Communication Optimizations for Irregular Scientific Computations on Distributed Memory Architectures

Das, Raja; Uysal, Mustafa; Saltz, Joel; Hwang, Yuan-Shin

doi:10.1006/jpdc.1994.1104

Cited by 166 publications

(76 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Previous work on inspectors in the context of languages such as HPF and Titanium [8,1,14] have suggested reordering loop iterations to differentiate local and non-local accesses. Consider a common sparse matrix-vector product shown in Figure 2, taken from the NAS Conjugate Gradient benchmark.…”

Section: Figure 4 Restructuring Of Sparse Matrixvector Multiplicatiomentioning

confidence: 99%

“…This execution scheme could also add significant overhead to serial sections, as these needed be executed on multiple nodes owning parts of the data. Handling irregular data was difficult and usually employed runtime schemes [8]. In contrast to HPF implementations, our execution model starts from the available parallelism specified through OpenMP directives.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Programming Distributed Memory Sytems Using OpenMP

Basumallik

Min

Eigenmann

2007

2007 IEEE International Parallel and Distributed Processing Symposium

View full text Add to dashboard Cite

show abstract

Section: Figure 4 Restructuring Of Sparse Matrixvector Multiplicatiomentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Programming Distributed Memory Sytems Using OpenMP

Basumallik

Min

Eigenmann

2007

2007 IEEE International Parallel and Distributed Processing Symposium

View full text Add to dashboard Cite

show abstract

“…Pingali et al manually regrouped work in integer computations, including tree traversals [17]. Dynamic regrouping is also part of the inspector-executor model, originally studied by Saltz and others for partitioning irregular computations on distributed-memory parallel machines [5]. Most of these studies considered data transformation in conjunction with computation reordering.…”

Section: Related Workmentioning

confidence: 99%

A Key-based Adaptive Transactional Memory Executor

Bai

Shen

Zhang

et al. 2007

2007 IEEE International Parallel and Distributed Processing Symposium

View full text Add to dashboard Cite

Software transactional memory systems enable a programmer to easily write concurrent data structures such as lists, trees, hashtables, and graphs, where non-conflicting operations proceed in parallel. Many of these structures take the abstract form of a dictionary, in which each transaction is associated with a search key. By regrouping transactions based on their keys, one may improve locality and reduce conflicts among parallel transactions.In this paper, we present an executor that partitions transactions among available processors. Our key-based adaptive partitioning monitors incoming transactions, estimates the probability distribution of their keys, and adaptively determines the (usually nonuniform) partitions. By comparing the adaptive partitioning with uniform partitioning and round-robin keyless partitioning on a 16-processor SunFire 6800 machine, we demonstrate that key-based adaptive partitioning significantly improves the throughput of fine-grained parallel operations on concurrent data structures.

show abstract

“…Initially, such transformations were incorporated into applications manually for parallelism [15]. Next, libraries with run-time transformation primitives were developed so that a programmer or compiler could insert calls to such primitives [16,55]. Currently, there are run-time reordering transformations for which a compiler can automatically analyze and generate the inspectors [74,19,40,26].…”

Section: Introductionmentioning

confidence: 99%

An approach for code generation in the Sparse Polyhedral Framework

et al. 2016

View full text Add to dashboard Cite

Applications that manipulate sparse data structures contain memory reference patterns that are unknown at compile time due to indirect accesses such as A [B[i]]. To exploit parallelism and improve locality in such applications, prior work has developed a number of run-time reordering transformations (RTRTs). This paper presents the Sparse Polyhedral Framework (SPF) for specifying RTRTs and compositions thereof and algorithms for automatically generating efficient inspector and executor code to implement such transformations. Experimental results indicate that the performance of automatically generated inspectors and executors competes with the performance of hand-written ones in some cases.

show abstract

Communication Optimizations for Irregular Scientific Computations on Distributed Memory Architectures

Cited by 166 publications

References 2 publications

Programming Distributed Memory Sytems Using OpenMP

Programming Distributed Memory Sytems Using OpenMP

A Key-based Adaptive Transactional Memory Executor

An approach for code generation in the Sparse Polyhedral Framework

Contact Info

Product

Resources

About