Search

Today’s High Performance Computing (HPC) systems contain thousand of nodes which work together to provide performance in the order of petaflops. The performance of these systems depends on various components like processors, memory, and interconnect. Among all, interconnect plays a major role as it glues together all the hardware components in an HPC system. A slow interconnect can impact a scientific application running on multiple processes severely as they rely on fast network messages to communicate and synchronize frequently. Unfortunately, the HPC community lacks a study that explores different interconnect errors, congestion events and applications characteristics on a large-scale HPC system. In our previous work, we process and analyze interconnect data of the Titan supercomputer to develop a thorough understanding of interconnects faults, errors, and congestion events. In this work, we first show how congestion events can impact application performance. We then investigate application characteristics interaction with interconnect errors and network congestion to predict applications encountering congestion with more than 90% accuracy.

Infrastructural development especially in the pavement sectors plays a key role in the economic development of developing countries. In general, pavement consists of a compacted subgrade layer as the lowest layer, wherein the soft or expansive soils in the subgrade are usually responsible for distress in pavements shortening the service life of the pavement. The present paper consists of a research work which was carried out to address some of the issues responsible for the pavement distress by treating the soil sample with the addition of agricultural waste ash such as cow dung and husk. Results of the laboratory tests showed that there was initially decrease in the optimum moisture content at 5% ash then it increased steadily afterwards and California Bearing Ratio (CBR) values of the soil increased 15 times to the original CBR values at 10% ash. An increment in the UCS values from 2.4 kg/cm2 (natural soil) to 6.3 kg/cm2 (15% ash) under 7-day curing conditions was observed. The cost analysis showed that there was a reduction in the overall cost of the whole pavement as well as a reduction in the thickness of the pavement layers.

There are conventional methods available for designing of flexible pavement based on the past experience whereas for the design of Cement Treated Base (CTB) and Cement Treated Sub Base (CTSB) layers no such conventional methods are available. Use of cement stabilization in granular material is an effective technique to increase the stiffness of base and sub base layers and also in the thickness reduction in comparison to conventional layers. Furthermore, cement base layers can improve the fatigue behaviour of asphalt surface layer and subgrade rutting over the short and long term. In our country, there is no performance data available for CTB and CTSB layers in in-situ conditions. This paper deals with the investigation of the resilient modulus (MR) of stabilized pavement layers through field data collection on in-service flexible pavement sections which may be useful for design purpose. The main objective of this research study is to make a comparison of MR values for cement stabilized layers using non-destructive method viz. Falling Weight Deflectometer (FWD) and destructives test method through cores. The field data has been collected from State Highway (SH) and Major District Road (MDR) around Lucknow, Uttar Pradesh. No visual distress was observed on both the road sections at the time of field surveys. Two study sections on SH and one on MDR of about 500 m length has been identified for the study. The in-situ cores were obtained from CTB and CTSB layers which were tested for determination of MR values through Unconfined Compressive Strength (UCS) as per IRC:37-2018. FWD deflection data was used for determination of MR values using backcalculation software for CTB and CTSB layers. Furthermore, the thickness and the properties of each layer were determined by conducting test pits on the existing pavement.

This article is based on dual fractal techniques to design the Koch–Sierpinski Fractal Microstrip (KSFM) antenna for wireless applications. The concept of the Koch fractal is applied on the outer p...

As parallel applications become more complex, auto-tuning becomes more desirable, challenging, and time-consuming. We propose, Bliss, a novel solution for auto-tuning parallel applications without requiring apriori information about applications, domain-specific knowledge, or instrumentation. Bliss demonstrates how to leverage a pool of Bayesian Optimization models to find the near-optimal parameter setting 1.64× faster than the state-of-the-art approaches.

We propose a resource allocation policy that uses (a) Power as a first class parameter as an indicator of the computational intensity of a task and its potential impact on peak power draw, and (b) Power Tolerance as an indicator of a task's sensitivity towards degradation of performance as a result of resource contention. Through experimentation and analysis, we present coarse-grained and fine-grained Power Tolerance assignment methods that can be employed to make smarter peak power performance trade-offs. Our experiments show that (a) cloud operators can benefit from a uniform workload-wide setting of Power Tolerance to achieve significant reduction in peak power consumption, (b) fine-grained Power Tolerance assignment methods show potential in making smarter peak power and performance trade-offs.

