检索结果-内蒙古大学图书馆

IEEE international symposium on parallel and distributed Processing Workshops and Phd Forum (IPDPSW)

作者： Jaemin Choi David F. Richards Laxmikant V. Kale Department of Computer Science University of Illinois at Urbana-Champaign Urbana Illinois USA Center for Applied Scientific Computing Lawrence Livermore National Laboratory Livermore California USA

ISBN: (数字)9781665497473

ISBN: (纸本)9781665497480

Asynchronous tasks, when created with over-decomposition, enable automatic computation-communication overlap which can substantially improve performance and scal-ability. This is not only applicable to traditional CPU-based systems, but also to modern GPU -accelerated platforms. While the ability to hide communication behind computation can be highly effective in weak scaling scenarios, performance begins to suffer with smaller problem sizes or in strong scaling due to fine-grained overheads and reduced room for overlap. In this work, we integrate G PU -aware communication into asynchronous tasks in addition to computation-communication overlap, with the goal of reducing time spent in communication and further increasing GPU utilization. We demonstrate the performance impact of our approach using a proxy application that performs the Jacobi iterative method, Jacobi3D. In addition to optimizations to minimize synchronizations between the host and GPU devices and increase the concurrency of GPU operations, we explore techniques such as kernel fusion and CUDA Graphs to mitigate fine-grained overheads at scale.

关键词： Jacobian matrices Performance evaluation Concurrent computing Scalability Graphics processing units Supercomputers Iterative methods

来源：评论

学校读者我要写书评

暂无评论

SSB-Tree: Making Persistent Memory B+- Trees Crash-Consistent and Concurrent by Lazy-Box

SSB-Tree: Making Persistent Memory B+- Trees Crash-Consisten...

引用

international symposium on parallel and distributed Processing (IPDPS)

作者： Tongliang Li Haixia Wang Airan Shao Dongsheng Wang Department of Computer Science Tsinghua University Beijing China Beijing National Research Center for Information Science and Technology Tsinghua University Beijing China

The 8-byte granularity of failure-atomicity brings two challenges to Persistent Memory (PM) B+-tree designs. The first is how to insert a key in a sorted node atomically. The second is how to atomically perform structural modification operations that involve multiple nodes, such as node splits and merges. The majority of current designs either have huge consistency cost or compromise recovery time. In this paper, we propose an in-node logging technology, named Lazy-Box, to update multiple parts of a node without exposing intermediate states. Lazy-Box aggregates successive modifications and selectively uses Copy-on-Write (CoW) to reduce consistency cost. Based on Lazy-Box, we propose a new variant of B+-tree on PM, named Side-to-Side B+-Tree (SSB- Tree). SSB- Tree relaxes the tree-structure requirement so that node splits and merges only modify one node. Taking advantage of Lazy-Box, all modification operations of SSB-Tree are committed through a single 8-byte write. Therefore, SSB- Tree not only enables efficient concurrency protocol but also achieves instant recovery. Last but most important, SSB- Tree doubles the node space and reuses the extra space to avoid expensive node allocations when performing CoW. Our experimental results show that SSB-Tree achieves up to 29%, 40%, and 14% higher throughput in Insert, Delete, and Scan benchmark respectively than other state-of-the-art PM B+-trees.

关键词： Concurrent computing distributed processing Protocols Costs Cows Throughput Computer crashes

来源：评论

学校读者我要写书评

暂无评论

Memory Access Granularity Aware Lossless Compression for GPUs

Memory Access Granularity Aware Lossless Compression for GPU...

引用

international symposium on parallel and distributed Processing (IPDPS)

作者： Sohan Lal Manuel Renz Julian Hartmer Ben Juurlink Technische Universit&#x00E4 t Hamburg Germany t Berlin Germany

High-bandwidth off-chip memory has played a key role in the success of Graphics Processing Units (GPUs) as an accelerator. However, as memory bandwidth scaling continues to lag behind the computational power, it remains a key bottleneck in computing systems. While memory compression has shown immense potential to increase the effective memory bandwidth by compressed data transfers between on-chip and off-chip memory, the large memory access granularity (MAG) of off-chip memory limits compression techniques from achieving a high effective compression ratio. Unfortunately, state-of-the-art lossless memory compression techniques do not take the large MAG of off-chip memory into account. A recent study has used MAG-aware approximation to increase the effective compression ratio, however, not all applications can tolerate errors, which limits its applicability. We propose extensions and GPU-specific optimizations to adapt a lossless memory compression technique to a MAG size to increase the effective compression ratio and performance gain. Our technique is based on the well-known Base-Delta-Immediate (BDI) compression technique that compresses a memory block to a common base and multiple deltas. We leverage the key observation that deltas often contain enough leading zeros to compress a block to a multiple of MAG without any loss of information. We show that MAG-aware BDI provides, on average, 48 % higher effective compression ratio, 10% (up to 27%) higher speedup, and 16% bandwidth reduction compared to normal BDI. While BDI, FPC, and CPACK have a similar compression ratio, MAG-aware BDI outperforms FPC, CPACK, and SLC by 56%, 47%, and 33%, respectively.

