TiDA: High-Level Programming Abstractions for Data Locality Management

Unat, Didem; Nguyen, Tan; Zhang, Weiqun; Farooqi, Muhammed Nufail; Bastem, Burak; Almgren, Ann S.; Shalf, John

doi:10.1007/978-3-319-41321-1_7

Cited by 21 publications

(20 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Wherever profitable, the research plan can redeem existing languages by amending or extending them (e.g., by changing the specifications or by introducing new APIs). Examples include Kokkos [19], TiDA [66], GridTools [7], hStreams [35], and DASH [22]. The Kokkos library supports expressing multidimensional arrays in C++, in which the polymorphic layout can be decided at compile time.…”

Section: State Of the Artmentioning

confidence: 99%

“…Being placed higher along an axis implies greater challenges in achieving locality. HPC applications typically fall into the fourth quadrant, both spatial and functional connectivities are high [67]. Applications communities have well known and valid concerns about wisely utilizing the developers time and protecting the investment already made in the mature production codes of today [13], [31].…”

Section: Applications Expectations From Abstractionsmentioning

confidence: 99%

“…This development has opened up a real opportunity for the higher level abstractions to gain traction in the applications communities, especially when the application developers are kept in the loop. We conducted a series of workshops on the topic of programming abstractions for data locality for high performance computing (HPC) that gather practitioners and researchers from all applicable areas, including the computational scientists from multiple science domains [55], [56], [67]. This survey paper distills the outcomes of the series thus far.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Trends in Data Locality Abstractions for HPC Systems

Unat

Dubey

Hoefler

et al. 2017

IEEE Trans. Parallel Distrib. Syst.

Self Cite

View full text Add to dashboard Cite

Abstract-The cost of data movement has always been an important concern in high performance computing (HPC) systems. It has now become the dominant factor in terms of both energy consumption and performance. Support for expression of data locality has been explored in the past, but those efforts have had only modest success in being adopted in HPC applications for various reasons. However, with the increasing complexity of the memory hierarchy and higher parallelism in emerging HPC systems, locality management has acquired a new urgency. Developers can no longer limit themselves to low-level solutions and ignore the potential for productivity and performance portability obtained by using locality abstractions. Fortunately, the trend emerging in recent literature on the topic alleviates many of the concerns that got in the way of their adoption by application developers. Data locality abstractions are available in the forms of libraries, data structures, languages and runtime systems; a common theme is increasing productivity without sacrificing performance. This paper examines these trends and identifies commonalities that can combine various locality concepts to develop a comprehensive approach to expressing and managing data locality on future large-scale high-performance computing systems.

show abstract

Section: State Of the Artmentioning

confidence: 99%

Section: Applications Expectations From Abstractionsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Trends in Data Locality Abstractions for HPC Systems

Unat

Dubey

Hoefler

et al. 2017

IEEE Trans. Parallel Distrib. Syst.

Self Cite

View full text Add to dashboard Cite

show abstract

“…In addition, MPI often requires additional memory to buffer messages at both the application and system levels. In future architectures we expect to see a significant reduction in memory capacity per core and a substantial increase in the number of NUMA domains per compute node [31]. Regional tiling allows one to respect NUMA effects while operating with one MPI process per compute node.…”

Section: Smcmentioning

confidence: 99%

“…Due to the irregularity of AMR algorithms, a fine-grained loop-level threading approach is not expected to provide efficient parallelism on a node with hundreds or thousands of threads. In addition, reducing the cost of data movement is expected be crucial for performance [31]. In response to these architectural challenges, 2 we are adopting a new programming model, tiling, in BoxLib to expose additional parallelism and optimize data access behavior.…”

mentioning

confidence: 99%

BoxLib with Tiling: An Adaptive Mesh Refinement Software Framework

Zhang¹,

Almgren²,

Day³

et al. 2016

SIAM J. Sci. Comput.

Self Cite

View full text Add to dashboard Cite

In this paper we introduce a block-structured adaptive mesh refinement (AMR) software framework that incorporates tiling, a well-known loop transformation. Because the multiscale, multiphysics codes built in BoxLib are designed to solve complex systems at high resolution, performance on current and next generation architectures is essential. With the expectation of many more cores per node on next generation architectures, the ability to effectively utilize threads within a node is essential, and the current model for parallelization will not be sufficient. We describe a new version of BoxLib in which the tiling constructs are embedded so that BoxLib-based applications can easily realize expected performance gains without extra effort on the part of the application developer. We also discuss a path forward to enable future versions of BoxLib to take advantage of NUMA-aware optimizations using the TiDA portable library.

show abstract