Noisy Intermediate-Scale Quantum (NISQ) computers are being increasingly used for executing early-stage quantum programs to establish the practical realizability of existing quantum algorithms. These quantum programs have uses cases in the realm of high-performance computing ranging from molecular chemistry and physics simulations to addressing NP-complete optimization problems. However, NISQ devices are prone to multiple types of errors, which affect the fidelity and reproducibility of the program execution. As the technology is still primitive, our understanding of these quantum machines and their error characteristics is limited. To bridge that understanding gap, this is the first work to provide a systematic and rich experimental evaluation of IBM Quantum Experience (QX) quantum computers of different scales and topologies. Our experimental evaluation uncovers multiple important and interesting aspects of benchmarking and evaluating quantum program on NISQ machines. We have open-sourced our experimental framework and dataset to help accelerate the evaluation of quantum computing systems.

HPC workload analysis and resource consumption characteristics are the key to driving better operation practices, system procurement decisions, and designing effective resource management techniques. Unfortunately, the HPC community does not have easy accessibility to long-term introspective work-load analysis and characterization for production-scale HPC systems. This study bridges this gap by providing detailed long-term quantification, characterization, and analysis of job characteristics on two supercomputers: Intrepid and Mira. This study is one of the largest of its kind – covering trends and characteristics for over three billion compute hours, 750 thousand jobs, and spanning a decade. We confirm several long-held conventional wisdom, and identify many previously undiscovered trends and its implications. We also introduce a learning based technique to predict the resource requirement of future jobs with high accuracy, using features available prior to the job submission and without requiring any application-specific tracing or application-intrusive instrumentation.

As we approach exascale computing, large-scale HPC systems are becoming increasingly power-constrained, requiring them to run HPC workloads in an energy-efficient manner. The first step toward achieving this goal is to better understand, analyze, and quantify the power consumption characteristics of HPC jobs. However, there is a lack of understanding of the power consumption characteristics of HPC jobs which run on production HPC systems. Such characterization is required to guide the design of the next generation of power-aware resource management. To the best of our knowledge, we are the first study to open-source the data and analysis of power-consumption characteristics of HPC jobs and users from two medium-scale production HPC clusters.

In recent years, the use of foamed bitumen technology along with Reclaimed Asphalt Pavement is gaining popularity across the world. The mechanical response of foamed bitumen mixtures containing reclaimed asphalt pavement is significantly influenced by constituent material properties and aggregate gradation. This article presents results from a study where foamed bitumen mixtures conforming to Indian specifications were evaluated. For this purpose, foamed bitumen mixtures using a different percentage of reclaimed asphalt pavement and bitumens were prepared. Initially, the foaming characteristics of virgin bitumens were evaluated to optimize for optimum water content and foaming temperature. In the second stage, mixture design was conducted to optimize for foamed bitumen content in foamed bitumen mixtures containing a different percentage of reclaimed asphalt pavement. Finally, these foamed bitumen mixtures were evaluated for their mechanical properties. The results from this laboratory study indicated properties of foamed bitumen and foamed mixtures are significantly influenced by properties of bitumen, the quantity of bitumen, and reclaimed asphalt pavement. Among the different mixtures, a mixture containing 50% reclaimed asphalt pavement exhibited best results in resilient modulus and resistance to moisture damage tests. A mixture containing 80% reclaimed asphalt pavement also shows acceptable strength and resistance to water susceptibility. Thus, it is possible to design high-quality bituminous mixes using higher reclaimed asphalt pavement percentages, which meet the required volumetric and desired performance criteria.

