, B, and C.

Abstract: Described herein are systems and methods that manage configuration updates for networking manager virtual machines. In one example, a method includes identifying an update for at least one networking manager virtual machine. In response to identifying the update, the method notifies a daemon on the host with the networking manager virtual machine to establish a channel with a control plane agent to receive communications in place of the networking manager virtual machine. The method further identifies when the configuration modification is complete for the networking manager virtual machine and notifies the daemon on the host to break the channel with the control plane agent.

Abstract: Some embodiments provide a method of providing distributed storage services to a host computer from a network interface card (NIC) of the host computer. At the NIC, the method accesses a set of one or more external storages operating outside of the host computer through a shared port of the NIC that is not only used to access the set of external storages but also for forwarding packets not related to an external storage. In some embodiments, the method accesses the external storage set by using a network fabric storage driver that employs a network fabric storage protocol to access the external storage set. The method presents the external storage as a local storage of the host computer to a set of programs executing on the host computer. In some embodiments, the method presents the local storage by using a storage emulation layer on the NIC to create a local storage construct that presents the set of external storages as a local storage of the host computer.