Palacios is a new open-source VMM under development at Northwestern University and the of New Mexico that enables applications executing in virtualized environment to achieve scalable high performance on large machines. functions as modularized extension Kitten, operating system being developed Sandia National Laboratories support large-scale supercomputing applications. Together, Kitten provide thin layer over hardware full-featured environments alongside Kitten's lightweight native...

Graphics processing units (GPUs) are increasingly critical for general-purpose parallel performance. GPU hardware is composed of many streaming multiprocessors, each which employs the single-instruction multiple-data (SIMD) execution style. This massively architecture allows GPUs to execute tens thousands threads in parallel. Thus, architectures efficiently heavily data-parallel applications. However, due this SIMD style, resource utilization and thus overall performance can be significantly...

In the inventory routing problem, supplier acts as a central decision maker who determines replenishment quantities and also, delivery times routes to all retailers. “Inventory Routing Problem Under Uncertainty”, Cui et al. develop novel framework for uncertain problem allow ambiguity in probability distribution of each retailer's demand. Adopting service-level viewpoint, they minimize risk levels violating prespecified acceptable range. They quantify that using new criterion, service...

Abstract Motivation: Biopathways are often modeled as systems of ordinary differential equations (ODEs). Such will usually have many unknown parameters and hence be difficult to calibrate. Since the data available for calibration limited precision, an approximate representation ODEs dynamics should suffice. One must, however, able efficiently construct such approximations large models perform model subsequent analysis. Results: We present a graphical processing unit (GPU) based scheme by...

It is now possible to allow VMs hosting HPC applications seamlessly bridge distributed cloud resources and tightly-coupled supercomputing cluster resources. However, achieve the application performance that are capable of, it important overlay network not introduce significant overhead relative native hardware, which case for current user-level tools, including our own existing VNET/U system. In response, we describe design, implementation, evaluation of a layer 2 virtual networking system...

To improve the unobtrusiveness and comfortableness of sleep monitoring, we proposed design implementation a smart mat which utilized flexible pressure sensor array printed electrodes to monitor physiological behavioral data during sleep. With novel soft seven-layer structure, system can measure distribution images calculate respiratory rate subjects. The function this was realized with acquisition circuit controlled by Arduino processor. Experiments for tests measurements were carried out...

Real-time 3D sound localization is an important technology for various applications such as camera steering systems, robotics audition, and gunshot direction. adds a new dimension, but also significantly increases the computational requirements. continuously processes large volumes of data each possible direction acoustic frequency range. Such highly demanding compute requirements outpace current CPU abilities. This paper develops real-time implementation on Graphical Processing Units...

We study a two‐stage resource pooling problem with multiple resources and customers. The central decision‐maker decides the capacity level of within total budget before realization uncertain demand. Then, fulfillment policy is determined by individual service‐level requirements. use robust satisficing framework to formulate allow ambiguity in distribution Moreover, we introduce new utility‐based probability distance allowing model be solved exactly using column constraint generation...

Real-time 3D sound localization is an important technology for various applications such as camera steering systems, robotics audition, and gunshot direction. adds a new dimension, but also significantly increases the computational requirements. continuously processes large volumes of data each possible direction acoustic frequency range. Such highly demanding compute requirements outpace current CPU abilities. This paper develops real-time implementation on Graphical Processing Units...

Overlay-based virtual networking provides a powerful model for realizing distributed and parallel computing systems with strong isolation, portability, recoverability properties. However, in extremely high throughput low latency networks, such overlays can suffer from bandwidth limitations, which is of particular concern if we want to apply the HPC environments. Through careful study an existing very performance overlay-based network system, have identified two core issues limiting...

GPUs are an increasingly popular implementation platform for a variety of general purpose applications from mobile and embedded devices to high performance computing. The CUDA OpenCL parallel programming models enable easy utilization the GPU's resources. However, tuning GPU applications' is complex labor intensive task. Software programmers employ optimization techniques explore tradeoffs between thread parallelism single thread. prior ignore register allocation, significant factor in and,...

Overlay-based virtual networking provides a powerful model for realizing distributed and parallel computing systems with strong isolation, portability, recoverability properties. However, in extremely high throughput low latency networks, such overlays can suffer from bandwidth limitations, which is of particular concern if we want to apply the HPC environments. Through careful study an existing very performance overlay-based network system, have identified two core issues limiting...

We present a novel code generation scheme for GPUs. Its key feature is the platform-aware of heterogeneous pool threads. This exposes more data-sharing opportunities among concurrent threads and reduces memory requirements that would otherwise exceed capacity on-chip memory. Instead conventional strategy focusing on exposing as much parallelism possible, our leverages phased nature access patterns found in many applications exhibit massive parallelism. demonstrate effectiveness computational...

Ethernet overlay networks are a powerful tool for virtualizing networked applications. Their performance suffers on advanced interconnects such as Infiniband, however, because of differences between the semantics and underlying network. In this paper, we demonstrate that providing virtual TCP offload device to guest operating system dramatically improves network like Infiniband. The enables leverage characteristics maximize performance. Our evaluation shows approach allows applications...

Ethernet overlay networks are a powerful tool for virtualizing networked applications. Their performance suffers on advanced interconnects such as Infiniband, however, because of differences between the semantics and underlying network. In this paper, we demonstrate that providing virtual TCP offload device to guest operating system dramatically improves network like Infiniband. The enables leverage characteristics maximize performance. Our evaluation shows approach allows applications...

We study an uncertain inventory routing problem with a finite horizon. The supplier acts as central planner who determines replenishment quantities and also the times routes for delivery to all retailers. allow ambiguity in probability distribution of each retailer's demand. Adopting service-level viewpoint, we minimize risk levels violating pre-specified acceptable range. quantify that using novel decision criterion, Service Violation Index, accounts how often severely requirement is...

In this paper, we propose a decision criterion that characterizes an enveloping bound on monetary risk measures and is computationally friendly. We start by extending the classical value at (VaR) measure. Whereas VaR evaluates threshold loss such from position exceeding given probability level, it fails to indicate performance guarantee other levels. define probabilistic measure (PEM) establish information for tail of all Using set normative properties, then generalize PEM (REM) general...