Abstract: A method for instantiating and transparently migrating executing containerized processes includes receiving, by a container engine executing on a first machine, an instruction to instantiate a container image on the first machine. The container engine transmits, to a modified container runtime process, executing on the first machine, the instruction to instantiate the container image on the first machine. The modified container runtime process generates, on the first machine, a shim process representing the instantiated container image. The shim process forwards the instruction to an agent executing on a second machine, via a proxy connected to the agent via a network connection. The agent directs instantiation of the container image as a containerized process. A scheduler component executing on the first machine determines to migrate the containerized process to a third machine.

Abstract: Embodiments disclosed include data center infrastructure management (DCIM) systems and methods configured to collect data center compute systems, power systems, and facility systems data, trigger an action or actions based on a diagnosed or predicted condition according to the collected data, and thereby control via a compute, power, and facilities module, the compute systems, power systems and facility systems in the data center. According to an embodiment, the control via the compute, power, and facilities module comprises calibrating the compute, power, and facility systems based on an estimated compute requirement, and an associated power, cooling, and network data resource requirement. The estimated compute requirement comprises estimating compute density per real-time power wattage, and storage density per real-time power wattage.

Abstract: Techniques for implementing secure GPU virtualization using sandboxing are provided. In one set of embodiments, a hypervisor of a host system can receive one or more first graphics/compute commands issued by a guest application running within a VM of the host system. The hypervisor can further communicate the one or more first graphics/compute commands to a sandboxed software process that is separate from the hypervisor. The sandboxed software process can then translate the one or more first graphics/compute commands into one or more second graphics/compute commands and issue the one or more second graphics/compute commands for execution on a physical GPU.

Abstract: A job scheduler system includes one or more hardware processors, a memory including a job group queue stored in the memory, and a job scheduler engine configured to create a first job group in the job group queue, the first job group includes a generation counter having an initial value, receive a first request to steal the first job group, determine a state of the first job group based at least in part on the generation counter, the state indicating that the first job group is available to steal, based on the determining the state of the first job group, atomically increment the generation counter, thereby making the first job group unavailable for stealing, and alter an execution order of the first job group ahead of at least one other job group in the job group queue.

Abstract: An apparatus comprising an interface to receive an identification of a function to be executed; and a scheduling engine comprising circuitry, the scheduling engine to select a candidate compute element from a plurality of candidate compute elements based on a combined burden, the combined burden based on an estimated burden to execute the function by the candidate compute element and an estimated burden of data movement over at least one interconnect identified for the candidate compute element.

Abstract: A method for translating memory addresses in a manycore system is provided, which is executed by one or more processors, and includes receiving identification information of a thread accessing a memory associated with one or more cores of a cluster that includes a plurality of cores, receiving a virtual address of data accessed by the thread, and determining a physical address of data in the memory based on the virtual address and the identification information of the thread.

Abstract: A computer-implemented method at a data management system comprises: generating, with one or more processors, a containerized runtime in a memory in communication with the one or more processors; instantiating, with the one or more processors, an app in the runtime; receiving, with the one or more processors, a request from the app for data; retrieving, with the one or more processors, a copy of the requested data from a data source; and transmitting, with the one or more processors, the data to the containerized runtime for the app to operate on.

Abstract: A method is provided for shaping a custom dental restoration from a preform, wherein the preform comprises a preform body and a preform stem. A method is further disclosed for generating one or more nesting positions for the restoration design within the geometry of the preform body relative to the position of the preform stem. A method is further disclosed for generating machining instructions based on the selected nesting position to optimize machining for chair-side applications.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for placement and scheduling of radio signal processing dataflow operations. An example method provides a primitive radio signal processing computational dataflow graph that comprises nodes representing operations and directed edges representing data flow. The nodes and directed edges of the primitive radio signal processing computational dataflow graph are partitioned to produce a set of software kernels that, when executed on processing units of a target hardware platform, achieve a specific optimization objective. Runtime resource scheduling, including data placement for individual software kernels in the set of software kernels to efficiently execute operations on the processing units of the target hardware platform. The resources of the processing units in the target hardware platform are then allocated according to the defined runtime resource scheduling.

Abstract: An automatic driving simulation task scheduling method and apparatus, a computer device, and a readable storage medium. The method includes: constructing a network flow block diagram on the basis of the number of cluster racks, the number of nodes, and the number of processes, and coding information of the network flow block diagram (S01); setting information of a task to be simulated, and performing shortest path computation on a simulation task flow by means of the network flow block diagram (S02); and updating the state of a scheduling system in real time according to a fair policy and a preemptive scheduling policy, and executing scheduling on the basis of the state of the scheduling system (S03).

Abstract: A neural processing device and a method for job scheduling are provided. The neural processing device configured to receive, by an address space ID (ASID) manager, first and second requests from at least one context, respectively, and determine whether ASIDs are allocated, store jobs of contexts to which the ASIDs have not been allocated from the ASID manager in entities, schedule, by a job scheduler, an execution order of the jobs stored in the entities and cause the ASID manager to allocate the ASIDs to the contexts to which the ASIDs have not been allocated among the at least one context, and sequentially receive, by a command queue, jobs of contexts to which the ASIDs have been allocated, store the jobs as standby jobs, and sequentially execute the standby jobs.