Foamed bitumen technology is gaining acceptance across the world as an approach towards sustainable road construction. This technology offers advantages in terms of reduced energy consumption, materials recycling and reduced bitumen requirement. The properties of foamed bitumen mixtures depend on various factors, including the properties of the constituent materials and production conditions. This paper concentrates on the influence of the properties of the binder on the foaming characteristics of bitumen and the properties of foamed bitumen mixtures. Initially, four binders of different grades were evaluated for physical properties and foaming characteristics. The optimum foam content was then used to prepare foamed bituminous mixtures containing recycled asphalt pavement. These mixtures were evaluated for their mechanical properties, including indirect tensile strength, retained tensile strength and resilient modulus. The results from this laboratory study indicate that the properties of foamed bitumen and foamed mixtures are significantly influenced by the physical properties of the binder used, the quantity of binder and the recycled asphalt content. For a particular binder content in the mixture, use of a higher viscosity grade binder resulted in higher indirect tensile strength. A higher recycled asphalt pavement content in the mixture also resulted in higher indirect tensile strength. For all binder grades, a larger resilient modulus was observed for binder content the range of 2·3–2·5%.

Online data-intensive (OLDI) services use anytime algorithms to compute over large amounts of data and respond quickly. Interactive response times are a priority, so OLDI services parallelize query execution across distributed software components and return best effort answers based on the data so far processed. Omitted data from slow components could lead to better answers, but tracing online how much better the answers could be is difficult. We propose Ubora, a design approach to measure the effect of slow-running components on the quality of answers. Ubora randomly samples online queries and executes them a second time. The first online execution omits data from slow components and provides interactive answers. The second execution uses mature results from intermediate components completed after the online execution finishes. Ubora uses memoization to speed up mature executions by replaying network messages exchanged between components. Our systems-level implementation works for a wide range of services, including Hadoop/Yarn, Apache Lucene, the EasyRec Recommendation Engine, and the OpenEphyra question-answering system. Ubora computes answer quality with more mature executions per second than competing approaches that do not use memoization. With Ubora, we show that answer quality is effective at guiding online admission control. While achieving the same answer quality on high-priority queries, our adaptive controller had 55% higher peak throughput on low-priority queries than a competing controller guided by the rate of timeouts.

We propose a new model for the study of resilience of coevolving multiplex scale-free networks. Our network model, called preferential interdependent networks, is a novel continuum over scale-free networks parameterized by their correlation $$\rho , 0 \le \rho \le 1$$ . Our failure and recovery model ties the propensity of a node, both to fail and to assist in recovery, to its importance. We show, analytically, that our network model can achieve any $$\gamma , 2 \le \gamma \le 3$$ for the exponent of the power law of the degree distribution; this is superior to existing multiplex models and allows us better fidelity in representing real-world networks. Our failure and recovery model is also a departure from the much studied cascading error model based on the giant component; it allows for surviving important nodes to send assistance to the damaged nodes to enable their recovery. This better reflects the reality of recovery in man-made networks such as social networks and infrastructure networks. Our main finding, based on simulations, is that resilient preferential interdependent networks are those in which the layers are neither completely correlated ( $$\rho = 1$$ ) nor completely uncorrelated ( $$\rho =0$$ ) but instead semi-correlated ( $$\rho \approx 0.1 - 0.3$$ ). This finding is consistent with the real-world experience where complex man-made networks typically bounce back quickly from stress. In an attempt to explain our intriguing empirical discovery, we present an argument for why semi-correlated multiplex networks can be the most resilient. Our argument can be seen as an explanation of plausibility or as an incomplete mathematical proof subject to certain technical conjectures that we make explicit.

Large-scale applications typically spend a large fraction of their execution time performing I/O to a parallel storage system. However, with rapid progress in compute and storage system stack of large-scale systems, it is critical to investigate and update our understanding of the I/O behavior of large-scale applications. Toward that end, in this work, we monitor, collect and analyze a year worth of storage system data from a large-scale production parallel storage system. We perform temporal, spatial and correlative analysis of the system and uncover surprising patterns which defy existing assumptions and have important implications for future systems.

Cloud platforms typically require users to provide resource requirements for applications so that resource managers can schedule containers with adequate allocations. However, the requirements for container resources often depend on numerous factors such as application input parameters, optimization flags, input files, and attributes that are specified for each run. So, it is complex for users to estimate the resource requirements for a given container accurately, leading to resource over-estimation that negatively affects overall utilization. We have designed a Resource Utilization Based Autoscaling System (RUBAS) that can dynamically adjust the allocation of containers running in a Kubernetes cluster. RUBAS improves upon the Kubernetes Vertical Pod Autoscaler (VPA) system non-disruptively by incorporating container migration. Our experiments use multiple scientific benchmarks. We analyze the allocation pattern of RUBAS with Kubernetes VPA. We compare the performance of container migration for in-place and remote node migration and we evaluate the overhead in RUBAS. Our results show that compared to Kubernetes VPA, RUBAS improves the CPU and memory utilization of the cluster by 10% and reduces the runtime by 15% with an overhead for each application ranging from 5% to 20%.

