The Use of BLAS and LAPACK in PETSc and external libraries#
BLAS 1 operations (and GPU equivalents) - vector operations such as
VecScale()are used extensively in PETSc. Depending on the simulation the size of the vectors may be from hundreds of entries to many millions.
BLAS 2 operations - dense matrix with vector operations, generally the dense matrices are very small.
Eigenvalue and SVD computations, generally for very small matrices
External packages such as MUMPS and SuperLU_DIST use BLAS 3 operations (and possibly BLAS 1 and 2). The dense matrices may be of modest size, going up to thousands of rows and columns.
For most PETSc simulations (that is not using certain external packages) using an optimized set of BLAS/LAPACK routines only provides a modest improvement in performance. For some external packages using optimized BLAS/LAPACK can make a dramatic improvement in performance.
32 or 64-bit BLAS/LAPACK integers#
BLAS/LAPACK libraries may use 32 or 64-bit integers. PETSc configure and compile handles this automatically
so long at the arguments to the BLAS/LAPACK routines are set to the type
PetscBLASInt. The routine
PetscBLASInt *) casts
PetscInt to the BLAS/LAPACK size. If the BLAS/LAPACK size is not large enough it generates an error. For the vast majority of
simulations, even very large ones, 64-bit BLAS/LAPACK integers are not needed, even when 64-bit PETSc integers are used, The configure
--with-64-bit-blas-indices attempts to locate and use a 64-bit integer version of BLAS/LAPACK library. Except for MKL CPardiso, the external packages do not support using
64-bit BLAS/LAPACK integers so if you are using such packages you cannot use 64-bit BLAS/LAPACK integers.
Available BLAS/LAPACK libraries#
Most systems (besides Microsoft Windows) come with pre-installed BLAS/LAPACK which are satisfactory for many PETSc simulations.
The freely available Intel MKL mathematics libraries provide BLAS/LAPACK that are generally better performing than the system provided libraries and are generally fine for most users.
For systems that do not provide BLAS/LAPACK, such as Microsoft Windows, PETSc provides the Fortran reference version
--download-fblaslapack and a f2c generated C version
--download-f2cblaslapack (which also supports 128 bit real number computations).
These libraries are less optimized but useful to get started with PETSc easily.
PETSc also provides access to OpenBLAS via the
--download-openblas configure option. OpenBLAS uses some highly optimized operations but falls back on reference
routines for many other operations. See the OpenBLAS manual for more information. The configure option
--download-openblas provides a full BLAS/LAPACK implementation.
BLIS does not bundle LAPACK with it so PETSc’s configure attempts to locate a compatible system LAPACK library to use if
selected. One can use
--download-f2cblaslapack --download-blis. This is recommended as a portable high-performance option.