Difference between revisions of "LAPACK"

From CBLFS
Jump to navigationJump to search
(Created page.)
 
 
(16 intermediate revisions by one other user not shown)
Line 2: Line 2:
 
|-
 
|-
 
!Download Source:
 
!Download Source:
−
|  
+
| http://www.netlib.org/lapack/lapack-{{Lapack-Version}}.tgz
 +
|}
 +
 
 +
----
 +
 
 +
{{Package-Introduction|LAPACK is a '''L'''inear '''A'''lgebra '''PACK'''age that provides routines for solving systems of simultaneous linear equations, least-squares solutions of linear systems of equations, eigenvalue problems, and singular value problems. The associated matrix factorizations (LU, Cholesky, QR, SVD, Schur, generalized Schur) are also provided, as are related computations such as reordering of the Schur factorizations and estimating condition numbers. Dense and banded matrices are handled, but not general sparse matrices. In all areas, similar functionality is provided for real and complex matrices, in both single and double precision.  LAPACK routines are written so that as much as possible of the computation is performed by calls to the [[ATLAS]].
 +
 
 +
LAPACK libraries can also be obtained from ATLAS (incomplete) or the microprocessor vendor's math library, such as AMD's [[ACML]].|http://www.netlib.org/lapack/}}
 +
 
 +
{{Note|These instructions do not install the resulting libraries.  Once the LAPACK libraries are built, build [[ATLAS]].  You will then have optimized ATLAS and LAPACK libraries installed.}}
 +
 
 +
== Dependencies ==
 +
 
 +
=== Required ===
 +
 
 +
One of the following is required to build LAPACK:
 +
 
 +
* [[GCC-3.3]] built with F77 support
 +
* [[GCC]] built with gfortran support
 +
 
 +
== Configuration Information ==
 +
 
 +
There is no configure script for LAPACK.  All options are set in the make.inc file.  You may wish to modify the following options.
 +
 
 +
{|
 +
|<i>PLAT=</i> appends the architecture to the library name.  Beneficial if building LAPACK for more than one architecture.
 
|-
 
|-
 +
|<i>OPTS=</i> sets the options passed to the FORTRAN compiler.
 
|}
 
|}
  
−
LAPACK is a '''L'''inear '''A'''lgebra '''PACK'''age that provides routines for solving systems of simultaneous linear equations, least-squares solutions of linear systems of equations, eigenvalue problems, and singular value problems. The associated matrix factorizations (LU, Cholesky, QR, SVD, Schur, generalized Schur) are also provided, as are related computations such as reordering of the Schur factorizations and estimating condition numbers. Dense and banded matrices are handled, but not general sparse matrices. In all areas, similar functionality is provided for real and complex matrices, in both single and double precision.  LAPACK routines are written so that as much as possible of the computation is performed by calls to the [[BLAS]].
+
{{Note|Do not remove the source tree if you plan to build the LAPACK 95 interface.}}
 +
 
 +
== Non-Multilib ==
 +
 
 +
First we create a make.inc to use in our build.
 +
 
 +
cp -v make.inc.example make.inc &&
 +
sed -i 's@../../blas$(PLAT).a@/usr/lib/blas/atlas/libblas.a@' make.inc &&
 +
sed -i "s/OPTS    = -funroll-all-loops/& ${CFLAGS} -fPIC/" make.inc
 +
sed -i "s/NOOPT    =/& -fPIC/" make.inc &&
 +
sed -i "s/EXT_ETIME/INT_CPU_TIME/" make.inc
 +
 
 +
If you haven't built the F77 compiler provided by gcc-3.3, you need to tell LAPACK that you're using gfortran.
 +
 
