检索结果-内蒙古大学图书馆

IEEE/ACM International Symposium on Cluster Computing and the Grid (CCGRID)

作者： Paolo Notaro Qiao Yu Soroush Haeri Jorge Cardoso Michael Gerndt Huawei Technologies Duesseldorf GmbH Munich Germany Chair of Computer Architecture and Parallel Systems Technical University of Munich Germany Chair of Distributed and Operating Systems Technical University of Berlin Germany Department of Informatics Engineering/CISUC University of Coimbra Portugal

The increasing demand for cloud computing drives the expansion in scale of datacenters and their internal optical network, in a strive for increasing bandwidth, high reliability, and lower latency. Optical transceivers are essential elements of optical networks, whose reliability has not been well-studied compared to other hardware components. In this paper, we leverage high quantities of monitoring data from optical transceivers and OS-level metrics to provide statistical insights about the occurrence of optical transceiver failures. We estimate transceiver failure rates and normal operating ranges for monitored attributes, correlate early-observable patterns to known failure symptoms, and finally develop failure prediction models based on our analyses. Our results enable network administrators to deploy early-warning systems and enact predictive maintenance strategies, such as replacement or traffic re-routing, reducing the number of incidents and their associated costs.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Towards the Munich Quantum Software Stack: Enabling Efficient Access and Tool Support for Quantum computers

Towards the Munich Quantum Software Stack: Enabling Efficien...

引用

Quantum Computing and Engineering (QCE), IEEE International Conference on

作者： Martin Schulz Laura Schulz Martin Ruefenacht Robert Wille Chair of Computer Architecture and Parallel Systems Technical University of Munich Munich Germany QCT Department Leibniz Supercomputing Centre Garching Germany Chair of Design Automation Technical University of Munich Munich Germany

As quantum computing systems mature and move from laboratories to production computing environments, corresponding software stacks are becoming key for their successful utilization. In particular, the expected use of quantum systems as HPC accelerators requires a deep integration with the existing and widely deployed HPC software stacks. Additionally, new requirements such as dynamic compilation and new challenges for tools and programming models must be considered. We tackle these challenges by developing the Munich Quantum Software Stack—a comprehensive initiative by the Munich Quantum Valley to offer a flexible, efficient, and user-oriented software environment. In this poster, we describe the core components and workflows, and how they will enable this transformation from quantum experiments to quantum accelerators.

关键词：

来源：评论

学校读者我要写书评

暂无评论

FedLesScan: Mitigating Stragglers in Serverless Federated Learning

FedLesScan: Mitigating Stragglers in Serverless Federated Le...

引用

IEEE International Conference on Big Data

作者： Mohamed Elzohairy Mohak Chadha Anshul Jindal Andreas Grafberger Jianfeng Gu Michael Gerndt Osama Abboud Chair of Computer Architecture and Parallel Systems Technische Universität München Garching (near Munich) Germany Huawei Technologies Munich Germany

ISBN: (纸本)9781665480468

Federated Learning (FL) is a machine learning paradigm that enables the training of a shared global model across distributed clients while keeping the training data local. While most prior work on designing systems for FL has focused on using stateful always running components, recent work has shown that components in an FL system can greatly benefit from the usage of serverless computing and Function-as-a-Service technologies. To this end, distributed training of models with severless FL systems can be more resource-efficient and cheaper than conventional FL systems. However, serverless FL systems still suffer from the presence of stragglers, i.e., slow clients due to their resource and statistical heterogeneity. While several strategies have been proposed for mitigating stragglers in FL, most methodologies do not account for the particular characteristics of serverless environments, i.e., cold-starts, performance variations, and the ephemeral stateless nature of the function instances. Towards this, we propose FedLesScan, a novel clustering-based semi-asynchronous training strategy, specifically tailored for serverless F L. FedLesScan dynamically adapts to the behavior of clients and minimizes the effect of stragglers on the overall system. We implement our strategy by extending an open-source serverless FL system called FedLess. Moreover, we comprehensively evaluate our strategy using the 2 nd generation Google Cloud Functions with four datasets and varying percentages of stragglers. Results from our experiments show that compared to other approaches FedLesScan reduces training time and cost by an average of 8% and 20% respectively while utilizing clients better with an average increase in the effective update ratio of 17.75%.

