Split (1)

train · 127k rows

text large_stringlengths 7 3.02k
Considering the diverse nature of real-world distributed applications that makes it hard to identify a representative subset of distributed benchmarks, we focus on their underlying distributed algorithms. We present and characterize a new kernel benchmark suite (named IMSuite) that simulates some of the classical dist...
Irregular memory access patterns pose performance and user productivity challenges on distributed-memory systems. They can lead to fine-grained remote communication and the data access patterns are often not known until runtime.
We describe a case study that uses grid computing techniques to support the collaborative learning of high school students investigating cosmic rays. Students gather and upload science data to our e-Lab portal.
We consider the fundamental problem of assigning distinct labels to agents in the probabilistic model of population protocols. Our protocols operate under the assumption that the size $n$ of the population is embedded in the transition function.
Given the proximity of many wireless users and their diversity in consuming local resources (e.g., data-plans, computation and even energy resources), device-to-device (D2D) resource sharing is a promising approach towards realizing a sharing economy. In the resulting networked economy, $n$ users segment themselves in...
Autoscaling is a hallmark of cloud computing as it allows flexible just-in-time allocation and release of computational resources in response to dynamic and often unpredictable workloads. This is especially important for web applications whose workload is time dependent and prone to flash crowds.
Data intensive applications often involve the analysis of large datasets that require large amounts of compute and storage resources. While dedicated compute and/or storage farms offer good task/data throughput, they suffer low resource utilization problem under varying workloads conditions.
In this paper, we investigate a distributed maximal independent set (MIS) reconfiguration problem, in which there are two maximal independent sets for which every node is given its membership status, and the nodes need to communicate with their neighbors in order to find a reconfiguration schedule that switches from th...
Internet-scale web applications are becoming increasingly storage-intensive and rely heavily on in-memory object caching to attain required I/O performance. We argue that the emerging serverless computing paradigm provides a well-suited, cost-effective platform for object caching.
This deliverable reports the results of white-box methodologies and early results of the first prototype of libraries and programming abstractions as available by project month 18 by Work Package 2 (WP2). It reports
In mobile database environments, multiple users may access similar data items irrespective of their physical location leading to concurrent access anomalies. As disconnections and mobility are the common characteristics in mobile environment, performing concurrent access to a particular data item leads to inconsistenc...
Low Earth Orbit (LEO) satellite constellations have seen significant growth and functional enhancement in recent years, which integrates various capabilities like communication, navigation, and remote sensing. However, the heterogeneity of data collected by different satellites and the problems of efficient inter-sate...
Cloud computing providers have setup several data centers at different geographical locations over the Internet in order to optimally serve needs of their customers around the world. However, existing systems do not support mechanisms and policies for dynamically coordinating load distribution among different Cloud-ba...
Iterative methods such as Jacobi, Gauss-Seidel, and Successive Over-Relaxation (SOR) are fundamental tools in solving large systems of linear equations across various scientific fields, particularly in the field of data science which has become increasingly relevant in the past decade. Iterative methods' use of ma...
This paper addresses the mutual visibility problem for a set of semi-synchronous, opaque robots occupying distinct positions in the Euclidean plane. Since robots are opaque, if three robots lie on a line, the middle robot obstructs the visions of the two other robots.
The rapid growth of data size and accessibility in recent years has instigated a shift of philosophy in algorithm design for artificial intelligence. Instead of engineering algorithms by hand, the ability to learn composable systems automatically from massive amounts of data has led to ground-breaking performance in i...
Complex scientific experiments from various domains are typically modeled as workflows and executed on large-scale machines using a Parallel Workflow Management System (WMS). Since such executions usually last for hours or days, some WMSs provide user steering support, i.e., they allow users to run data analyses and, ...
Active replication following the state machine replication (SMR) approach is a way to make existing systems and services more reliable and fault-tolerant. The additional communication overhead has a negative impact on the system's throughput and overall request latency.
Cloud computing is a model for enabling convenient, on-demand network access to a shared pool of configurable computing resources. To provide cloud computing services economically, it is important to optimize resource allocation under the assumption that the required resource can be taken from a shared resource pool.
This paper reports on an in-depth evaluation of the performance portability frameworks Kokkos and RAJA with respect to their suitability for the implementation of complex particle-in-cell (PIC) simulation codes, extending previous studies based on codes from other domains. At the example of a particle-in-cell model, w...