 +
sed -i 's@g77@gfortran@g' make.inc
 +
 
 +
Compile the LAPACK libraries:
 +
 
 +
make lib
 +
 
 +
Install the documentation that comes with the LAPACK library:
 +
 
 +
install -v -m755 -d /usr/share/doc/LAPACK-{{Lapack-Version}}/html &&
 +
install -v INSTALL/lawn81.{pdf,ps} /usr/share/doc/LAPACK-{{Lapack-Version}}/ &&
 +
for doc in html/*.html; do
 +
  cp -v $doc /usr/share/doc/LAPACK-{{Lapack-Version}}/html
 +
done
 +
 
 +
== Multilib ==
 +
 
 +
First we create a make.inc to use in our build.
 +
 
 +
cp -v make.inc.example make.inc &&
 +
sed -i "s/NOOPT    =/& -fPIC/" make.inc &&
 +
sed -i "s/EXT_ETIME/INT_CPU_TIME/" make.inc
 +
 
 +
If you haven't built the F77 compiler provided by gcc-3.3, you need to tell LAPACK that you're using gfortran.
 +
 
 +
sed -i 's@g77@gfortran@g' make.inc
 +
 
 +
=== 32Bit ===
 +
 
 +
Add your build options to the make file:
 +
 
 +
sed -i 's@../../blas$(PLAT).a@/usr/lib/blas/atlas/libblas.a@' make.inc &&
 +
sed -i "s/OPTS    = -funroll-all-loops/& ${BUILD32} -fPIC/" make.inc
 +
 
 +
Compile the LAPACK libraries:
 +
 
 +
  make lib
 +
 
 +
=== N32 ===
 +
 
 +
Add your build options to the make file:
 +
 
 +
sed -i 's@../../blas$(PLAT).a@/usr/lib32/blas/atlas/libblas.a@' make.inc &&
 +
sed -i "s/OPTS    = -funroll-all-loops/& ${BUILDn32} -fPIC/" make.inc
 +
 
 +
Compile the LAPACK libraries:
  
−
ATLAS (Automatically Tuned Linear Algebra Software) will provide C and FORTRAN interfaces to a LAPACK implementation. The LAPACK libraries provided by ATLAS are not optimized for a specific machine architecture.  The ideal source of BLAS and LAPACK libraries are those provided by the microprocesor vendor such as AMD's [[ACML]].  However, these are not provided under a license that allows re-distribution without written consent from the vendor.
+
  make lib
  
−
== Dependencies ==
+
=== 64Bit ===
 +
 
 +
Add your build options to the make file:
 +
 
 +
sed -i 's@../../blas$(PLAT).a@/usr/lib64/blas/atlas/libblas.a@' make.inc &&
 +
sed -i "s/OPTS    = -funroll-all-loops/& ${BUILD64} -fPIC/" make.inc
 +
 
 +
Compile the LAPACK libraries:
  
−
=== Optional ===
+
make lib
  
−
[[BLAS]] is recommended otherwise LAPACK will install an unoptimized version of the libraries.
+
Install the documentation that comes with the LAPACK library:
  
−
== Non-Multilib or Multilib ==
+
install -v -m755 -d /usr/share/doc/LAPACK-{{Lapack-Version}}/html &&
 +
install -v INSTALL/lawn81.{pdf,ps} /usr/share/doc/LAPACK-{{Lapack-Version}}/ &&
 +
for doc in html/*.html; do
 +
  cp -v $doc /usr/share/doc/LAPACK-{{Lapack-Version}}/html
 +
done
  
−
  To do.
+
[[Category:Science_Engineering]]
−
== Contents ==
 

Latest revision as of 16:39, 19 March 2009

Download Source: http://www.netlib.org/lapack/lapack-3.1.1.tgz

Introduction to LAPACK

LAPACK is a Linear Algebra PACKage that provides routines for solving systems of simultaneous linear equations, least-squares solutions of linear systems of equations, eigenvalue problems, and singular value problems. The associated matrix factorizations (LU, Cholesky, QR, SVD, Schur, generalized Schur) are also provided, as are related computations such as reordering of the Schur factorizations and estimating condition numbers. Dense and banded matrices are handled, but not general sparse matrices. In all areas, similar functionality is provided for real and complex matrices, in both single and double precision. LAPACK routines are written so that as much as possible of the computation is performed by calls to the ATLAS.