关键词： distributed processing Graphics processing units Bandwidth Performance gain Data transfer Encoding System-on-chip

来源：评论

学校读者我要写书评

暂无评论

Temporal Data Analytics on COVID-19 Data with Ubiquitous computing 18

Temporal Data Analytics on COVID-19 Data with Ubiquitous Com...

引用

18th IEEE Int Symp on parallel and distributed Proc with Applicat (ISPA) / 10th IEEE Int Conf on Big Data and Cloud Comp (BDCloud) / IEEE Int Symp on Social Comp and Networking (SocialCom) / IEEE Int Conf on Sustainable Comp and Commun (SustainCom)

作者： Chen, Yubo Leung, Carson K. Shang, Siyuan Wen, Qi Univ Manitoba Dept Comp Sci Winnipeg MB Canada

ISBN: (纸本)9781665414852

With technological advancements in computing and communications, huge amounts of big data are generated and collected at a very rapid rate from a wide variety of rich data sources. Embedded in these big data are useful information and valuable knowledge. An example is healthcare and epidemiological data such as data related to patients who suffered from viral diseases like the coronavirus disease 2019 (COVID-19). Knowledge discovered from these epidemiological data via data science helps researchers, epidemiologists and policy makers to get a better understanding of the disease, which may inspire them to come up ways to detect, control and combat the disease. In this paper, we present a temporal data science algorithm for analyzing big COVID-19 epidemiological data, with focus on the temporal data analytics with ubiquitous computing. The algorithm helps users to get a better understanding of information about the confirmed cases of COVID-19. Evaluation results show the benefits of our system in temporal data analytics of big COVID-19 data with ubiquitous computing. Although the algorithm is designed for temporal data analytics of big epidemiological data, it would be applicable to other temporal data analytics of big data in many real-life applications and services.

关键词： data science coronavirus disease COVID-19 big data temporal data data mining ubiquitous computing

来源：评论

学校读者我要写书评

暂无评论

As easy as ABC: Optimal (A)ccountable (B)yzantine (C)onsensus is easy!

As easy as ABC: Optimal (A)ccountable (B)yzantine (C)onsensu...

引用

international symposium on parallel and distributed Processing (IPDPS)

作者： Pierre Civit Seth Gilbert Vincent Gramoli Rachid Guerraoui Jovan Komatovic CNRS Sorbonne University NUS Singapore EPFL University of Sydney &#x00C9 cole Polytechnique Federale de Lausanne (EPFL)

It is known that the agreement property of the Byzantine consensus problem among $n$ processes can be violated in a non-synchronous system if the number of faulty processes exceeds $t_distributed$ = ┌ $n$ /3┐ − 1 [10], [19]. In this paper, we investigate the accountable Byzantine consensus problem in non-synchronous systems: the problem of solving Byzantine consensus whenever possible (e.g., when the number of faulty processes does not exceed $t_distributed$ ) and allowing correct processes to obtain proof of culpability of (at least) $t_distributed+ 1$ faulty processes whenever correct processes disagree. We present four complementary contributions: 1) We introduce ABC: a simple yet efficient transformation of any Byzantine consensus protocol to an accountable one. ABC introduces an overhead of only two all-to-all communication rounds and $O(n^distributed)$ additional bits in executions with up to $t_distributed$ faults (i.e., in the common case). 2) We define the accountability complexity, a complex-ity metric representing the number of accountability-specific messages that correct processes must send. Fur-thermore, we prove a tight lower bound. In particular, we show that any accountable Byzantine consensus protocol incurs cubic accountability complexity. Moreover, we illustrate that the bound is tight by applying the ABC transformation to any Byzantine consensus protocol. 3) We demonstrate that, when applied to an optimal Byzan-tine consensus protocol, ABC constructs an accountable Byzantine consensus protocol that is (1) optimal with respect to the communication complexity in solving consensus whenever consensus is solvable, and (2) op-timal with respect to the accountability complexity in obtaining accountability whenever disagreement occurs. 4) We generalize ABC to other distributed computing prob-lems besides the classic consensus problem. We charac-terize a class of agreement tasks, including reliable and consistent broadcast [5], that ABC renders accountable.

