droundup_lwork()

double precision function droundup_lwork

(

integer

lwork

)

DROUNDUP_LWORK

Purpose:

 DROUNDUP_LWORK deals with a subtle bug with returning LWORK as a Float.
 This routine guarantees it is rounded up instead of down by
 multiplying LWORK by 1+eps when it is necessary, where eps is the relative machine precision.
 E.g.,

        float( 9007199254740993            ) == 9007199254740992
        float( 9007199254740993 ) * (1.+eps) == 9007199254740994

 \return DROUNDUP_LWORK
 \verbatim
         DROUNDUP_LWORK >= LWORK.
         DROUNDUP_LWORK is guaranteed to have zero decimal part.

Parameters

[in]

LWORK

Workspace size.

Author

Weslley Pereira, University of Colorado Denver, USA

Further Details:

  This routine was inspired in the method `magma_zmake_lwork` from MAGMA.
  \see https://bitbucket.org/icl/magma/src/master/control/magma_zauxiliary.cpp

Definition at line 58 of file droundup_lwork.f.

*
*  -- LAPACK auxiliary routine --
*  -- LAPACK is a software package provided by Univ. of Tennessee,    --
*  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
*
*     .. Scalar Arguments ..
      INTEGER          LWORK
*     ..
*
* =====================================================================
*     ..
*     .. Intrinsic Functions ..
      INTRINSIC         epsilon, dble, int
*     ..
*     .. Executable Statements ..
*     ..
      droundup_lwork = dble( lwork )
*
      IF( int( droundup_lwork ) .LT. lwork ) THEN
*         Force round up of LWORK
          droundup_lwork = droundup_lwork * ( 1.0d+0 + epsilon(0.0d+0) )
      ENDIF
*
      RETURN
*
*     End of DROUNDUP_LWORK
*

Here is the call graph for this function:

Here is the caller graph for this function: