Abstract | ||
---|---|---|
An important number of Numerical Linear Algebra methods to tackle problems in diverse fields of science and engineering, rely heavily on the solution of one or many sparse triangular linear systems. Since the early years, this has motivated numerous efforts that seek to produce efficient implementations of this kernel for most hardware platforms. However, this operation implies strong data dependencies and unbalanced computations that difficult the concurrency, specially when massively-parallel processors such as GPUs are employed. In this work we review the different techniques to expose the data parallelism in this operation with special attention to the many-core based proposals. Additionally, we experimentally evaluate the two most successful approaches, namely the routine that is included in CUSPARSE library and the synchronization free method of W. Liu et al. [1]. Finally, we advance in the characterization of the triangular sparse linear systems to select the best solver in each case. |
Year | DOI | Venue |
---|---|---|
2017 | 10.1109/SBAC-PADW.2017.15 | 2017 International Symposium on Computer Architecture and High Performance Computing Workshops (SBAC-PADW) |
Keywords | Field | DocType |
graphic processors,multi-core processors,sparse triangular linear systems,high performance | Kernel (linear algebra),Linear system,Computer science,Concurrency,Parallel computing,Sparse approximation,Data parallelism,Solver,Numerical linear algebra,Sparse matrix | Conference |
ISBN | Citations | PageRank |
978-1-5386-4820-9 | 2 | 0.43 |
References | Authors | |
14 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Daniel Erguiz | 1 | 2 | 0.43 |
Ernesto Dufrechou | 2 | 25 | 11.02 |
Pablo Ezzatti | 3 | 124 | 28.24 |