Sparse matricized tensor times Khatri-Rao product (MTTKRP) is one of the most computationally expensive kernels in sparse tensor computations. This work focuses on optimizing the MTTKRP operation on GPUs, addressing both performance and storage requirements.
Replicated services accessed via {\em quorums} enable each access to be performed at only a subset (quorum) of the servers, and achieve consistency across accesses by requiring any two quorums to intersect. Recently, $b$-masking quorum systems, whose intersections contain at least $2b+1$ servers, have been proposed to...
This paper is about partitioning in parallel and distributed simulation. That means decomposing the simulation model into a numberof components and to properly allocate them on the execution units.
Existing general purpose frameworks for gigantic model training, i.e., dense models with billions of parameters, cannot scale efficiently on cloud environment with various networking conditions due to large communication overheads. In this paper, we propose MiCS, which Minimizes the Communication Scale to bring down c...
Distributed real-time and embedded (DRE) systems executing mixed criticality task sets are increasingly being deployed in mobile and embedded cloud computing platforms, including space applications. These DRE systems must not only operate over a range of temporal and spatial scales, but also require stringent assuranc...
State-of-the-art simulations of detailed neural models follow the Bulk Synchronous Parallel execution model. Execution is divided in equidistant communication intervals, equivalent to the shortest synaptic delay in the network.
There are existing standard solvers for tackling discrete optimization problems. However, in practice, it is uncommon to apply them directly to the large input space typical of this class of problems.
In this paper, we present a novel parallel dimension-independent node positioning algorithm that is capable of generating nodes with variable density, suitable for meshless numerical analysis. A very efficient sequential algorithm based on Poisson disc sampling is parallelized for use on shared-memory computers, such ...
With the increasing popularity of Internet-based services and applications, power efficiency is becoming a major concern for data center operators, as high electricity consumption not only increases greenhouse gas emissions, but also increases the cost of running the server farm itself. In this paper we address the pr...
Process anonymity has been studied for a long time. Memory anonymity is more recent.
As an alternative to current wired-based networks, wireless sensor networks (WSNs) are becoming an increasingly compelling platform for engineering structural health monitoring (SHM) due to relatively low-cost, easy installation, and so forth. However, there is still an unaddressed challenge: the application-specific ...
A distributed algorithm performs local computations on pieces of input and communicates the results through given communication links. When processing a massive graph in a distributed algorithm, local outputs must be configured as a solution to a graph problem without shared memory and with few rounds of communication...
Distributed Stream Processing (DSP) focuses on the near real-time processing of large streams of unbounded data. To increase processing capacities, DSP systems are able to dynamically scale across a cluster of commodity nodes, ensuring a good Quality of Service despite variable workloads.
Naor, Parter, and Yogev [SODA 2020] recently designed a compiler for automatically translating standard centralized interactive protocols to distributed interactive protocols, as introduced by Kol, Oshman, and Saxena [PODC 2018]. In particular, by using this compiler, every linear-time algorithm for deciding the membe...
This paper presents the first hardware implementation of bittide, a decentralized clock synchronization mechanism for achieving logical synchrony in distributed systems. We detail the design and implementation of an 8-node bittide network using off-the-shelf FPGA boards and adjustable clock sources.
Deep neural networks (DNNs) have been shown to outperform conventional machine learning algorithms across a wide range of applications, e.g., image recognition, object detection, robotics, and natural language processing. However, the high computational complexity of DNNs often necessitates extremely fast and efficien...
The interconnection network comprises a significant portion of the cost of large parallel computers, both in economic terms and power consumption. Several previous proposals exploit large-radix routers to build scalable low-distance topologies with the aim of minimizing these costs.
In serial batch (s-batch) scheduling, jobs are grouped in batches and processed sequentially within their batch. This paper considers multiple parallel machines, nonidentical job weights and release times, and sequence-dependent setup times between batches of different families.
OpenFMO framework, an open-source software (OSS) platform for Fragment Molecular Orbital (FMO) method, is extended to multi-physics simulations (MPS). After reviewing the several FMO implementations on distributed computer environments, the subsequent development planning corresponding to MPS is presented.
We consider blockchain in dynamic networks. We define the Blockchain Decision Problem.
We present GSPMD, an automatic, compiler-based parallelization system for common machine learning computations. It allows users to write programs in the same way as for a single device, then give hints through a few annotations on how to distribute tensors, based on which GSPMD will parallelize the computation.
