Re: Structure of large link libraries in f95




Charles Russell wrote:

Is there any way in f95 to structure a large link library such as LAPACK
without either 1) having to link the whole library when calling only a
few routines or 2) having to create and maintain a large collection of
explicit interfaces?

It could depend on the OS you are using, but most operating systems
still support archives.

LAPACK is structured as separately compileable routines. If you
put the routines in an archive, programs linked against the archive
will extract only the routines the program needs from the archive.

LAPACK , at least up to the third edition of LAPACK, uses
FORTRAN 77 compatible interfaces, so you don't need explicit
interfaces. If you want to use explicit interfaces, you need to
exercise a bit of caution. Some LAPACK implementations come
with include files that contain explicit interfaces for all the user
accessible routines in the implementation. There are
compiler/ linker combinations that will pull in all of the routines
named in the include file into the executable file.

Bob Corbett

.



Relevant Pages

  • Re: Matrix Multiplication in BLAS or Lapack
    ... > so you need not to link the lapack library. ... > LAPACK library includes the blas routines already due to ... > BLAS offers the generic and simple routines, e.g. daxpy, dgemv, dgemm, ... > playing around with the optimization options of the compiler - ...
    (sci.math.num-analysis)
  • Re: sin(10^300), ignorance, collusion. was Re: TILE64 embedded multicore processors - Guy Macon
    ... And LAPACK gives reliable answers so long as the arithmetic isn't ... count routines for bisection, ... slower fall-backs for crap arithmetics. ... And there may be some optimizations ...
    (comp.arch)
  • Re: Matrix Multiplication in BLAS or Lapack
    ... BLAS-library and not of LAPACK; ... LAPACK library includes the blas routines already due to ... BLAS offers the generic and simple routines, e.g. daxpy, dgemv, dgemm, ... playing around with the optimization options of the compiler - ...
    (sci.math.num-analysis)
  • Re: LAPACK, et al., vs. Numerical Recipes
    ... Recipes and LAPACK, preferably including ATLAS and/or Goto BLAS. ... the solver routines in Lapack are probably better ... Victor Eijkhout -- eijkhout at tacc utexas edu ...
    (sci.math.num-analysis)
  • Re: Lapack ?
    ... Martin Raspaud writes: ... > Does anyone know if there is a 100% Common lisp implementation of Lapack ... > routines for matrix inversion? ...
    (comp.lang.lisp)