-
1
المؤلفون: Johannes Pietrzyk, Patrick Damme, Wolfgang Lehner, Dirk Habich, Erich Focht
المصدر: Datenbank-Spektrum. 19:183-197
مصطلحات موضوعية: Coprocessor, Computer science, 05 social sciences, Vectorization, NEC SX-Aurora TSUBASA, Column stores, Experimental evaluation, SIMD extension, Vektorisierung, NEC SX-Aurora TSUBASA, Säulenspeicher, Experimentelle Auswertung, SIMD-Erweiterung, Joins, 02 engineering and technology, Parallel computing, Supercomputer, In-memory database, Parallel processing (DSP implementation), 020204 information systems, Vectorization (mathematics), 0202 electrical engineering, electronic engineering, information engineering, SIMD, ddc:004, 0509 other social sciences, 050904 information & library sciences, Data compression
الوصف: In-memory column-store database systems are state of the art for the efficient processing of analytical workloads. In these systems, data compression as well as vectorization play an important role. Currently, the vectorized processing is done using regular SIMD (Single Instruction Multiple Data) extensions of modern processors. For example, Intel’s latest SIMD extension supports 512-bit vector registers which allows the parallel processing of 8× 64-bit values. From a database system perspective, this vectorization technique is not only very interesting for compression and decompression to reduce the computational overhead, but also for all database operators like joins, scan, as well as groupings. In contrast to these SIMD extensions, NEC Corporation has recently introduced a novel pure vector engine (supercomputer) as a co-processor called SX-Aurora TSUBASA. This vector engine features a vector length of 16.384 bits with the world’s highest bandwidth of up to 1.2 TB/s, which perfectly fits to data-intensive applications like in-memory database systems. Therefore, we describe the unique architecture and properties of this novel vector engine in this paper. Moreover, we present selected in-memory column-store-specific evaluation results to show the benefits of this vector engine compared to regular SIMD extensions. Finally, we conclude the paper with an outlook on our ongoing research activities in this direction.
URL الوصول: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::dc313bef924be1ea3d530f508ebe10b9
https://doi.org/10.1007/s13222-019-00323-w -
2
المؤلفون: Florentin Dörre, Alexander Krause, Martin Junghanns, Dirk Habich
المصدر: GRADES-NDA@SIGMOD
مصطلحات موضوعية: Graph analytics, Java, Computer science, Programming language, 010103 numerical & computational mathematics, 02 engineering and technology, computer.software_genre, 01 natural sciences, Matrix (mathematics), 020204 information systems, Linear algebra, 0202 electrical engineering, electronic engineering, information engineering, Graph computation, Graph algorithms, 0101 mathematics, computer, Implementation, Java Programming Language, computer.programming_language
الوصف: Analyzing connected data in forms of graphs is more relevant than ever. To allow users to write their own custom graph algorithms, graph computation models such as GraphBLAS have been developed. Unfortunately, the popular Java programming language was mostly neglected by existing GraphBLAS implementations so far. To overcome that issue, we present our implementation of essential GraphBLAS concepts in the Java programming language in this paper. For our purpose, we extended the linear algebra library Efficient Java Matrix Library (EJML). To show the benefits of our implementation, we compare us against existing graph algorithm libraries in Java using real world graphs and three graph algorithms.