Large-scale computing systems are becoming increasingly more power-constrained, but these systems employ hardware over- provisioning to achieve higher system throughput because applications often do not consume the peak power capacity of nodes. Unfortunately, focusing on system throughput alone can lead to severe unfairness among multiple concurrently-running applications. This paper introduces PERQ, a new feedback-based principled approach to improve system throughput while achieving fairness among concurrent applications.

Vibration generated in modern computing environments such as autonomous vehicles, edge computing infrastructure, and data center systems is an increasing concern. In this paper, we systematically measure, quantify and characterize the impact of vibration on the performance of SSD devices. Our experiments and analysis uncover that exposure to both short-term and long-term vibration, even within the vendor-specified limits, can significantly affect SSD I/O performance and reliability.

The booming of cloud computing, online services and big data applications have resulted in dramatic expansion of storage systems. Meanwhile, disk drives are reported to be the most commonly replaced hardware component. Disk failures cause service downtime and even data loss, costing enterprises multi-trillion dollars per year. Existing disk failure management approaches are mostly reactive and incur high overheads. To overcome these problems, in this paper, we present a proactive, cost-effective solution to managing large-scale production storage systems. We aim to uncover the entire process in which disk's health deteriorates and forecast when disk drives will fail in the future. Due to a common lack of diagnostic information of disk failures, we rely on the Self-Monitoring, Analysis and Reporting Technology (SMART) data and explore statistical analysis techniques to identify the start of disk degradation. We then model the disk degradation processes as functions of SMART attributes, which eliminates the dependency on time and thus I/O workload. Experimental results from over 23,000 enterprise-class disk drives in a production data center show that our derived models can accurately quantify the degradation of disk health, which enables us to proactively protect data against disk failures. We also investigate several types of disk failures and propose remediation mechanisms to prolong disk lifetime.

Academic cloud infrastructures require users to specify an estimate of their resource requirements. The resource usage for applications often depends on the input file sizes, parameters, optimization flags, and attributes, specified for each run. Incorrect estimation can result in low resource utilization of the entire infrastructure and long wait times for jobs in the queue. We have designed a Resource Utilization based Migration (RUMIG) system to address the resource estimation problem. We present the overall architecture of the two-stage elastic cluster design, the Apache Mesos-specific container migration system, and analyze the performance for several scientific workloads on three different cloud/cluster environments. In this paper we (b) present a design and implementation for container migration in a Mesos environment, (c) evaluate the effect of right-sizing and cluster elasticity on overall performance, (d) analyze different profiling intervals to determine the best fit, (e) determine the overhead of our profiling mechanism. Compared to the default use of Apache Mesos, in the best cases, RUMIG provides a gain of 65% in runtime (local cluster), 51% in CPU utilization in the Chameleon cloud, and 27% in memory utilization in the Jetstream cloud.

Reliability has become a first-class design objective for GPU devices due to increasing soft-error rate. To assess the reliability of GPU programs, researchers rely on software fault-injection methods. Unfortunately, software fault-injection process is prohibitively expensive, requiring multiple days to complete a statistically sound fault-injection campaign.Therefore, to address this challenge, this paper proposes a novel fault-injection method, PCFI, that reduces the number of fault injections by exploiting the predictability in fault-injection outcome based on the program counter of the soft-error affected instruction. Evaluation on a variety of GPU programs covering a wide range of application domains shows that PCFI reduces the time to complete fault-injection campaigns by 22% on average, without sacrificing accuracy.

