Modern parallel architectures have both heterogeneous processors and deep, complex memory hierarchies. We present Legion, a programming model and runtime system for achieving high performance on these machines. Legion is organized around logical regions, which express both locality and independence of program data, and tasks, functions that perform computations on regions. We describe a runtime system that dynamically extracts parallelism from Legion programs, using a distributed, parallel scheduling algorithm that identifies both independent tasks and nested parallelism. Legion also enables explicit, programmer controlled movement of data through the memory hierarchy and placement of tasks based on locality information via a novel mapping interface. We evaluate our Legion implementation on three applications: fluid-flow on a regular grid, a three-level AMR code solving a heat diffusion equation, and a circuit simulation.

http://theory.stanford.edu/~aiken/publications/papers/sc12.pdf

Legion: expressing locality and independence with logical regions

A method comprising providing a plurality of links to a plurality of end-user devices communicatively coupled to a network system, a particular link of the plurality of links supporting control-plane communications between the network system and a particular end-user device of the plurality of end-user devices over one or more wireless access networks; receiving a message from a server communicatively coupled to the network system, the message comprising payload for delivery to the particular end-user device; generating an encrypted message comprising the payload and an identifier identifying a particular device agent of a plurality of device agents on the particular end-user device, the identifier configured to assist in delivering at least a portion of the payload to the particular device agent on the particular end-user device; and sending the encrypted message to the particular end-user device over the particular link.

Automated Device Provisioning and Activation

Disclosed herein are methods, systems, and apparatuses to enable subscribers of mobile wireless communication devices to view, research, select and customize service plans; to create and manage device groups, share and set permission controls for service plans among devices in device groups; to manage communication services through graphical user interfaces; to sponsor and promote service plans; and to design, manage, and control communication services through application programming interfaces.

Service Plan Design, User Interfaces, Application Programming Interfaces, and Device Management

Device Assisted Services (DAS) for protecting network capacity is provided. In some embodiments, DAS for protecting network capacity includes monitoring a network service usage activity of the communications device in network communication; classifying the network service usage activity for differential network access control for protecting network capacity; and associating the network service usage activity with a network service usage control policy based on a classification of the network service usage activity to facilitate differential network access control for protecting network capacity.

Device-Assisted Services for Protecting Network Capacity

A method performed by an end user device associated with a service plan having a limit on usage of a network service, the method comprising storing one or more notification actions corresponding to one or more notification requests; performing a device action that reflects a past or intended use of the network service; receiving one of the one or more notification requests from a network element in response to the device action; performing one of the one or more notification actions in response to the notification request, the notification action causing the end user device to retrieve at least a portion of a notification message associated with a status of the use, the at least a portion of the notification message being separate from the one of the one or more notification requests; and presenting the notification message on a user interface of the end user device.

System and Method for Providing User Notifications

Methods, apparatus, and products are disclosed for providing policy-based operating system services in a hypervisor on a computing system. The computing system includes at least one compute node. The compute node includes an operating system and a hypervisor. The operating system includes a kernel. The hypervisor comprising a kernel proxy and a plurality of operating system services of a service type. Providing policy-based operating system services in a hypervisor on a computing system includes establishing, on the compute node, a kernel policy specifying one of the operating system services of the service type for use by the kernel proxy, and accessing, by the kernel proxy, the specified operating system service. The computing system may also be implemented as a distributed computing system that includes one or more operating system service nodes. One or more of the operating system services may be distributed among the operating system service nodes.