关键词： Measurement distributed processing Lattices Complexity theory Consensus protocol Reliability Task analysis

来源：评论

学校读者我要写书评

暂无评论

Efficient task pruning mechanism to improve robustness of heterogeneous computing systems

引用

JOURNAL OF parallel AND distributed computing 2020年 142卷 46-61页

作者： Denninnart, Chavit Gentry, James Mokhtari, Ali Salehi, Mohsen Amini Univ Louisiana Lafayette Sch Comp & Informat Lafayette LA 70504 USA Spectra Log Boulder CO USA

In heterogeneous distributed computing (HC) systems, diversity can exist in both computational resources and arriving tasks. In an inconsistently heterogeneous computing system, task types have different execution times on heterogeneous machines. A method is required to map arriving tasks to machines based on machine availability and performance, maximizing the number of tasks meeting deadlines (defined as robustness). For tasks with hard deadlines (e.g., those in live video streaming), tasks that miss their deadlines are dropped. The problem investigated in this research is maximizing the robustness of an oversubscribed HC system. A way to maximize this robustness is to prune (i.e., defer or drop) tasks with low probability of meeting their deadlines to increase the probability of other tasks meeting their deadlines. In this paper, we first provide a mathematical model to estimate a task's probability of meeting its deadline in the presence of task dropping. We then investigate methods for engaging probabilistic dropping. We propose methods to dynamically determine task dropping and deferring threshold probabilities. Next, we develop a pruning system and a pruning-aware mapping heuristic, which we extend to engender fairness across various task types. We present the pruning mechanism as an independent component that can be applied to any mapping heuristic to improve the system robustness. To reduce overhead of the pruning mechanism, we propose approximation methods that remarkably reduce the number of mathematical calculations and improve the practicality of deploying the mechanism in heterogeneous or even homogeneous computing systems. We show the cost and energy gains of the pruning mechanism. Simulation results, harnessing a selection of mapping heuristics, show efficacy of the pruning mechanism in improving robustness (on average by similar or equal to 22%) and cost in an oversubscribed HC system by up to similar or equal to 33%. (C) 2020 Elsevier Inc. All righ

关键词： Heterogeneous computing (HC) Probabilistic pruning Mapping heuristic Robustness

来源：评论

学校读者我要写书评

暂无评论

PokéMem: Taming Wild Memory Consumers in Apache Spark

PokéMem: Taming Wild Memory Consumers in Apache Spark

引用

international symposium on parallel and distributed Processing (IPDPS)

作者： Minhyeok Kweun Goeun Kim Byungsoo Oh Seongho Jung Taegeon Um Woo-Yeon Lee Samsung Research Korea Republic of Kakao Corp.

Apache Spark is a widely used in-memory processing system due to its high performance. For fast data processing, Spark manages in-memory data such as cached or shuffling (aggregate and sorting) data in its own managed memory pools. However, despite its sophisticated memory management scheme, we found that Spark still suffers from out-of-memory (OOM) exceptions and high garbage collection (GC) overheads when wild memory consumers, who are not tracked by Spark and execute external codes, use a large amount of memory. To resolve the problems, we propose PokéMem, which is an enhanced Spark that incorporates wild memory consumers into the managed ones to prevent them from taking up memory spaces excessively in stealth. Our main idea is to open the black-box of unmanaged memory regions in external codes by providing customized data collections. PokéMem enables fine-grained controls of created objects within running tasks, by spilling and reloading the objects of custom data collections based on the memory pressure and access patterns. To further reduce memory pressures, PokéMem exploits pre-built memory estimation models to predict the external code's memory usage and proactively acquires memory before the execution of external code, and also performs JVM heap-usage monitoring to avoid critical memory pressures. With the help of these techniques, our evaluations show that PokéMem outperforms vanilla Spark with at most 3× faster execution with 3.9× smaller GC overheads, and successfully runs workloads without OOM exception that vanilla Spark has failed to run.

关键词： Codes Runtime Machine learning algorithms Memory management Cluster computing Predictive models Sparks

来源：评论

学校读者我要写书评

暂无评论

Union: An Automatic Workload Manager for Accelerating Network Simulation 34

Union: An Automatic Workload Manager for Accelerating Networ...

引用

34th IEEE international parallel and distributed Processing symposium (IPDPS)

作者： Wang, Xin Mubarak, Misbah Kang, Yao Ross, Robert B. Lan, Zhiling IIT Dept Comp Sci Chicago IL 60616 USA Argonne Natl Lab Math & Comp Sci Div 9700 S Cass Ave Argonne IL 60439 USA Amazon Web Serv Seattle WA USA

ISBN: (数字)9781728168760

ISBN: (纸本)9781728168760

With the rapid growth of the machine learning applications, the workloads of future HPC systems are anticipated to be a mix of scientific simulation, big data analytics, and machine learning applications. Simulation is a great research vehicle to understand the performance implications of co-running scientific applications with big data and machine learning workloads on large-scale systems. In this paper, we present Union, a workload manager that provides an automatic framework to facilitate hybrid workload simulation in CODES. Furthermore, we use Union, along with CODES, to investigate various hybrid workloads composed of traditional simulation applications and emerging learning applications on two dragonfly systems. The experiment results show that both message latency and communication time are important performance metrics to evaluate network interference. Network interference on HPC applications is more reflected by the message latency variation, whereas ML application performance depends more on the communication time.

关键词： High-performance computing interference heterogeneous workloads

来源：评论

学校读者我要写书评

暂无评论

Secure acceleration on cloud-based FPGAs - FPGA enclaves 34

Secure acceleration on cloud-based FPGAs - FPGA enclaves

引用

34th IEEE international parallel and distributed Processing symposium (IPDPS)

作者： Englund, Hakan Lindskog, Niklas Ericsson Research Lund Sweden

ISBN: (纸本)9781728174457

FPGAs are becoming a common sight in cloud environments and new usage paradigms, such as FPGA-as-a-Service, have emerged. This development poses a challenge to traditional FPGA security models, as these are assuming trust between the user and the hardware owner. Currently, the user cannot keep bitstream nor data protected from the hardware owner in an FPGA-as-a-service setting. This paper proposes a security model where the chip manufacturer takes the role of root-of-trust to remedy these security problems. We suggest that the chip manufacturer creates a Public Key Infrastructure (PKI), used for user bitstream protection and data encryption, on each device. The chip manufacturer, rather than the hardware owner, also controls certain security-related peripherals. This allows the user to take control over a predefined part of the programmable logic and set up a protected enclave area. Hence, all user data can be provided in encrypted form and only be revealed inside the enclave area. In addition, our model enables secure and concurrent multi-tenant usage of remote FPGAs. To also consider the needs of the hardware owner, our solution includes bitstream certification and affirming that uploaded bitstreams have been vetted against maliciousness.

关键词： Cloud security Confidential computing Enclaves FPGA Hardware security System-on-chip

来源：评论

学校读者我要写书评

暂无评论

Caching Resource Allocation and Mobility Aware in Information-Centric Networking with Mobile Edge computing 18

Caching Resource Allocation and Mobility Aware in Informatio...

引用

作者： Khan, Daud Haopeng, Chen Din, Najamul Shanghai Jiao Tong Univ Sch Elect Informat & Elect Engn Shanghai Peoples R China

ISBN: (纸本)9781665414852

The rapid rise in communication delay and the speed of downloading Wireless Communication devices with Fifth Generation Wireless Networks (5-G) wireless networks, mobile edge servers may use the high-speed local link to reduce the demand of their core networks by caching or prefetching content. The Information-Centric Networking (ICN) with Mobile Edge computing (MEC) aims to upgrade its system to overcome the lack of network capacity. The aim is to provide a network service that is more suitable today (particularly resource allocation and mobility) and resilient to interruptions and failures. Cache resource allocation and mobility is one of the problematic and realistically major issues of Information-Centric-Network and in Mobile Edge computing. Though several studies have been optimized. A minimum consideration has been devoted to minimizing user energy consumption to improve caching efficiency and service quality. To address this deficiency, the focus of this study is optimized cache resource allocation for closer users and connected networks with minimum time delay and energy consumption considering mobility. For this reason, the problem of caching resource allocation (CRA) is divided into two subproblems: the problem of Base Station caching ability (BSCC) and the challenge of Request MAtching (RMAT). Then we can one by one solve BSCC and RMAT and merge their CRA solutions.

关键词： Mobile Edge computing Resource Allocation Real-Time Caching and Mobility

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：