Flextream: Adaptive Compilation of Streaming Applications for Heterogeneous Architectures

Hormati, Amir; Choi, Yoonseo; Kudlur, Manjunath; Rabbah, Rodric; Mudge, Trevor; Mahlke, Scott

doi:10.1109/pact.2009.39

Cited by 75 publications

(54 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Analytic Modeling Analytic models have been widely used to tackle complex optimization problems, such as autoparallelization [7,40], runtime estimation [12,27,58], and task mappings [28]. A particular problem with them, however, is that the model has to be re-tuned for each new targeted hardware [50].…”

Section: Related Workmentioning

confidence: 99%

Minimizing the cost of iterative compilation with active learning

William

Petoumenos

Wang

et al. 2017

2017 IEEE/ACM International Symposium on Code Generation and Optimization (CGO)

View full text Add to dashboard Cite

Since performance is not portable between platforms, engineers must fine-tune heuristics for each processor in turn. This is such a laborious task that high-profile compilers, supporting many architectures, cannot keep up with hardware innovation and are actually out-of-date. Iterative compilation driven by machine learning has been shown to be efficient at generating portable optimization models automatically. However, good quality models require costly, repetitive, and extensive training which greatly hinders the wide adoption of this powerful technique.In this work, we show that much of this cost is spent collecting training data, runtime measurements for different optimization decisions, which contribute little to the final heuristic. Current implementations evaluate randomly chosen, often redundant, training examples a pre-configured, almost always excessive, number of times -a large source of wasted effort. Our approach optimizes not only the selection of training examples but also the number of samples per example, independently. To evaluate, we construct 11 high-quality models which use a combination of optimization settings to predict the runtime of benchmarks from the SPAPT suite. Our novel, broadly applicable, methodology is able to reduce the training overhead by up to 26x compared to an approach with a fixed number of sample runs, transforming what is potentially months of work into days.

show abstract

Section: Related Workmentioning

confidence: 99%

Minimizing the cost of iterative compilation with active learning

William

Petoumenos

Wang

et al. 2017

2017 IEEE/ACM International Symposium on Code Generation and Optimization (CGO)

View full text Add to dashboard Cite

show abstract

“…In the academic arena, there are a number of systems that support different stream abstractions such as Stream [47], [48], StreamMIT [49], Aurora [50], Flextream [51], River [52], Cayuga [53] and Naiad [54]. None of those academic approaches seems to be focused on meeting deadlines, although some of them could be easily extended with the infrastructure proposed in our article to support end-toend deadlines.…”

Section: Distributed Stream Processing Patternsmentioning

confidence: 99%

Patterns for Distributed Real-Time Stream Processing

Val

Fernandez-Garcia

Sánchez-Fernandez

et al. 2017

IEEE Trans. Parallel Distrib. Syst.

View full text Add to dashboard Cite

Abstract-In recent yeais, big data systems have become an active area of research and development. Stream processing is one of the potential application scenarios of big data systems where the goal is to process aconlinoous, high velochyflow of information items. High frequencytradirg (HFT) in stock markets ortrendirgtopicdetection in Twitter are som e examples of stream processing applications. In some cases (like, for instance, in HFT), these applications have end-t�nd qualhy-of-service reqlirements and may benefh from the usage� real-time techniques. Taking this into account, the present articl e analyzes, from the point� view of real-time systems, a set of patterns that can be used when implementing a stream processing application. For each pattern, we discuss its advantages and dsadvanlages, as well as its impact in application performance, measured as response time, maximum ill)Ut frequency and changes in utilization demands due to the pattern.

show abstract

“…Analytic Modelling Analytic models have been widely used to tackle complex optimization problems, such as auto-parallelization [17,18], runtime estimation [19][20][21], and task mappings [22]. A particular problem with them, however, is the model has to be re-tuned whenever it is targeted at new hardware [23].…”

Section: Related Workmentioning

confidence: 99%

Fast Automatic Heuristic Construction Using Active Learning

Ogilvie

Petoumenos

Wang

et al. 2015

Languages and Compilers for Parallel Computing

View full text Add to dashboard Cite

Abstract. Building effective optimization heuristics is a challenging task which often takes developers several months if not years to complete. Predictive modelling has recently emerged as a promising solution, automatically constructing heuristics from training data. However, obtaining this data can take months per platform. This is becoming an ever more critical problem and if no solution is found we shall be left with out of date heuristics which cannot extract the best performance from modern machines. In this work, we present a low-cost predictive modelling approach for automatic heuristic construction which significantly reduces this training overhead. Typically in supervised learning the training instances are randomly selected to evaluate regardless of how much useful information they carry. This wastes effort on parts of the space that contribute little to the quality of the produced heuristic. Our approach, on the other hand, uses active learning to select and only focus on the most useful training examples. We demonstrate this technique by automatically constructing a model to determine on which device to execute four parallel programs at differing problem dimensions for a representative Cpu-Gpu based heterogeneous system. Our methodology is remarkably simple and yet effective, making it a strong candidate for wide adoption. At high levels of classification accuracy the average learning speed-up is 3x, as compared to the stateof-the-art.

show abstract

Flextream: Adaptive Compilation of Streaming Applications for Heterogeneous Architectures

Abstract: ABSTRACT

Cited by 75 publications

References 15 publications

Minimizing the cost of iterative compilation with active learning

Minimizing the cost of iterative compilation with active learning

Patterns for Distributed Real-Time Stream Processing

Fast Automatic Heuristic Construction Using Active Learning

Contact Info

Product

Resources

About