Providing policy-based operating system services in a hypervisor on a computing system

This paper presents novel cache optimizations for massively parallel, throughput-oriented architectures like GPUs. L1 data caches (L1 D-caches) are critical resources for providing high-bandwidth and low-latency data accesses. However, the high number of simultaneous requests from single-instruction multiple-thread (SIMT) cores makes the limited capacity of L1 D-caches a performance and energy bottleneck, especially for memory-intensive applications. We observe that the memory access streams to L1 D-caches for many applications contain a significant amount of requests with low reuse, which greatly reduce the cache efficacy. Existing GPU cache management schemes are either based on conditional/reactive solutions or hit-rate based designs specifically developed for CPU last level caches, which can limit overall performance. To overcome these challenges, we propose an efficient locality monitoring mechanism to dynamically filter the access stream on cache insertion such that only the data with high reuse and short reuse distances are stored in the L1 D-cache. Specifically, we present a design that integrates locality filtering based on reuse characteristics of GPU workloads into the decoupled tag store of the existing L1 D-cache through simple and cost-effective hardware extensions. Results show that our proposed design can dramatically reduce cache contention and achieve up to 56.8% and an average of 30.3% performance improvement over the baseline architecture, for a range of highly-optimized cache-unfriendly applications with minor area overhead and better energy efficiency. Our design also significantly outperforms the state-of-the-art CPU and GPU bypassing schemes (especially for irregular applications), without generating extra L2 and DRAM level contention.

/pdf/locality-driven-dynamic-gpu-cache-bypassing-54wcrnrsoq.pdf

Locality-Driven Dynamic GPU Cache Bypassing

Methods, apparatus, and products are disclosed for providing policy-based operating system services in an operating system on a computing system. The computing system includes at least one compute node. The compute node includes an operating system that includes a kernel and a plurality of operating system services of a service type. Providing policy-based operating system services in an operating system on a computing system includes establishing, on the compute node, a kernel policy specifying one of the operating system services of the service type for use in the operating system, and accessing, by the kernel, the specified operating system service. The computing system may also be implemented as a distributed computing system that includes one or more operating system service nodes. One or more of the operating system services may be distributed among the operating system service nodes.

Providing policy-based operating system services in an operating system on a computing system

GPUs have been proven effective for structured applications that map well to the rigid 1D-3D grid of threads in modern bulk synchronous parallel (BSP) programming languages. However, less success has been encountered in mapping data intensive irregular applications such as graph analytics, relational databases, and machine learning. Recently introduced nested device-side kernel launching functionality in the GPU is a step in the right direction, but still falls short of being able to effectively harness the GPUs performance potential. We propose a new mechanism called Dynamic Thread Block Launch (DTBL) to extend the current bulk synchronous parallel model underlying the current GPU execution model by supporting dynamic spawning of lightweight thread blocks. This mechanism supports the nested launching of thread blocks rather than kernels to execute dynamically occurring parallel work elements. This paper describes the execution model of DTBL, device-runtime support, and microarchitecture extensions to track and execute dynamically spawned thread blocks. Experiments with a set of irregular data intensive CUDA applications executing on a cycle-level simulator show that DTBL achieves average 1.21x speedup over the original flat implementation and average 1.40x over the implementation with device-side kernel launches using CUDA Dynamic Parallelism.

/pdf/dynamic-thread-block-launch-a-lightweight-execution-30fm2cwl40.pdf

Dynamic thread block launch: a lightweight execution mechanism to support irregular applications on GPUs

Embodiments off the invention provide a mechanism for process migration on a massively parallel computer system. In particular, embodiments of the invention may be used to update process state data for a migrated compute node, such as MPI (or other communication library) state data, across a full collection of compute nodes present in a given parallel system executing a parallel task. Migrating a process form one compute node to another may be useful to address a variety of sub-optimal operating conditions. For example, one or more processes may be migrated to cure network congestion resulting from a poorly mapped task or when a compute node is predicted to experience a hardware failure.

Albert Sidelnik

Papers

Providing policy-based operating system services in a hypervisor on a computing system

Locality-Driven Dynamic GPU Cache Bypassing

Providing policy-based operating system services in an operating system on a computing system

Dynamic thread block launch: a lightweight execution mechanism to support irregular applications on GPUs

Mechanism for process migration on a massively parallel computer