This article presents Clover, a compiler-directed soft error detection and recovery scheme for lightweight soft error resilience. The compiler carefully generates soft-error-tolerant code based on idempotent processing without explicit checkpoints. During program execution, Clover relies on a small number of acoustic wave detectors deployed in the processor to identify soft errors by sensing the wave made by a particle strike. To cope with DUEs (detected unrecoverable errors) caused by the sensing latency of error detection, Clover leverages a novel selective instruction duplication technique called tail-DMR (dual modular redundancy) that provides a region-level error containment. Once a soft error is detected by either the sensors or the tail-DMR, Clover takes care of the error as in the case of exception handling. To recover from the error, Clover simply redirects program control to the beginning of the code region where the error is detected. The experimental results demonstrate that the average runtime overhead is only 26%, which is a 75% reduction compared to that of the state-of-the-art soft error resilience technique. In addition, this article evaluates an alternative technique called tail-wait, comparing it to Clover. According to the evaluation with the different processor configurations and the various error detection latencies, Clover turns out to be a superior technique, achieving 1.06 to 3.49 × speedup over the tail-wait.

India is developing her national highway network through widening and rehabilitation of existing highways along with the construction of expressways in different phases, since 1999. Unprecedented growth of road traffic, high variations in pavement temperature and need of long lasting pavements have increased the use of modified bitumen specifically in wearing courses of many flexible pavement road sections of national highway network in entire country. Crumb rubber-modified bitumen (CRMB) and polymer-modified bitumen (PMB) of different grades are mostly used modified binders under different climatic and environmental conditions. During the design life, bituminous road sections show different rates of initiation and propagation of distresses under varying traffic and climatic conditions. In this study, an effort has been made to calibrate the internationally recognised Highway Development & Management (HDM-4) road deterioration models for the selected flexible pavement sections over time with traff...

Modern computer systems are designed to balance performance and energy consumption. Several run-time factors, such as concurrency levels, thread mapping strategies, and dynamic voltage and frequency scaling (DVFS) should be considered in order to achieve optimal energy efficiency for a workload. Selecting appropriate run-time factors, however, is one of the most challenging tasks because the run-time factors are architecture-specific and workload-specific.While most existing works concentrate on either static analysis of the workload or run-time prediction results, in this paper, we present a hybrid two-step method that utilizes concurrency levels and DVFS settings to achieve the energy efficiency configuration for a workload. The experimental results based on a Xeon E5620 server with NPB and PARSEC benchmark suites show that the model is able to predict the energy efficient configuration accurately. On average, an additional 10 % EDP (Energy Delay Product) saving is obtained by using run-time DVFS for the entire system. An off-line optimal solution is used to compare with the proposed scheme. The experimental results show that the average extra EDP saved by the optimal solution is within 5 % on selective parallel benchmarks. We present a hybrid method to achieve an energy efficiency configuration.Our method utilizes concurrency levels, thread allocation, and DVFS settings.We propose a model to capture the relationship between C , P , and T in detail.We apply an analytical speedup model to predict an optimal/nearoptimal configuration.

The Indian Road Network is the second largest in the world after U.S. and flexible pavement is the most sought pavement in India. In fact, majority of roads in India (more then 90%) are flexible pavement due to ease of construction and initial cost of construction. The design of flexible pavement in India mainly consider rutting and fatigue life as per Indian Road Congress (IRC:37-2001). There are many factors that affects the rutting of the flexible pavements namely speed, contact pressure, temperature, quality of construction, wandering of the vehicle and layer thickness etc. The present paper focuses on rutting itself which is governed by direction of traffic, while keeping other parameters nearly same. The effect of movement of heavy traffic in a channelized as well non channelized way on pavement resulting in rutting has been studied with the state of art equipment, Accelerated Pavement Testing Facility (APTF), by CSIR-CRRI, with the aim of finding the performance through evaluation for uni and bi-directional rutting characteristic of pavement. Rutting is the longitudinal depression in the wheel path of the pavement. It can either be instability or structural rutting. Rutting can cause hydroplaning, increase in vehicle operating cost and psychological problems while driving. The amount of rutting under unidirectional loading was compared with the bidirectional loading case with almost the same number of passes. The change in air voids were also calculated from the rutted as well as non rutted portion. The magnitude of rutting under unidirectional loading was more as compared to the one under bi-directional loading. The comparative study has clearly demonstrated that the damage to flexible pavement by rutting with unidirectional load is more critical and nearly twice as compared to the damage by bi-directional loading. This aspect stresses the need for accounting the directional movements in our pavement design guidelines and depending on whether testing uni-directional or bi-directional, a different “transfer function” or relationship to real in service loading is required.

