Towards Optimization of Hybrid CPU/GPU Query Plans in Database Systems

Breß, Sebastian; Schallehn, Eike; Geist, Ingolf

doi:10.1007/978-3-642-32518-2_3

Cited by 14 publications

(32 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Since graphics cards lack support for virtual function calls -and are notoriously bad at runing the complex control logic that would be neccesary to emulate them -this model is unsuited for a GDBMS. Furthermore, we identified in prior work that tuple-wise processing is not possible on the GPU, due to lacking support for inter-kernel communication [15]. We therefore argue that a GDBMS should utilize an operator-at-a-time model.…”

Section: Functional Propertiesmentioning

confidence: 95%

“…They used a Selinger-style optimizer to create initial query plans and then used heuristics and an analytical cost-model to split a workload between CPU and GPU. In our previous work, we proposed a framework that can perform cost-based operation-wise scheduling and cost-based optimization of hybrid CPU/GPU query plans, which is designed to be used with operator-at-a-time bulk processing [15]. Przymus and others developed a query planner that is capable of optimizing for two goals simultaneously (e.g., query response time and energy consumption) [51].…”

Section: Functional Propertiesmentioning

confidence: 99%

“…Hence, a GPU-aware optimizer has to identify sub plans of a query plan, which it can process on the CPU or the GPU [29]. Furthermore, the resulting plan should minimize the number of copy operations [15]. Since optimizers are typically cost based, a GDBMS needs for each GPU operator a cost model.…”

Section: Potential Optimizations For Gdbmssmentioning

confidence: 99%

“…Additionally, block-oriented processing is a necessary prerequisite for overlapping processing and data transfer for single operations and allows for a more fine grained workload distribution on available processing devices [63]. Note that traditional pipelining of blocks between GPU operators is not possible, because inter-kernel communication is undefined [15]. While launching a new kernel for each block is likely to be expensive, query compilation and kernel fusion are promising ways to allow block-oriented processing on the GPU as well.…”

Section: Potential Optimizations For Gdbmssmentioning

confidence: 99%

“…While there has been prior work in this direction [14,15,29], GPU-aware query optimization remains an open challenge. Query Compilation for Multiple Devices: With the upcoming trend of query compilation, the basic problem of processing-device allocation remains the same as in traditional query optimization.…”

Section: Generic Cost Modelmentioning

confidence: 99%

See 4 more Smart Citations

GPU-Accelerated Database Systems: Survey and Open Challenges

Breß

Heimel

Siegmund

et al. 2014

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

Abstract. The vast amount of processing power and memory bandwidth provided by modern graphics cards make them an interesting platform for data-intensive applications. Unsurprisingly, the database research community identified GPUs as effective co-processors for data processing several years ago. In the past years, there were many approaches to make use of GPUs at different levels of a database system. In this paper, we explore the design space of GPU-accelerated database management systems. Based on this survey, we present key properties, important trade-offs and typical challenges of GPU-aware database architectures, and identify major open challenges. Additionally, we survey existing GPU-accelerated DBMSs and classify their architectural properties. Then, we summarize typical optimizations implemented in GPU-accelerated DBMSs. Finally, we propose a reference architecture, indicating how GPU acceleration can be integrated in existing DBMSs.

show abstract

Section: Functional Propertiesmentioning

confidence: 95%

Section: Functional Propertiesmentioning

confidence: 99%

Section: Potential Optimizations For Gdbmssmentioning

confidence: 99%

Section: Potential Optimizations For Gdbmssmentioning

confidence: 99%

Section: Generic Cost Modelmentioning

confidence: 99%

See 3 more Smart Citations

GPU-Accelerated Database Systems: Survey and Open Challenges

Breß

Heimel

Siegmund

et al. 2014

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

show abstract

A Scalable Sparse Matrix-Based Join for SPARQL Query Processing

Zhang

Peng

et al. 2019

Database Systems for Advanced Applications

View full text Add to dashboard Cite

Resource Description Framework (RDF) has been widely used to represent information on the web, while SPARQL is a standard query language to manipulate RDF data. Given a SPARQL query, there often exist many joins which are the bottlenecks of efficiency of query processing. Besides, the real RDF datasets often reveal strong data sparsity, which indicates that a resource often only relates to a few resources even the number of total resources is large. In this paper, we propose a sparse matrix-based (SM-based) SPARQL query processing approach over RDF datasets which considers both join optimization and data sparsity. Firstly, we present a SM-based storage for RDF datasets to lift the storage efficiency, where valid edges are stored only, and then introduce a predicatebased hash index on the storage. Secondly, we develop a scalable SM-based join algorithm for SPARQL query processing. Finally, we analyze the overall cost by accumulating all intermediate results and design a query plan generated algorithm. Besides, we extend our SM-based join algorithm on GPU for parallelizing SPARQL query processing. We have evaluated our approach compared with the state-of-the-art RDF engines over benchmark RDF datasets and the experimental results show that our proposal can significantly improve SPARQL query processing with high scalability.

show abstract

Exploring the Design Space of a GPU-Aware Database Architecture

Breß

Heimel

Siegmund

et al. 2014

Advances in Intelligent Systems and Computing

Self Cite

View full text Add to dashboard Cite

Abstract. The vast amount of processing power and memory bandwidth provided by modern graphics cards make them an interesting platform for data-intensive applications. Unsurprisingly, the database research community has identified GPUs as effective co-processors for data processing several years ago. In the past years, there were many approaches to make use of GPUs at different levels of a database system. In this paper, we summarize the major findings of the literature on GPU-accelerated data processing. Based on this survey, we present key properties, important trade-offs and typical challenges of GPU-aware database architectures, and identify major open research questions.

show abstract

Towards Optimization of Hybrid CPU/GPU Query Plans in Database Systems

Cited by 14 publications

References 8 publications

GPU-Accelerated Database Systems: Survey and Open Challenges

GPU-Accelerated Database Systems: Survey and Open Challenges

A Scalable Sparse Matrix-Based Join for SPARQL Query Processing

Exploring the Design Space of a GPU-Aware Database Architecture

Contact Info

Product

Resources

About