Abstract
In this paper, we present a numerical algorithm for computing products of the form R K RT, where R, RT, and K are sparse matrices. By reformulating the problem into the simultaneous processing of a sequential data and control stream, cache miss penalties are significantly reduced. Even though the algorithm increases memory requirements, it accelerates sparse matrix products on recent processor architectures by a factor of up to 4 compared to previous approaches. We apply the algorithm to compute consistent system matrices at different resolution levels in a dynamic multigrid elasticity simulation, and we show its efficiency for nested and non-nested mesh hierarchies.
Original language | English |
---|---|
Pages (from-to) | 263-275 |
Number of pages | 13 |
Journal | Electronic Transactions on Numerical Analysis |
Volume | 37 |
State | Published - 2010 |
Keywords
- Cache-awareness
- Galerkin update
- Multigrid
- Sparse matrix products