Basker: Parallel sparse LU factorization utilizing hierarchical parallelism and data layouts

Booth, Joshua Dennis; Ellingwood, Nathan David; Thornquist, Heidi K.; Rajamanickam, Sivasankaran

doi:10.1016/j.parco.2017.06.003

Cited by 13 publications

(6 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…“NIRB” for “near-infrared branding” [56,65–68], produces landmarks with low pixel intensity in EM and can damage ultrastructure [63,69]. To facilitate ultrastructural measurements in non-obscured identified dendrites, we took advantage of the photo-oxidability of DAB [70,71]. We immersed the samples in DAB solution and applied focalized UV light at user-defined positions (Fig 1A) to imprint osmiophilic DAB landmarks around targeted dendrites (see panels B-E in S1 Fig) and pattern the tissue with localized electron-dense DAB precipitates (Fig 1B).…”

Section: Resultsmentioning

confidence: 99%

Morphologically constrained modeling of spinous inhibition in the somato-sensory cortex

Gemin

Serna

Assendorp

et al. 2020

Preprint

View full text Add to dashboard Cite

Pyramidal neurons are covered by thousands of dendritic spines receiving excitatory synaptic inputs. The ultrastructure of dendritic spines shapes signal compartmentalization but ultrastructural diversity is rarely taken into account in computational models of synaptic integration. Here, we developed a 3D correlative light-electron microscopy (3D-CLEM) approach allowing the analysis of specific populations of synapses in genetically defined neuronal types in intact brain circuits. We used it to reconstruct segments of basal dendrites of layer 2/3 pyramidal neurons of adult mouse somatosensory cortex and quantify spine ultrastructural diversity. We found that 10% of spines were dually-innervated and 38% of inhibitory synapses localize to spines. Using our morphometric data to constrain a model of synaptic signal compartmentalization, we assessed the impact of spinous versus dendritic shaft inhibition. Our results indicate that spinous inhibition is locally more efficient than shaft inhibition and that it can decouple voltage and calcium signaling, potentially impacting synaptic plasticity.

show abstract

Section: Resultsmentioning

confidence: 99%

Morphologically constrained modeling of spinous inhibition in the somato-sensory cortex

Gemin

Serna

Assendorp

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…In particular, the preconditioners in Ifpack2 are mostly serial and/or MPI parallel, while ShyLU was designed to take advantage of on-node (shared memory) parallelism. The primary focus on exascale readiness has been on developing accelerator focused iterative solvers in Belos, multithreaded and GPU-ready multigrid methods in MueLu, Kokkos-based direct factorizations such as Basker [66] and Tacho [67] in ShyLU, GPU-ready domain decomposition methods in Ifpack2 and ShyLU.…”

Section: Trilinos and Kokkos Kernelsmentioning

confidence: 99%

Preparing sparse solvers for exascale computing

Anzt

Boman

Falgout

et al. 2020

Phil. Trans. R. Soc. A.

Self Cite

View full text Add to dashboard Cite

Sparse solvers provide essential functionality for a wide variety of scientific applications. Highly parallel sparse solvers are essential for continuing advances in high-fidelity, multi-physics and multi-scale simulations, especially as we target exascale platforms. This paper describes the challenges, strategies and progress of the US Department of Energy Exascale Computing project towards providing sparse solvers for exascale computing platforms. We address the demands of systems with thousands of high-performance node devices where exposing concurrency, hiding latency and creating alternative algorithms become essential. The efforts described here are works in progress, highlighting current success and upcoming challenges. This article is part of a discussion meeting issue ‘Numerical algorithms for high-performance computational science’.

show abstract

“…Consider the graph and the elimination tree from Figure 8. Since there exists the edge (1,6) in G(A), therefore another (fill) edge (5, 6) must exist (here not a fill edge, but a regular edge). Similarly, the same conclusion can be drawn from the existence of the edge (4, 6).…”

Section: Path Compressionmentioning

confidence: 99%

“…Next we turn our attention to parallel performance. While KLU remains a benchmark for serial solvers, for parallel solvers, MKL-PARDISO is often cited as the benchmark [6,8]. To give the reader a sense of the progress in parallel sparse matrix methods, in Figure 6 we compare KLU, PARDISO (Version 6.2) to MKL-PARDISO on up to 16 cores on an Intel Xeon E7-4880 architecture with 2.5 GHz processors.…”

Section: Application -Circuit Simulationmentioning

confidence: 99%

“…Others may be related to implementation. Research on sparse matrix factorization for circuit simulation continues to draw attention, especially in the area of acceleration with Intel's many integrated core (MIC) architecture [6] and GPUs [7,31]. Other techniques for acceleration include improved preconditioners for iterative solvers [44].…”

Section: Application -Circuit Simulationmentioning

confidence: 99%

See 1 more Smart Citation

State-of-The-Art Sparse Direct Solvers

Bollhöfer¹,

Schenk²,

Janalík³

et al. 2019

Preprint

View full text Add to dashboard Cite

In this chapter we will give an insight into modern sparse elimination methods. These are driven by a preprocessing phase based on combinatorial algorithms which improve diagonal dominance, reduce fill-in and improve concurrency to allow for parallel treatment. Moreover, these methods detect dense submatrices which can be handled by dense matrix kernels based on multi-threaded level-3 BLAS. We will demonstrate for problems arising from circuit simulation how the improvement in recent years have advanced direct solution methods significantly.

show abstract

Basker: Parallel sparse LU factorization utilizing hierarchical parallelism and data layouts

Cited by 13 publications

References 18 publications

Morphologically constrained modeling of spinous inhibition in the somato-sensory cortex

Morphologically constrained modeling of spinous inhibition in the somato-sensory cortex

Preparing sparse solvers for exascale computing

State-of-The-Art Sparse Direct Solvers

Contact Info

Product

Resources

About