Next:
4.1 BLAS libraries
Up:
Parallelization of numerical algorithms
Previous:
3.4.3 Communication expenditure
Contents
4. Basic routines of numerical algorithms
Subsections
4.1 BLAS libraries
4.1.1 Vector-by-Vector operations (BLAS1)
4.1.1.1 Determining the inner product on a vector unit
4.1.1.2 Inner product on a parallel machine with distributed memory
4.1.2 Matrix-by-Vector operations (BLAS2)
4.1.2.1 Parallel machines with distributed memory
4.1.3 Matrix-by-Matrix-operations (BLAS3)
4.1.3.1 Matrix-by-Matrix multiplication
4.1.3.2 Algorithms for
4.1.3.3 Parallelization of
4.2 Operations with sparse matrices
4.2.1 Storing sparse matrices
4.2.1.1 Compressed Row Storage (CRS)
4.2.1.2 Skyline Storage
4.2.2 Operations with sparse matrices on a vector unit
4.3 Domain decomposition and basic numerical routines
4.3.1 Non-overlapping elements
4.3.1.1 Generation of stiffness matrix.
4.3.1.2 Change of vector types
4.3.1.3 Inner product.
4.3.1.4 Matrix-times-Vector multiplication.
4.3.2 Non-overlapping nodes
4.3.2.1 Generation of stiffness matrix.
4.3.2.2 Change of vector type.
4.3.2.3 Inner product.
4.3.2.4 Matrix-times-Vector multiplication.
4.3.3 Overlapping elements
4.3.3.1 Generation of stiffness matrix.
4.3.3.2 Change of vector types.
4.3.3.3 Inner product.
4.3.3.4 Matrix-by-Vector multiplication.
4.3.4 Overlapping nodes
Gundolf Haase 2000-03-20