关键词： Training Costs Federated learning Computational modeling Training data Serverless computing Big Data

来源：评论

学校读者我要写书评

暂无评论

City models: past, present and future prospects

引用

Frontiers of Urban and Rural Planning 2025年第1期3卷 1-24页

作者： Ritter, Helge Herzog, Otthein Rothermel, Kurt Cohn, Anthony G. Wu, Zhiqiang Faculty of Technology Bielefeld University Bielefeld Germany College of Architecture and Urban Planning Tongji University Shanghai People’s Republic of China FB 03 University of Bremen Bremen Germany Institute of Parallel and Distributed Systems University of Stuttgart Stuttgart Germany School of Computer Science University of Leeds Leeds UK The Alan Turing Institute London UK

This paper attempts to take a comprehensive look at the challenges of representing the spatio-temporal structures and dynamic processes that define a city’s overall characteristics. For the task of urban planning and urban operation, we take the stance that even if the necessary representations of these structures and processes can be achieved, the most important representation of the relevant mindsets of the citizens are, unfortunately, mostly neglected. After a review of major “traditional” urban models of structures behind urban scale, form, and dynamics, we turn to major recent modeling approaches triggered by recent advances in AI that enable multimodal generative models. Some of these models can create representations of geometries, networks and images, and reason flexibly at a human-compatible semantic level. They provide huge amounts of knowledge extracted from huge collections of text and image documents and cover the required rich representation spectrum including geographic knowledge by different knowledge sources, degrees of granularity and scales. We then discuss what these new opportunities mean for coping with the modeling challenges posed by cities, in particular with regard to the role and impact of citizens and their interactions within the city infrastructure. We propose to integrate these possibilities with existing approaches, such as agent-based models, which opens up new modeling spaces including rich citizen models which are able to also represent social interactions. Finally, we put forward some thoughts about a vision of a “social AI in a city ecosystem” that adds relevant citizen models to state-of-the-art structural and process models. This extended city representation will enable urban planners to establish citizen-oriented planning of city infrastructures, to make them into inviting environments that reconcile and foster human culture, city resilience and sustainability.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Leveraging Hybrid Classical-Quantum Methods for Efficient Load Rebalancing in HPC

Leveraging Hybrid Classical-Quantum Methods for Efficient Lo...

引用

High Performance Computing, Networking, Storage and Analysis, SC-W: Workshops of the International Conference for

作者： Justyna Zawalska Minh Chung Katarzyna Rycerz Laura Schulz Martin Schulz Dieter Kranzlmüller Institute of Computer Science AGH University of Krakow Krakow Poland Academic Computer Center CYFRONET AGH Krakow Poland Leibniz Supercomputing Centre (LRZ) Garching bei München Germany MNM-Team Ludwig-Maximilians-Universität München (LMU) Germany Chair for Computer Architecture and Parallel Systems Technical University of Munich (TUM) Germany

ISBN: (数字)9798350355543

ISBN: (纸本)9798350355550

Load imbalance is a challenge for parallel applications in High Performance Computing (HPC). It is caused by processes having different execution times or load values, leading to idle or wait times at synchronization points, where faster processes must wait for the slowest process to catch up. To mitigate this issue, applications can employ load balancing (LB) strategies, which migrate load between processes to even out load. This is often referred to as the Load Rebalancing Problem (LRP). While many approaches solving the LRP exist, they can only be heuristics and hence further optimization potential exists. In our work, we turn to a novel approach by using hybrid classical-quantum approaches and present two versions of the constrained quadratic model for solving the LRP; the two differ in how they balance the number of qubits required with the types of applied constraints. We compare the quantum-based methods with classical methods using heuristic algorithms Greedy, Karmarkar–Karp, and ProactLB. We evaluate our approaches using imbalance ratio and speedup as metrics, as well as the number of migrated tasks to indicate overhead caused by migrations. Our results show that the quantum-based methods outperform the classic methods. For example, we need only 1/4 of the number of migrated tasks in a realistic use case compared with classical methods, particularly Greedy and KK, to balance the load.