Autoregressive models, despite their commendable performance in a myriad of generative tasks, face challenges stemming from their inherently sequential structure. Inference on these models, by design, harnesses a temporal dependency, where the current token's probability distribution is conditioned on preceding to...
Broadcast networks are often used in modern communication systems. A common broadcast network is a single hop shared media system, where a transmitted message is heard by all neighbors, such as some LAN networks.
We propose a framework for deployment and subsequent autonomic management of component-based distributed applications. An initial deployment goal is specified using a declarative constraint language, expressing constraints over aspects such as component-host mappings and component interconnection topology.
The disjoint set union problem is a basic problem in data structures with a wide variety of applications. We extend a known efficient sequential algorithm for this problem to obtain a simple and efficient concurrent wait-free algorithm running on an asynchronous parallel random access machine (APRAM).
The exponential growth in use of large deep neural networks has accelerated the need for training these deep neural networks in hours or even minutes. This can only be achieved through scalable and efficient distributed training, since a single node/card cannot satisfy the compute, memory, and I/O requirements of toda...
We present the ParaPlan tool which provides the reachability analysis of planar hybrid systems defined by differential inclusions (SPDI). It uses the parallelized and optimized version of the algorithm underlying the SPeeDI tool.
Blockchain-based federated learning has gained significant interest over the last few years with the increasing concern for data privacy, advances in machine learning, and blockchain innovation. However, gaps in security and scalability hinder the development of real-world applications.
As the cloud computing paradigm has gained prominence, the need for verifiable computation has grown increasingly urgent. The concept of verifiable computation enables a weak client to outsource difficult computations to a powerful, but untrusted, server.
In this work we detail a novel open source library, called MMLSpark, that combines the flexible deep learning library Cognitive Toolkit, with the distributed computing framework Apache Spark. To achieve this, we have contributed Java Language bindings to the Cognitive Toolkit, and added several new components to the S...
Particle Swarm Optimization (PSO) is a stochastic technique for solving the optimization problem. Attempts have been made to shorten the computation times of PSO based algorithms with massive threads on GPUs (graphic processing units), where thread groups are formed to calculate the information of particles and the co...
A fundamental ambition of grid and distributed systems is to be capable of sustaining evolution and allowing for adaptability ((F. Losavio et al., 2002), (S. Radhakrishnan, 2005)). Furthermore, as the complexity and sophistication of theses structures increases, so does the need for adaptability of each component.
Efficient runtime task scheduling on complex memory hierarchy becomes increasingly important as modern and future High-Performance Computing (HPC) systems are progressively composed of multisocket and multi-chiplet nodes with nonuniform memory access latencies. Existing locality-aware scheduling schemes either require...
Various kinds of applications take advantage of GPUs through automation tools that attempt to automatically exploit the available performance of the GPU's parallel architecture. Directive-based programming models, such as OpenACC, are one such method that easily enables parallel computing by just adhering code ann...
Partitioning and deploying Deep Neural Networks (DNNs) across edge nodes may be used to meet performance objectives of applications. However, the failure of a single node may result in cascading failures that will adversely impact the delivery of the service and will result in failure to meet specific objectives.
Co-locating latency-critical services (LCSs) and best-effort jobs (BEJs) constitute the principal approach for enhancing resource utilization in production. Nevertheless, the co-location practice hurts the performance of LCSs due to resource competition, even when employing isolation technology.
As emerging deep neural network (DNN) models continue to grow in size, using large GPU clusters to train DNNs is becoming an essential requirement to achieving acceptable training times. In this paper, we consider the case where future increases in cluster size will cause the global batch size that can be used to trai...
In a virtualized environment, contextualization is the process of configuring a VM instance for the needs of various deployment use cases. Contextualization in CernVM can be done by passing a handwritten context to the user data field of cloud APIs, when running CernVM on the cloud, or by using CernVM web interface wh...
A survey is given of approaches to the problem of distributed consensus, focusing particularly on methods based on cellular automata and related systems. A variety of new results are given, as well as a history of the field and an extensive bibliography.
Sparse matrix-vector multiplication (SpMV) operations are commonly used in various scientific applications. The performance of the SpMV operation often depends on exploiting regularity patterns in the matrix.