URL الوصول: https://explore.openaire.eu/search/publication?articleId=doi_________::3d4e1ea2db14a8640caec6c34ee8e68c
https://doi.org/10.1145/3461837.3464627 -
3
المؤلفون: Wolfgang Lehner, Dirk Habich, Annett Ungethüm, Johannes Fett
المصدر: DaMoN
مصطلحات موضوعية: Single instruction, multiple threads, Computer science, 02 engineering and technology, Parallel computing, ComputerSystemsOrganization_PROCESSORARCHITECTURES, Column (database), Porting, 020202 computer hardware & architecture, Set (abstract data type), 020204 information systems, Vectorization (mathematics), 0202 electrical engineering, electronic engineering, information engineering, SIMD, State (computer science), Execution model
الوصف: Data-level parallelism (DLP) is a heavily used hardware-driven parallelization technique to optimize the analytical query processing, especially in in-memory column stores. This kind of parallelism is characterized by executing essentially the same operation on different data elements simultaneously. Besides Single Instruction Multiple Data (SIMD) extensions on common x86-processors, GPUs also provide DLP but with a different execution model called Single Instruction Multiple Threads (SIMT), where multiple scalar threads are executed in a SIMD manner. Unfortunately, a complete GPU-specific implementation of all query operators has to be set up, since the state of the vectorized implementations cannot be ported from x86-processors to GPUs right now. To avoid this implementation effort, we present our vision to virtualize GPUs as virtual vector engines with software-defined SIMD instructions and to specialize hardware-oblivious vectorized operators to GPUs using our Template Vector Library (TVL) in this paper.
URL الوصول: https://explore.openaire.eu/search/publication?articleId=doi_________::7841c4b21c37254c4d78ac5b589d6d92
https://doi.org/10.1145/3465998.3466015 -
4
المصدر: ICDE Workshops
مصطلحات موضوعية: Coprocessor, Computer science, Memory bandwidth, 02 engineering and technology, 020202 computer hardware & architecture, Vector processor, Abstraction layer, Parallel processing (DSP implementation), Computer engineering, 020204 information systems, Vectorization (mathematics), 0202 electrical engineering, electronic engineering, information engineering, SIMD, Abstraction (linguistics)
الوصف: NEC Corporation offers a vector engine as a specialized co-processor having two unique features. On the one hand, it operates on vector registers multiple times wider than those of recent mainstream x86-processors. On the other hand, this accelerator provides a memory bandwidth of up to 1.2TB/s for 48GB of main memory. Both features are interesting for analytical query processing: First, vectorization based on the Single Instruction Multiple Data (SIMD) paradigm is a state-of-the-art technique to improve the query performance on x86-processors. Thus, for this accelerator we are able to use the same programming, processing, and optimization concepts as for the host x86-processor. Second, this vector engine is an optimal platform for investigating the efficient vector processing on wide vector registers. To achieve that, we describe an approach to master this co-processor for analytical query processing using a column-store specific abstraction layer for vectorization in this paper. We also detail on selected evaluation results to show the benefits and shortcomings of our approach as well as of the coprocessor compared to x86-processors. We conclude the paper with a discussion on interesting future research activities.
URL الوصول: https://explore.openaire.eu/search/publication?articleId=doi_________::140015657a54e209578b316a87acdcfc
https://doi.org/10.1109/icdew53142.2021.00018 -
5
المؤلفون: Mikhail Zarubin, Thomas Kissinger, Wolfgang Lehner, Dirk Habich, Thomas Willhalm
المصدر: The VLDB Journal. 29:775-795
مصطلحات موضوعية: Emulation, Computer science, Distributed computing, Node (networking), 020206 networking & telecommunications, 02 engineering and technology, Data structure, Replication (computing), Hardware and Architecture, High availability, 0202 electrical engineering, electronic engineering, information engineering, Overhead (computing), 020201 artificial intelligence & image processing, Non-volatile random-access memory, Latency (engineering), Information Systems
الوصف: The long-awaited nonvolatile random-access memory technology NVRAM is finally publicly available on the market and requires significant changes to the architecture of in-memory database systems. Since such hybrid DRAM–NVRAM database systems may be able to keep the primary data solely persistent in the NVRAM, efficient replication mechanisms need to be considered to prevent base data losses and to guarantee high availability in case of various persistent memory failures. In this article, we argue for a software-based replication approach and present compute node-local mechanisms to provide the building blocks—generally available for most platforms—for an efficient NVRAM replication with a low latency and minimal throughput penalty. Within our evaluation, based on both real NVRAM hardware and DRAM-backed emulation, we measured up to 10$$\times $$ less overhead for our optimized replication mechanisms compared to the basic replication mechanism of the Intel persistent memory development kit PMDK. Finally, we present a lightweight switching approach for enabling the adaptive online selection of the best replication mechanism for a given situation.
