Making TifaMMy fit for tomorrow: Towards future shared memory systems and beyond
Authors:
- 1 July 2011
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 517-524
Abstract
In this paper, we present the recent port to and latest results of our cache-oblivious algorithms and implementations of parallel LU decomposition code TifaMMy on two new architectures: SGI's UltraViolet distributed shared memory machine, and Intel's latest x86 architecture Sandy Bridge. TifaMMy's matrix multiplication and LU decomposition routines have been further optimized with regard to these new architectures. Results are discussed and compared with Intel's architecture specific and optimized numerical Math Kernel Library (MKL) for both the standard C++ version with vectorization compiler switches and TifaMMy's highly optimized vector intrinsics version.Keywords
This publication has 9 references indexed in Scilit:
- Porting existing cache-oblivious linear algebra HPC modules to larrabee architecture2010
- Towards many-core implementation of LU decomposition using Peano Curves2009
- High-performance implementation of the level-3 BLASACM Transactions on Mathematical Software, 2008
- Hardware-Oriented Implementation of Cache Oblivious Matrix Operations Based on Space-Filling Curves2008
- Parallel matrix multiplication based on space-filling curves on shared memory multicore platforms2008
- An experimental comparison of cache-oblivious and cache-conscious programs2007
- Nonlinear array layouts for hierarchical memory systems1999
- Recursion leads to automatic variable blocking for dense linear-algebra algorithmsIBM Journal of Research and Development, 1997
- A set of level 3 basic linear algebra subprogramsACM Transactions on Mathematical Software, 1990