A multi-band, high gain, miniaturized fractal microstrip antenna is described in this article for modern communication systems. Space-utilizing and structure-similarity are the two main characteristics of fractal miniaturization technique. In the proposed design, a hexagonal Sierpinski gasket structure is loaded on a square microstrip antenna. Using fractal miniaturization techniques, the patch area is reduced by 68.4% along with 168.8% increase in the perimeter by loading several triangular slots of different dimensions in various iterations. The simulation and optimization of various iterations are analysed using Ansoft HFSS simulation tool and verified with fabricated antenna results. The proposed design presents the hexa-band, dual-polarization performance at 3.46, 8.28, 12.26, 17.21, 23.40, 26.01 frequencies (GHz) with Peak Gain (dB) of 6, 8.37, 9.65, 9, 7.84 and 9.34 respectively. The designed miniaturized hexagonal fractal antenna exhibits high gain, high directivity, omnidirectional radiation pattern entire multi-resonant frequency and useful for modern wireless communications like 5G and IoT application, satellite and Radar applications etc.

In recent years, NAND flash-based solid state drives (SSD) have been widely used in datacenters due to their better performance compared with the traditional hard disk drives. However, little is known about the reliability characteristics of SSDs in production systems. Existing works study the statistical distributions of SSD failures in the field. However, they do not go deep into SSD drives and investigate the unique error types and health dynamics that distinguish SSDs from hard disk drives. In this paper, we explore the SSD-specific SMART (Self-Monitoring, Analysis, and Reporting Technology) attributes to conduct an in-depth analysis of SSD reliability in a production environment. Data is collected from a scalable production system having several physical locations. Our dataset contains over a million records with more than twenty attributes. We leverage machine learning technologies, specifically data clustering and correlation analysis methods, to discover groups of SSDs which have different health status and relations among SSD-specific SMART attributes. Our results show that 1) Media wear affects the reliability of SSDs more than any other factors, and 2) SSDs transit from one health group to another which infers the reliability degradation of those drives. To the best of our knowledge, this is the first study that investigates SSD-specific SMART data to characterize SSD reliability in a production environment.

We propose a new model for the study of resilience of coevolving multiplex scale-free networks. Our network model, called preferential interdependent networks, is a novel continuum over scale-free networks parameterized by their correlation p, 0 ≤ $p$ ≤1. Our failure and recovery model ties the propensity of a node, both to fail and to assist in recovery, to its importance. We show, analytically, that our network model can achieve any γ, 2 ≤ γ ≤ 3 for the exponent of the power law of the degree distribution; this is superior to existing multiplex models and allows us better fidelity in representing real-world networks. Our failure and recovery model is also a departure from the much studied cascading error model based on the giant component; it allows for surviving important nodes to send assistance to the damaged nodes to enable their recovery. This better reflects the reality of recovery in man-made networks such as social networks and infrastructure networks. Our main finding, based on simulations, is that resilient preferential interdependent networks are those in which the layers are neither completely correlated (p = 1) nor completely uncorrelated (p= 0) but instead semi-correlated (p ≈ 0.1 – 0.3). This finding is consistent with the real-world experience where complex man-made networks typically bounce back quickly from stress. In an attempt to explain our intriguing empirical discovery we present an argument for why semi-correlated multiplex networks can be the most resilient. Our argument can be seen as an explanation of plausibility or as an incomplete mathematical proof subject to certain technical conjectures that we make explicit.

In high-performance computing (HPC) workflows, data analytics is typically utilized to gain insights from scientific simulations. Approaching the era of exascale, online analysis is gaining popularity due to the savings of I/O to persistent storage. As computing capability keeps growing, power consumption is becoming critical to HPC facilities. Enforcing power limits is emerging as a practical trend for power-constrained HPC facilities. However, it remains unclear how to choose the appropriate power limits for various HPC workflows and how to distribute the power limit of a workflow between simulation and analysis. In addition, given a power limit, it is unclear what the optimal scales and power capping levels are for various workflows, especially when taking reliability into account. In order to resolve these issues in power-constrained HPC, in this paper, we propose a reliability-aware model to determine the aforementioned platform configurations for HPC workflows. We also validate our model and present model-driven studies for a wide range of real-system scenarios. Our study reveals interesting insights about how platform configuration affects the performance and energy efficiency of HPC workflows under power constraints.