-
6
المؤلفون: Genia Kostka, Sabrina Habich-Sobiegalla, Niklas Anzinger
المصدر: Journal of Cleaner Production. 205:188-200
مصطلحات موضوعية: Consumption (economics), business.product_category, Social network, Public economics, Renewable Energy, Sustainability and the Environment, business.industry, 020209 energy, Strategy and Management, Public policy, 02 engineering and technology, 010501 environmental sciences, 01 natural sciences, Industrial and Manufacturing Engineering, Purchasing, Sustainable transport, Greenhouse gas, Electric vehicle, 0202 electrical engineering, electronic engineering, information engineering, China, business, 0105 earth and related environmental sciences, General Environmental Science
الوصف: The adoption of electric vehicles is key to lowering the consumption of fossil fuels and emission of greenhouse gases. Cross-national surveys studying citizens' purchase intentions regarding electric vehicles (EVs) remain limited, especially when it comes to combining individual micro-level factors and contextual macro-level forces. Based on a cross-national dataset with 2806 respondents from China (n = 1078), Brazil (n = 929), and Russia (n = 799), this study analyzes variations and determinants of purchase intentions for EVs in these three countries. The survey results indicate that purchase intentions for EVs among Chinese citizens is higher than amongst Brazilian and Russian citizens. The purchasing intention of citizens in all three countries is especially high for people who have a wide social network, and if they already know somebody with an EV. Other macro-level factors, including pollution and charging infrastructure, only impact on purchasing intention in Brazil, while government policy initiatives for EVs seem to have limited effects in all three countries. Micro-level factors, such as age and education, do not have any statistically significant effect in Russia and Brazil, and only a weak effect in China. Based on these results, we provide recommendations for business and policy makers who need to anticipate citizens' demand for EVs and design policies suitable to accelerate the adoption of sustainable transport solutions.
-
7
المؤلفون: Johannes Pietrzyk, Wolfgang Lehner, Dirk Habich
المصدر: DaMoN
مصطلحات موضوعية: Computer science, Distributed computing, Materialized view, Process (computing), 020207 software engineering, 02 engineering and technology, Query optimization, Set (abstract data type), 020204 information systems, Vectorization (mathematics), 0202 electrical engineering, electronic engineering, information engineering, Redundancy (engineering), SIMD, Line (text file)
الوصف: Query execution techniques constantly adapt to novel hardware features to achieve high query performance, in particular for analytical queries. In recent years, vectorization based on the Single Instruction Multiple Data (SIMD) parallel paradigm has been established as a state-of-the-art approach to increase single-query performance. However, since concurrent analytical queries are executed independently potentially invoking a set of fully vectorized operators, the same data accesses and computations among different queries may be executed redundantly. Various techniques have already been proposed to avoid such redundancy, ranging from concurrent scans via the construction of materialized views to applying multiple query optimization techniques. Continuing this line of research, we now investigate the opportunity of sharing vector registers for concurrently running queries in analytical scenarios. In particular, our core sharing approach is to process data elements of different queries together within a single vector register. As we are going to show, sharing vector registers to optimize the execution of concurrent queries can be very beneficial in many cases. We therefore demonstrate the feasibility of a new work sharing strategy and thus open up a wide spectrum of future research opportunities.
URL الوصول: https://explore.openaire.eu/search/publication?articleId=doi_________::18e46381457a9adfa5815c6b8687969d
https://doi.org/10.1145/3399666.3399923 -
8
المؤلفون: Wolfgang Lehner, Dirk Habich, Annett Ungethüm, Patrick Damme, Alexander Krause, Johannes Pietrzyk
مصطلحات موضوعية: FOS: Computer and information sciences, Computer science, General Engineering, Process (computing), Databases (cs.DB), 020207 software engineering, 02 engineering and technology, Data_CODINGANDINFORMATIONTHEORY, computer.software_genre, Base (topology), Operator (computer programming), Computer Science - Databases, 020204 information systems, Compression (functional analysis), 0202 electrical engineering, electronic engineering, information engineering, Memory footprint, Data mining, computer, Integer (computer science), Data compression
الوصف: In this paper, we present MorphStore, an open-source in-memory columnar analytical query engine with a novel holistic compression-enabled processing model. Basically, compression using lightweight integer compression algorithms already plays an important role in existing in-memory column-store database systems, but mainly for base data. In particular, during query processing, these systems only keep the data compressed until an operator cannot process the compressed data directly, whereupon the data is decompressed, but not recompressed. Thus, the full potential of compression during query processing is not exploited. To overcome that, we developed a novel compression-enabled processing model as presented in this paper. As we are going to show, the continuous usage of compression for all base data and all intermediates is very beneficial to reduce the overall memory footprint as well as to improve the query performance.
