A Split Execution Model for SpTRSV

Ahmad, Najeeb; Yilmaz, Buse; Unat, Didem

doi:10.1109/tpds.2021.3074501

Cited by 7 publications

(1 citation statement)

References 33 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…“Block‐diagonal based methods” is an umbrella term which is used for such partitioning methods for SpTRSV on CPUs 6 . Partitioning approach is a natural approach for SpTRSV and it has many successful examples in the literature 7‐12 . In a previous work, 13 we proposed exactly the opposite approach in which instead of partitioning the matrix according to the differences in the sparsity pattern of matrix parts (hence different degree of parallelism), the dependency graph of the matrix is transformed to make the sparsity pattern of the matrix more homogeneous.…”

Section: Introductionmentioning

confidence: 99%

A novel graph transformation strategy for optimizing SpTRSV on CPUs

Yilmaz

2023

Concurrency and Computation

Self Cite

View full text Add to dashboard Cite

SummarySparse triangular solve (SpTRSV) is an extensively studied computational kernel. An important obstacle in parallel SpTRSV implementations is that in some parts of a sparse matrix the computation is serial. By transforming the dependency graph, it is possible to increase the parallelism of the parts that lack it. In this work, we present a novel graph transformation strategy to increase the parallelism degree of a sparse matrix and compare it to our previous strategy. It is seen that our transformation strategy can provide a speedup as high as .

show abstract

Section: Introductionmentioning

confidence: 99%