We present an efficient approach for writing architecture-agnostic parallel high-performance stencil computations in Julia, which is instantiated in the package <a href="http://ParallelStencil.jl" rel="external noopener nofollow" class="link-external link-http">this http URL</a>. Powerful metaprogramming, costless abs...
The well-known randomized consensus algorithm by Aspnes and Herlihy for asynchronous shared-memory systems was proved to work, even against a strong adversary, under the assumption that the registers that it uses are atomic registers. With atomic registers, every read or write operation is instantaneous (and thus indi...
Companies are rushing to deliver their services and solutions through the cloud. The scheduling process is very critical in reducing delays.
Neural Radiance Fields (NeRF) is an emerging technique to synthesize 3D objects from 2D images with a wide range of potential applications. However, rendering existing NeRF models is extremely computation intensive, making it challenging to support real-time interaction on mobile devices.
To achieve scalability with today's heterogeneous HPC resources, we need a dramatic shift in our thinking; MPI+X is not enough. Asynchronous Many Task (AMT) runtime systems break down the global barriers imposed by the Bulk Synchronous Programming model.
To enable the pre-trained models to be fine-tuned with local data on edge devices without sharing data with the cloud, we design an efficient split fine-tuning (SFT) framework for edge and cloud collaborative learning. We propose three novel techniques in this framework.
Heterogeneous multi-core systems such as big/little architectures have been introduced as an attractive server design option with the potential to improve performance under power constraints in data centres. Since both big high-performing and little power-efficient cores can run on the same system sharing the workload...
Clustering of image is one of the important steps of mining satellite images. In our experiment we have simultaneously run multiple K-means algorithms with different initial centroids and values of k in the same iteration of MapReduce jobs.
We present new, simple, fully distributed, practical algorithms with linear time communication cost for irregular gather and scatter operations in which processors contribute or consume possibly different amounts of data. In a linear cost transmission model with start-up latency $\alpha$ and cost per unit $\beta$, the...
The most essential component of every Distributed Ledger Technology (DLT) is the Consensus Algorithm (CA), which enables users to reach a consensus in a decentralized and distributed manner. Numerous CA exist, but their viability for particular applications varies, making their trade-offs a crucial factor to consider ...
Federated Learning (FL) emerges as a distributed machine learning paradigm without end-user data transmission, effectively avoiding privacy leakage. Participating devices in FL are usually bandwidth-constrained, and the uplink is much slower than the downlink in wireless networks, which causes a severe uplink communic...
Message brokers often mediate communication between data producers and consumers by adding variable-sized messages to ordered distributed queues. Our goal is to determine the number of consumers and consumer-partition assignments needed to ensure that the rate of data consumption keeps up with the rate of data product...
Popular blockchains such as Ethereum and several others execute complex transactions in blocks through user-defined scripts known as smart contracts. Serial execution of smart contract transactions/atomic-units (AUs) fails to harness the multiprocessing power offered by the prevalence of multi-core processors.
This paper investigates the design of parallel general matrix multiplication (GEMM) for a Versal Adaptive Compute Accelerated Platform (ACAP) equipped with a VC1902 system-on-chip and multiple Artificial Intelligence Engines (AIEs). Our efforts aim to port standard optimization techniques applied in the high-performan...
The proliferation of agentic Large Language Models (LLMs) on personal devices introduces a new class of workloads characterized by a dichotomy of objectives. Reactive tasks, initiated by users, demand immediate, low-latency responses, while proactive tasks operate invisibly and prioritize throughput.
Do Linux distribution package managers need the privileged operations they request to actually happen? Apparently not, at least for building container images for HPC applications.
Reliability in cloud AI infrastructure is crucial for cloud service providers, prompting the widespread use of hardware redundancies. However, these redundancies can inadvertently lead to hidden degradation, so called "gray failure", for AI workloads, significantly affecting end-to-end performance and conceali...
The rapidly growing popularity and scale of data-parallel workloads demand a corresponding increase in raw computational power of GPUs (Graphics Processing Units). As single-GPU systems struggle to satisfy the performance demands, multi-GPU systems have begun to dominate the high-performance computing world.
This paper studies the difference in computational power between the mesh-connected parallel computers equipped with dynamically reconfigurable bus systems and those with static ones. The mesh with separable buses (MSB) is the mesh-connected parallel computer with dynamically reconfigurable row/column buses.