GPUs are widely deployed on large-scale HPC systems to provide powerful computational capability for scientific applications from various domains. As those applications are normally long-running, investigating the characteristics of GPU errors becomes imperative for reliability. In this paper, we first study the system conditions that trigger GPU errors using six-month trace data collected from a large-scale, operational HPC system. Then, we use machine learning to predict the occurrence of GPU errors, by taking advantage of temporal and spatial dependencies of the trace data. The resulting machine learning prediction framework is robust and accurate under different workloads.

Today's High Performance Computing (HPC) systems are capable of delivering performance in the order of petaflops due to the fast computing devices, network interconnect, and back-end storage systems. In particular, interconnect resilience and congestion resolution methods have a major impact on the overall interconnect and application performance. This is especially true for scientific applications running multiple processes on different compute nodes as they rely on fast network messages to communicate and synchronize frequently. Unfortunately, the HPC community lacks state-of-practice experience reports that detail how different interconnect errors and congestion events occur on large-scale HPC systems. Therefore, in this paper, we process and analyze interconnect data of the Titan supercomputer to develop a thorough understanding of interconnects faults, errors and congestion events. We also study the interaction between interconnect, errors, network congestion and application characteristics.

In this paper, we describe the GUIDE framework used to collect, federate, and analyze log data from the Oak Ridge Leadership Computing Facility (OLCF), and how we use that data to derive insights into facility operations. We collect system logs and extract monitoring data at every level of the various OLCF subsystems, and have developed a suite of pre-processing tools to make the raw data consumable. The cleansed logs are then ingested and federated into a central, scalable data warehouse, Splunk, that offers storage, indexing, querying, and visualization capabilities. We have further developed and deployed a set of tools to analyze these multiple disparate log streams in concert and derive operational insights. We describe our experience from developing and deploying the GUIDE infrastructure, and deriving valuable insights on the various subsystems, based on two years of operations in the production OLCF environment.

Resilience is one of the key challenges in maintaining high efficiency of future extreme scale supercomputers. Researchers and system practitioners rely on field-data studies to understand reliability characteristics and plan for future HPC systems. In this work, we compare and contrast the reliability characteristics of multiple large-scale HPC production systems. Our study covers more than one billion compute node hours across five different systems over a period of 8 years. We confirm previous findings which continue to be valid, discover new findings, and discuss their implications.

GPUs have become part of the mainstream high performance computing facilities that increasingly require more computational power to simulate physical phenomena quickly and accurately. However, GPU nodes also consume significantly more power than traditional CPU nodes, and high power consumption introduces new system operation challenges, including increased temperature, power/cooling cost, and lower system reliability. This paper explores how power consumption and temperature characteristics affect reliability, provides insights into what are the implications of such understanding, and how to exploit these insights toward predicting GPU errors using neural networks.

HPC (high-performance computing) applications usually show bursty I/O behaviors. In order to expedite the applications, permanent storage systems are usually provisioned to serve such I/O bursts. Approaching the era of exascale computing, non-volatile RAM is introduced as burst buffers, to absorb the bursty bulk data and relax the I/O provisioning requirement of the permanent storage systems. However, without judiciously draining the burst buffers, I/O bursts are passed down to the underlying storage systems, which causes severe I/O contention issues.In order to minimize the I/O provisioning requirement and resolve the issues caused by I/O bursts, we propose a proactive draining scheme to manage the draining process of distributed node-local burst buffers. In addition, we develop an I/O provisioning model to predict the minimized I/O provisioning requirement for permanent storage systems. Evaluation results show that applying the proactive draining scheme largely relaxes the I/O provisioning requirement while preserving the I/O performance of underlying storage systems.

