Description of problem: All calls made to dlamc1 subroutine of lapack never return. Try the following test program: PROGRAM aaa INTEGER BETA, T LOGICAL RND, IEEE1 CALL DLAMC1( BETA, T, RND, IEEE1 ) END This behavior is compiler dependent and does not occur when the library is compiled with lower optimization options. I cannot tell whether it is a compiler bug (I tested with native gcc-g77-3.4.2-6.fc3) or actually a bug in algorithm: it seems to do pretty ugly things and - I think - rely on IEEE arithmetics. Older gcc versions (eg. gcc-g77-3.3.3-7) do not expose the bug. Version-Release number of selected component (if applicable): lapack-3.0-25 How reproducible: Always Steps to Reproduce: 1. Compile test program with "g77 test.f -llapack -lblas" and run it Actual Results: Program will never stop. Expected Results: Program should return instantenously. Additional info: Recompiling the routine with current g77 and only -O option makes the problem dissappear.
I seem to be unable to attach a file to the report. Anyway, the URL locates a file that contains relevant subroutines from lapack and servers as a test case. Compiling it with -O give a code that works as expected. Code compiled with -O2 will never stop.
*** This bug has been marked as a duplicate of 138447 ***
Changed to 'CLOSED' state since 'RESOLVED' has been deprecated.