关键词： Measurement Upper bound NP-hard problem High performance computing Heuristic algorithms Qubit Load management Synchronization Optimization Load modeling

来源：评论

学校读者我要写书评

暂无评论

Modelling DVFS and UFS for Region-Based Energy Aware Tuning of HPC Applications

arXiv

引用

arXiv 2021年

作者： Chadha, Mohak Gerndt, Michael Computer Architecture and Parallel Systems Technische Universität München Garching [near Munich Germany

—Energy efficiency and energy conservation are one of the most crucial constraints for meeting the 20MW power envelope desired for exascale systems. Towards this, most of the research in this area has been focused on the utilization of user-controllable hardware switches such as per-core dynamic voltage frequency scaling (DVFS) and software controlled clock modulation at the application level. In this paper, we present a tuning plugin for the Periscope Tuning Framework which integrates fine-grained autotuning at the region level with DVFS and uncore frequency scaling (UFS). The tuning is based on a feed-forward neural network which is formulated using Performance Monitoring Counters (PMC) supported by x86 systems and trained using standardized benchmarks. Experiments on five standardized hybrid benchmarks show an energy improvement of 16.1% on average when the applications are tuned according to our methodology as compared to 7.8% for static tuning. Copyright © 2021, The Authors. All rights reserved.

关键词： Dynamic frequency scaling

来源：评论

学校读者我要写书评

暂无评论

architecture-specific performance optimization of compute-intensive FaaS functions

arXiv

引用

arXiv 2021年

作者： Chadha, Mohak Jindal, Anshul Gerndt, Michael Computer Architecture and Parallel Systems Technische Universität München Garching [near Munich Germany

FaaS allows an application to be decomposed into functions that are executed on a FaaS platform. The FaaS platform is responsible for the resource provisioning of the functions. Recently, there is a growing trend towards the execution of compute-intensive FaaS functions that run for several seconds. However, due to the billing policies followed by commercial FaaS offerings, the execution of these functions can incur significantly higher costs. Moreover, due to the abstraction of underlying processor architectures on which the functions are executed, the optimization of these functions is challenging. As a result, most FaaS functions use pre-compiled libraries generic to x86-64 leading to performance degradation. In this paper, we examine the underlying processor architectures for Google Cloud Functions (GCF) and determine their prevalence across the 19 available GCF regions. We modify, adapt, and optimize a representative set of six compute-intensive FaaS workloads written in Python using Numba, a JIT compiler based on LLVM, and present results wrt performance, memory consumption, and costs on GCF. Results from our experiments show that the optimization of FaaS functions can improve performance by 18.2x (geometric mean) and save costs by 76.8% on average for the six functions. Our results show that optimization of the FaaS functions for the specific architecture is very important. We achieved a maximum speedup of 1.79x by tuning the function especially for the instruction set of the underlying processor architecture. Copyright © 2021, The Authors. All rights reserved.

关键词： Cost functions

来源：评论

学校读者我要写书评

暂无评论

Time series mining at petascale performance 35th

Time series mining at petascale performance

引用

35th International Conference on High Performance Computing, ISC High Performance 2020

作者： Raoofy, Amir Karlstetter, Roman Yang, Dai Trinitis, Carsten Schulz, Martin Computer Architecture and Parallel Systems Technical University of Munich Garching Germany IfTA GmbH Puchheim Germany NVIDIA Munich Germany