In high-performance computing (HPC), end-to-end workflows are typically utilized to gain insights from scientific simulations. An end-to-end workflow consists of scientific simulation and data analysis, and can be executed in-situ, in-transit, and offline. Existing studies on end-to-end workflows have largely focused on the high-performance execution approaches. However, the emerging heterogeneous architectures and energy concerns lead to the rethinking of workflow execution approaches. As a guide to the rethinking, this paper evaluates how to run end-to-end HPC workflows efficiently in terms of performance, energy, and error resilience. The evaluation covers emerging heterogeneous processor architectures, processor power capping techniques, and heterogeneous-reliability memory.

Transient faults continue to be a critical concern in a range of computing domains including: High-Performance Computing (HPC), scientific computing, and the automotive industry. While radiation-induced faults have been well studied and understood in microprocessors, their impact on computations on Graphic Processing Units (GPU) has received less attention. GPUs are now being used in a large number of HPC and automotive markets. Mitigating the effects of transient faults requires a thorough understanding of the interaction between applications, system software, and the underlying hardware. Developing this understanding is quite challenging mainly due to our limited ability to capture and study cross-layer reliability interactions. In this paper, we consider the combination of neutron beam testing experiments with architectural fault injection experiments to gain a deeper understanding of the relationship between the vulnerability of GPUs and the underlying workload characteristics of applications targeted for GPU devices.

State of the art schedulers use workload profiles to help determine which resources to allocate. Traditionally, threads execute on every available core, but increasingly, too much power is consumed by using every core. Because peak power can occur at any point in time during the workload, workloads are commonly profiled to completion multiple times in an offline architecture. In practice, this process is too time consuming for online profiling and alternate approaches are used, such as profiling for k% of the workload or predicting peak power from similar workloads. We studied the effectiveness of these methods for core scaling. Core scaling is a technique which executes threads on a subset of available cores, allowing unused cores to enter low-power operating modes. Schedulers can use core scaling to reduce peak power, but must have an accurate profile across potential settings for number of active cores in order to know when to make this decision. We devised an accurate, fast and adaptive approach to profile peak power under core scaling. Our approach uses short profiling runs to collect instantaneous power traces for a workload under each core scaling setting. The duration of profiling varies for each power trace and depends on the desired accuracy. Compared to k% profiling of peak power, our approach reduced the profiling duration by up to 93% while keeping accuracy within 3%.

Checkpoint and restart mechanisms have been widely used in large scientific simulation applications to make forward progress in case of failures. However, none of the prior works have considered the interaction of power-constraint with temperature, reliability, performance, and checkpointing interval. It is not clear how power-capping may affect optimal checkpointing interval. What are the involved reliability, performance, and energy trade-offs? In this paper, we develop a deep understanding about the interaction between power-capping and scientific applications using checkpoint/restart as resilience mechanism, and propose a new model for the optimal checkpointing interval (OCI) under power-capping. Our study reveals several interesting, and previously unknown, insights about how power-capping affects the reliability, energy consumption, performance.

Resilience is an important challenge for extreme-scale supercomputers. Today, failures in supercomputers are assumed to be uniformly distributed in time. However, recent studies show that failures in high-performance computing systems are partially correlated in time, generating periods of higher failure density. Our study of the failure logs of multiple supercomputers show that periods of higher failure density occur with up to three times more than the average. We design a monitoring system that listens to hardware events and forwards important events to the runtime to detect those regime changes. We implement a runtime capable of receiving notifications and adapt dynamically. In addition, we build an analytical model to predict the gains that such dynamic approach could achieve. We demonstrate that in some systems, our approach can reduce the wasted time by over 30%.

Online data-intensive services parallelize query execution across distributed software components. Interactive response time is a priority, so online query executions return answers without waiting for slow running components to finish. However, data from these slow components could lead to better answers. We propose Ubora, an approach to measure the effect of slow running components on the quality of answers. Ubora randomly samples online queries and executes them twice. The first execution elides data from slow components and provides fast online answers; the second execution waits for all components to complete. Ubora uses memoization to speed up mature executions by replaying network messages exchanged between components. Our systems-level implementation works for a wide range of platforms, including Hadoop/Yarn, Apache Lucene, the EasyRec Recommendation Engine, and the OpenEphyra question answering system. Ubora computes answer quality much faster than competing approaches that do not use memoization. With Ubora, we show that answer quality can and should be used to guide online admission control. Our adaptive controller processed 37% more queries than a competing controller guided by the rate of timeouts., Comment: Technical Report