Abstract: A hardware functional module in a System On Chip (SOC) performs a given task, where the given task is one of a plurality of tasks that are associated with a work item and the work item is managed by firmware in the SOC. A first notification that the given task has completed is sent from the hardware functional module to a scoreboard module, where the scoreboard module is implemented in hardware on the SOC. The scoreboard module: in response to receiving the first notification, increments a counter; determines whether the counter exceeds a threshold; and in the event it is determined that the counter exceeds the threshold, sends, from the scoreboard module to the firmware, a second notification indicating that the plurality of tasks have completed.

Abstract: A data transmission method and device for data transmission between an internal memory of a system-on-chip and an external memory coupled to the system-on-chip. The method is executed by a processor of the system-on-chip, including steps of: adding execution information of at least one data transmission task to be executed to a task queue; sending execution information of one data transmission task in the task queue to a direct access device for the direct access device to execute the data transmission task; and receiving an interrupt request sent by the direct access device, where the interrupt request is used to indicate completion of the execution of the data transmission task, so that the processor sends execution information of a next data transmission task to be executed in the task queue to the direct access device.

Abstract: The present disclosure is related to methods, systems, and machine-readable media for cloneless snapshot reversion. A request can be received to revert to a past snapshot of a virtual computing instance in a snapshot chain of a snapshot tree provided by a software defined data center. A live snapshot can be created at an end of the snapshot chain comprising the past snapshot. An intervening snapshot in the snapshot chain can be indicated as abandoned in a snapshot map associated with the snapshot tree based on the reversion.

Abstract: The present application discloses a task allocation method and system based on a resource management platform. The method comprises: receiving an artificial intelligence model training and/or testing task and a name of data set required for processing the task; acquiring data set distribution information of a plurality of nodes; judging if the node has the required data sets according to names of the data sets in the node; and selecting a node with the size of the required data set meeting preset requirements for task allocation according to the size of the required data set in the node if the node has the required data set.

Abstract: Contents of a virtual machine of a primary storage are received at a first point in time. A full snapshot of the virtual machine is stored on a secondary storage using the received contents of the virtual machine. One or more cloud storage volumes of the virtual machine are caused to be created. Contents of the full snapshot for storage on the one or more cloud storage volumes are provided from the secondary storage. Changed data of the virtual machine of the primary storage is received at a second point in time. An incremental snapshot of the virtual machine is stored using the received changed data. The one or more cloud storage volumes are updated using the incremental snapshot of the virtual machine. An image of at least one of the cloud storage volumes is caused to be created for standby deployment of another instance of the virtual machine.

Abstract: Aspects of the present disclosure enable data protection operations including differential and incremental backups by performing changed-block tracking in network or cloud computing systems with architectures that do not natively support changed-block tracking or do not expose changed-block tracking functionality to an information management system. In certain aspects, an identity of changed blocks may be obtained by using a hypervisor configured to interface with the cloud computing architecture. The identified changed blocks may be used to generate a map of the changed blocks. The maps of the changed blocks can be used by a virtual server agent to extract the changed blocks from a copy of a virtual machine disk and backed up to perform a differential or incremental backup.

Abstract: An interposer circuit is used between an interrupt controller and a processor core to facilitate direct injection of a virtual interrupt into a guest executing on the processor core, even though the interrupt controller does not support the direct injection. The interposer circuit can convert a command received from the interrupt controller for a physical interrupt into another command for a virtual interrupt to make the processor core believe that the processor core has received a virtual interrupt even though the interrupt controller is not able to provide the virtual interrupt. The virtual interrupt can be directly injected into the processor core without the intervention of a hypervisor executing on the processor core.

Abstract: Techniques for executing workgroups are provided. The techniques include executing, for a first workgroup of a first kernel dispatch, a workgroup dependency instruction that includes an indication to prioritize execution of a second workgroup of a second kernel dispatch, and in response to the workgroup dependency instruction, dispatching the second workgroup of the second kernel dispatch prior to dispatching a third workgroup of the second kernel dispatch, wherein no workgroup dependency instruction including an indication to prioritize execution of the third workgroup has been executed.

Abstract: A method for executing instructions embedded in two threads stored in a system including two operating units and a virtual managing entity for holding queues for virtual objects (VO) waiting to use a respective operating unit and diverting them between queues. Each VO is associated with two virtual timers, one measuring a time period during which the VO is held in the queue (TIQ) and the other providing time period during which the VO will remain in an alive state (TTL). The method includes receiving information relating to VOs associated with the two threads; operating on VOs for which their TTLs have expired; upon emerging from its respective queue, determining whether each VO should be diverted to another queue; upon diverting the VO, resetting its TIQ timer; and allocating an access time to each VO based on a number of threads requiring that VO and the TIQ associated therewith.