Submitted to PVLDBURL الوصول: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::228e0653ae80064812346c834d40d7b7
http://arxiv.org/abs/2004.09350 -
9
المؤلفون: Wolfgang Lehner, Dirk Habich, Alexander Krause
المصدر: Communications in Computer and Information Science ISBN: 9783030611323
SFDI/LSGDA@VLDBمصطلحات موضوعية: 020203 distributed computing, Data processing, Social graph, business.industry, Computer science, Multiprocessing, 02 engineering and technology, Parallel computing, Data structure, Set (abstract data type), Analytics, 020204 information systems, Scalability, 0202 electrical engineering, electronic engineering, information engineering, Feature (machine learning), business
الوصف: Graph-structured data can be found in nearly every aspect of today’s world which contributes to an increasing importance of this data structure for storing and processing data. From a processing perspective, finding comprehensive patterns in graph-structured data is a processing primitive in a variety of applications, such as fraud detection, biological engineering or social graph analytics. On the hardware side, multiprocessor systems—consisting of multiple processors in a single scale-up server—are the next important wave on top of multi-core systems. In particular, symmetric multiprocessor systems (SMP) are characterized by the fact, that each processor has the same architecture, e.g., every processor is a multi-core and all multiprocessors share a common and huge main memory space. Moreover, large SMPs will feature a non-uniform memory access (NUMA), whose impact on the design of efficient data processing concepts is considerable. In this paper, we give an overview of NeMeSys, our system for scalable near-memory graph pattern matching (GPM) on SMPs. NeMeSys is built on a synthesis of well-known concepts of database systems including a set of graph-tailored and hardware-oriented optimization techniques for scalable GPM on SMPs.
URL الوصول: https://explore.openaire.eu/search/publication?articleId=doi_________::eda9a11e43a571a0978c1c0fa3a482bb
https://doi.org/10.1007/978-3-030-61133-0_4 -
10
المؤلفون: Wolfgang Lehner, Dirk Habich, Johannes Luong
المصدر: IEEE BigData
مصطلحات موضوعية: Computer science, Distributed computing, Interface (computing), Perspective (graphical), Integration platform, 020206 networking & telecommunications, Programmprozessoren, Strukturierte Abfragesprache, Laufzeit, Große Daten, Planung, Motoren, Datenintegration, 02 engineering and technology, computer.software_genre, Set (abstract data type), 020204 information systems, Program processors, Structured Query Language, Runtime, Big Data, Planning, Engines, Data integration, 0202 electrical engineering, electronic engineering, information engineering, ddc:004, computer, Data integration
الوصف: Storing and processing data at different locations using a heterogeneous set of formats and data managements systems is state-of-the-art in many organizations. However, data analyses can often provide better insight when data from several sources is integrated into a combined perspective. In this paper we present an overview of our data integration system DataCalc. DataCalc is an extensible integration platform that executes adhoc analytical queries on a set of heterogeneous data processors. Our novel platform uses an expressive function shipping interface that promotes local computation and reduces data movement between processors. In this paper, we provide a discussion of the overall architecture and the main components of DataCalc. Moreover, we discuss the cost of integrating additional processors and evaluate the overall performance of the platform.
URL الوصول: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::87d6cd819ae95016b7637acbce76da4e
https://doi.org/10.1109/bigdata47090.2019.9006252