State-of-the-art language and vision models are routinely trained across thousands of GPUs, often spanning multiple data-centers, yet today's distributed frameworks still assume reliable connections (e.g., InfiniBand or RoCE). The resulting acknowledgment traffic and retransmissions inflate tail latencies and limi...
This article presents a theoretical investigation of computation beyond the Turing barrier from emergent behavior in distributed systems. In particular, we present an algorithmic network that is a mathematical model of a networked population of randomly generated computable systems with a fixed communication protocol....
Communication primitives play a central role in modern computing. They offer a panel of reliability and ordering guarantees for messages, enabling the implementation of complex distributed interactions.
Wireless communication networks are becoming highly virtualized with two-layer hierarchies, in which controllers at the upper layer with tasks to achieve can ask a large number of agents at the lower layer to help realize computation, storage, and transmission functions. Through offloading data processing to the agent...
We consider the External Clock Synchronization problem in dynamic sensor networks. Initially, sensors obtain inaccurate estimations of an external time reference and subsequently collaborate in order to synchronize their internal clocks with the external time.
In this paper, we introduce DLB, a Deep Learning based load Balancing mechanism, to effectively address the data skew problem. The key idea of DLB is to replace hash functions in the load balancing mechanisms with deep learning models, which are trained to be able to map different distributions of workloads and data t...
Fat-tree networks have been widely adopted to High Performance Computing (HPC) clusters and to Data Center Networks (DCN). These parallel systems usually have a large number of servers and hosts, which generate large volumes of highly-volatile traffic.
Everyday a huge amount of data is generated in Cloud Computing. The maintenance of this electronic data needs some extremely efficient services.
There exists a plethora of consensus protocols in literature. The reason is that there is no one-size-fits-all solution, since every protocol is unique and its performance is directly tied to the deployment settings and workload configurations.
Big data analytics requires high programmer productivity and high performance simultaneously on large-scale clusters. However, current big data analytics frameworks (e.g. Apache Spark) have prohibitive runtime overheads since they are library-based.
Energy consumption is increasingly becoming a limiting factor to the design of faster large-scale parallel systems, and development of energy-efficient and energy-aware applications is today a relevant issue for HPC code-developer communities. In this work we focus on energy performance of the Knights Landing (KNL) Xe...
In this short paper, we present an innovative approach to limit the required bandwidth when transferring data during in transit analysis. This approach is called hybrid because it combines existing in situ and in transit solutions.
We develop a novel parallel decomposition strategy for unweighted, undirected graphs, based on growing disjoint connected clusters from batches of centers progressively selected from yet uncovered nodes. With respect to similar previous decompositions, our strategy exercises a tighter control on both the number of clu...
Deep neural network (DNN) inference has become an important part of many data-center workloads. This has prompted focused efforts to design ever-faster deep learning accelerators such as GPUs and TPUs.
The Internet stack is not a complete description of the resources and services needed to implement distributed applications, as it only accounts for communication services and the protocols that are defined to deliver them. This paper presents an account of the current distributed application architecture using a form...
Routing is a challenging problem for wireless ad hoc networks, especially when the nodes are mobile and spread so widely that in most cases multiple hops are needed to route a message from one node to another. In fact, it is known that any online routing protocol has a poor performance in the worst case, in a sense th...
The ATLAS Event Streaming Service (ESS) at the LHC is an approach to preprocess and deliver data for Event Service (ES) that has implemented a fine-grained approach for ATLAS event processing. The ESS allows one to asynchronously deliver only the input events required by ES processing, with the aim to decrease data tr...
This short papers discusses the issues of teaching cloud computing from a software engineering rather than a business perspective. It discusses what topics might be covered in a senior course on cloud software engineering.
The problem of MPI programs execution time prediction on a certain set of computer installations is considered. This problem emerges with orchestration and provisioning a virtual infrastructure in a cloud computing environment over a heterogeneous network of computer installations: supercomputers or clusters of server...
This paper investigates an open problem introduced in [14]. Two or more mobile agents start from different nodes of a network and have to accomplish the task of gathering which consists in getting all together at the same node at the same time.
The Metaverse is a virtual environment where users are represented by avatars to navigate a virtual world, which has strong links with the physical one. State-of-the-art Metaverse architectures rely on a cloud-based approach for avatar physics emulation and graphics rendering computation.

Subsets and Splits

No community queries yet

The top public SQL queries from the community will appear here once available.