LAPACK Archives

[Lapack] second possible bugfix for another floating point exception in

Hello,

the lapack 3.1.0 testing breaks on a CRAY with a floating point 
exception in zunmrz.f, too.

So i am proposing the following similar patch, which seems to fix the 
problem:

kite:/ha/rull/lapack-3.1.0/SRC$ cat zunmrz.f.patch
*** zunmrz.f.orig       Thu Jan 25 17:50:24 2007
--- zunmrz.f    Thu Jan 25 18:02:28 2007
***************
*** 170,177 ****
        END IF
  *
        IF( INFO.EQ.0 ) THEN
           IF( M.EQ.0 .OR. N.EQ.0 ) THEN
!             LWKOPT = 1
  *
  *           Determine the block size.  NB may be at most NBMAX, where
  *           NBMAX is used to define the local array T.
--- 170,178 ----
        END IF
  *
        IF( INFO.EQ.0 ) THEN
+          LWKOPT = 1
           IF( M.EQ.0 .OR. N.EQ.0 ) THEN
! ******            LWKOPT = 1
  *
  *           Determine the block size.  NB may be at most NBMAX, where
  *           NBMAX is used to define the local array T.
kite:/ha/rull/lapack-3.1.0/SRC$

With regards
Ingo


<Prev in Thread] Current Thread [Next in Thread>
  • [Lapack] second possible bugfix for another floating point exception in LAPACK 3.1.0 test, Dr.-Ing. Ingo D. Rullhusen <=


For additional information you may use the LAPACK/ScaLAPACK Forum.
Or one of the mailing lists, or