[BOOK][B] Communication-avoiding Krylov subspace methods in theory and practice
EC Carson - 2015 - search.proquest.com
Advancements in the field of high-performance scientific computing are necessary to
address the most important challenges we face in the 21st century. From physical modeling …
address the most important challenges we face in the 21st century. From physical modeling …
Reducing communication in the conjugate gradient method: a case study on high-order finite elements
Currently, a major bottleneck for several scientific computations is communication, both
communication between different processors, so-called horizontal communication, and …
communication between different processors, so-called horizontal communication, and …
Preconditioning strategies for stochastic elliptic partial differential equations
N Venkovic - 2023 - theses.hal.science
We are interested in the Monte Carlo (MC) sampling of discretized elliptic partial differential
equations (PDEs) with random variable coefficients. The dominant computational load of …
equations (PDEs) with random variable coefficients. The dominant computational load of …
Avoiding communication in primal and dual block coordinate descent methods
Primal and dual block coordinate descent methods are iterative methods for solving
regularized and unregularized optimization problems. Distributed-memory parallel …
regularized and unregularized optimization problems. Distributed-memory parallel …
Varying the s in your s-step GMRES
D Imberti, J Erhel - Electronic Transactions on Numerical Analysis, 2017 - inria.hal.science
Krylov subspace methods are commonly used iterative methods for solving large sparse
linear systems, however they suffer from communication bottlenecks on parallel computers …
linear systems, however they suffer from communication bottlenecks on parallel computers …
[PDF][PDF] Prospectus for the Next LAPACK and ScaLAPACK Libraries: Basic ALgebra LIbraries for Sustainable Technology with Interdisciplinary Collaboration …
The convergence of several unprecedented changes, including formidable new system
design constraints and revolutionary levels of heterogeneity, has made it clear that much of …
design constraints and revolutionary levels of heterogeneity, has made it clear that much of …
Communication-Efficient, 2D Parallel Stochastic Gradient Descent for Distributed-Memory Optimization
Distributed-memory implementations of numerical optimization algorithm, such as stochastic
gradient descent (SGD), require interprocessor communication at every iteration of the …
gradient descent (SGD), require interprocessor communication at every iteration of the …
StressBench: a configurable full system network and I/O benchmark framework
We present StressBench, a network benchmarking framework written for testing MPI
operations and file I/O concurrently. It is designed specifically to execute MPI communication …
operations and file I/O concurrently. It is designed specifically to execute MPI communication …
s-Step Enlarged Krylov Subspace Conjugate Gradient Methods
SM Moufawad - SIAM Journal on Scientific Computing, 2020 - SIAM
Recently, enlarged Krylov subspace methods, which consist of enlarging the Krylov
subspace by a maximum of t vectors per iteration based on the domain decomposition of the …
subspace by a maximum of t vectors per iteration based on the domain decomposition of the …
Deflation strategies to improve the convergence of communication-avoiding GMRES
The generalized minimum residual (GMRES) method is a popular method for solving a large-
scale sparse nonsymmetric linear system of equations. On modern computers, especially on …
scale sparse nonsymmetric linear system of equations. On modern computers, especially on …