ISBN: (纸本)9783030507428

The mining of time series data plays an important role in modern information retrieval and analysis systems. In particular, the identification of similarities within and across time series has garnered significant attention and effort over the last few years. For this task, the class of matrix profile algorithms, which create a generic structure that encodes correlations among records and dimensions—the matrix profile—is a promising approach, as it allows simplified post-processing and analysis steps by examining the resulting matrix profile structure. However, it is expensive to create a matrix profile: it requires significant computational power to evaluate the distance among all subsequence pairs in a time series, especially for very long and multi-dimensional time series with a large dimensionality. Existing approaches are limited in their scalability, as they do not target High Performance Computing systems, and—for most realistic problems—are suited only for datasets with a small dimensionality. In this paper, we introduce a novel MPI-based approach for the calculation of a matrix profile for multi-dimensional time series that pushes these limits. We evaluate the efficiency of our approach using an analytical performance model combined with experimental data. Finally, we demonstrate our solution on a 128-dimensional time series dataset of 1 million records, solving 274 trillion sorts at a sustained 1.3 Petaflop/s performance on the SuperMUC-NG system. © Springer Nature Switzerland AG 2020.

关键词： Time series

来源：评论

学校读者我要写书评

暂无评论

FedLess: Secure and Scalable Federated Learning Using Serverless Computing

FedLess: Secure and Scalable Federated Learning Using Server...

引用

IEEE International Conference on Big Data

作者： Andreas Grafberger Mohak Chadha Anshul Jindal Jianfeng Gu Michael Gerndt Chair of Computer Architecture and Parallel Systems Technische Universität München Garching (Near Munich) Germany

ISBN: (纸本)9781665445993

The traditional cloud-centric approach for Deep Learning (DL) requires training data to be collected and processed at a central server which is often challenging in privacy-sensitive domains like healthcare. Towards this, a new learning paradigm called Federated Learning (FL) has been proposed that brings the potential of DL to these domains while addressing privacy and data ownership issues. FL enables clients to learn a shared ML model while keeping the data local. However, conventional FL systems face challenges such as scalability, complex infrastructure management, and wasted compute and incurred costs due to idle clients. These challenges of FL systems closely align with the core problems that serverless computing and Function-as-a-Service (FaaS) platforms aim to solve. These include rapid scalability, no infrastructure management, automatic scaling to zero for idle clients, and a pay-per-use billing model. To this end, we present a novel system and framework for serverless FL, called FedLess. Our system supports multiple commercial and self-hosted FaaS providers and can be deployed in the cloud, on-premise in institutional data centers, and on edge devices. To the best of our knowledge, we are the first to enable FL across a large fabric of heterogeneous FaaS providers while providing important features like security and Differential Privacy. We demonstrate with comprehensive experiments that the successful training of DNNs for different tasks across up to 200 client functions and more is easily possible using our system. Furthermore, we demonstrate the practical viability of our methodology by comparing it against a traditional FL system and show that it can be cheaper and more resource-efficient.

关键词： Training Scalability Computational modeling Serverless computing Training data FAA Big Data

来源：评论

学校读者我要写书评

暂无评论

Towards Autoscaling with Guarantees on Kubernetes Clusters

Towards Autoscaling with Guarantees on Kubernetes Clusters

引用

Autonomic Computing and Self-Organizing systems Companion (ACSOS-C), IEEE International Conference on

作者： Stephen Burroughs Helge Dickel Martin van Zijl Vladimir Podolskiy Michael Gerndt Robi Malik Panos Patros ORKA Cloud and Adaptive Systems Lab University of Waikato New Zealand Chair of Computer Architecture and Parallel Systems Technical University of Munich Germany

Autoscaling is used by cloud providers, microser-vices, and edge computing applications to respond to dynamic load fluctuations. A critical direction of research has focused on providing guarantees under uncertainty that the auto scaling system will work as intended-both at design-time and more importantly, at runtime. In this work, we evaluate the efficacy of three complementary methods: A) deterministic finite automata, B) probabilistic process algebra and C) Proportional-Integral-Derivative (PID) control. We experimentally evaluate their efficacy in modelling and verify autoscaling on clusters managed by Kubernetes. Our results suggest that deterministic modelling can provide theoretically optimal guarantees for small deployments; probabilistic algebras are able to capture stochastic behaviours, but benefit from deterministic templates; and control theory benefits by providing a black-box approach for modelling, verification and control.

关键词：

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：