LAPACK libraries can also be obtained from ATLAS (incomplete) or the microprocessor vendor's math library, such as AMD's ACML.

Project Homepage: http://www.netlib.org/lapack/

Caution.png

Note

These instructions do not install the resulting libraries. Once the LAPACK libraries are built, build ATLAS. You will then have optimized ATLAS and LAPACK libraries installed.

Dependencies

Required

One of the following is required to build LAPACK:

  • GCC-3.3 built with F77 support
  • GCC built with gfortran support

Configuration Information

There is no configure script for LAPACK. All options are set in the make.inc file. You may wish to modify the following options.

PLAT= appends the architecture to the library name. Beneficial if building LAPACK for more than one architecture.
OPTS= sets the options passed to the FORTRAN compiler.
Caution.png

Note

Do not remove the source tree if you plan to build the LAPACK 95 interface.

Non-Multilib

First we create a make.inc to use in our build. 

cp -v make.inc.example make.inc &&
sed -i 's@../../blas$(PLAT).a@/usr/lib/blas/atlas/libblas.a@' make.inc &&
sed -i "s/OPTS     = -funroll-all-loops/& ${CFLAGS} -fPIC/" make.inc
sed -i "s/NOOPT    =/& -fPIC/" make.inc &&
sed -i "s/EXT_ETIME/INT_CPU_TIME/" make.inc

If you haven't built the F77 compiler provided by gcc-3.3, you need to tell LAPACK that you're using gfortran. 

sed -i 's@g77@gfortran@g' make.inc

Compile the LAPACK libraries: 

make lib

Install the documentation that comes with the LAPACK library: 

install -v -m755 -d /usr/share/doc/LAPACK-3.1.1/html &&
install -v INSTALL/lawn81.{pdf,ps} /usr/share/doc/LAPACK-3.1.1/ &&
for doc in html/*.html; do
  cp -v $doc /usr/share/doc/LAPACK-3.1.1/html
done

Multilib

First we create a make.inc to use in our build. 

cp -v make.inc.example make.inc &&
sed -i "s/NOOPT    =/& -fPIC/" make.inc &&
sed -i "s/EXT_ETIME/INT_CPU_TIME/" make.inc

If you haven't built the F77 compiler provided by gcc-3.3, you need to tell LAPACK that you're using gfortran. 

sed -i 's@g77@gfortran@g' make.inc

32Bit

Add your build options to the make file: 

sed -i 's@../../blas$(PLAT).a@/usr/lib/blas/atlas/libblas.a@' make.inc &&
sed -i "s/OPTS     = -funroll-all-loops/& ${BUILD32} -fPIC/" make.inc

Compile the LAPACK libraries: 

make lib

N32

Add your build options to the make file: 

sed -i 's@../../blas$(PLAT).a@/usr/lib32/blas/atlas/libblas.a@' make.inc &&
sed -i "s/OPTS     = -funroll-all-loops/& ${BUILDn32} -fPIC/" make.inc

Compile the LAPACK libraries: 

make lib

64Bit

Add your build options to the make file: 

sed -i 's@../../blas$(PLAT).a@/usr/lib64/blas/atlas/libblas.a@' make.inc &&
sed -i "s/OPTS     = -funroll-all-loops/& ${BUILD64} -fPIC/" make.inc

Compile the LAPACK libraries: 

make lib

Install the documentation that comes with the LAPACK library: 

install -v -m755 -d /usr/share/doc/LAPACK-3.1.1/html &&
install -v INSTALL/lawn81.{pdf,ps} /usr/share/doc/LAPACK-3.1.1/ &&
for doc in html/*.html; do
  cp -v $doc /usr/share/doc/LAPACK-3.1.1/html
done
Retrieved from "?title=LAPACK&oldid=17725"