Incorrect handling of EQUIVALENCE in ifx

Simon_Richards — Fri, 30 Sep 2022 10:16:47 GMT

The following example code gives incorrect results with ifx version 2022.1.0

PROGRAM ifx_equiv IMPLICIT NONE REAL :: a(5) = [1.0, 2.0, 3.0, 4.0, 5.0] REAL :: b(5) = [6.0, 7.0, 8.0, 9.0,10.0] REAL :: c(10) EQUIVALENCE (a(1), c(1)) EQUIVALENCE (b(1), c(6)) ! DATA a /1.0, 2.0, 3.0, 4.0, 5.0/ ! DATA b /6.0, 7.0, 8.0, 9.0, 10.0/ ! a = [1.0, 2.0, 3.0, 4.0, 5.0] ! b = [6.0, 7.0, 8.0, 9.0,10.0] WRITE(*,'("a = ", 5F5.1)') a WRITE(*,'("b = ", 5F5.1)') b WRITE(*,'("c = ", 10F5.1)') c END PROGRAM ifx_equiv

The (incorrect) output when compiled with ifx is:

a = 1.0 2.0 3.0 4.0 5.0 b = 0.0 0.0 0.0 0.0 0.0 c = 1.0 2.0 3.0 4.0 5.0 0.0 0.0 0.0 0.0 0.0

The (correct) output when compiled with ifort or gfortran is:

a = 1.0 2.0 3.0 4.0 5.0 b = 6.0 7.0 8.0 9.0 10.0 c = 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

If you replace the default initialization of the arrays with the commented out DATA statements the same error occurs. However, if you explicitly initialize the arrays, as in the commented out lines 15 and 16, then the error does not occur.

This looks to me like an ifx compiler bug.

I realize that EQUIVALENCE is an obsolete language feature which can be unsafe and should be avoided. This minimum example is based on a legacy routine in a production code which was found to give incorrect results with the ifx compiler as a consequence of this error.

Re: Incorrect handling of EQUIVALENCE in ifx

Steve_Lionel — Fri, 30 Sep 2022 13:16:43 GMT

Still fails in ifx 2022.2.0.

Re: Incorrect handling of EQUIVALENCE in ifx

Ron_Green — Fri, 30 Sep 2022 16:35:19 GMT

We'll get a bug report on this. Thanks for catching it.

Re: Incorrect handling of EQUIVALENCE in ifx

Devorah_H_Intel — Wed, 19 Oct 2022 18:02:45 GMT

This was fixed in future release.

ifort equiv.f90 -o eq $ ./eq a = 1.0 2.0 3.0 4.0 5.0 b = 6.0 7.0 8.0 9.0 10.0 c = 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 $ ifx equiv.f90 -o eq $ ./eq a = 1.0 2.0 3.0 4.0 5.0 b = 6.0 7.0 8.0 9.0 10.0 c = 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 $

topic Re: Incorrect handling of EQUIVALENCE in ifx in Intel® Fortran Compiler

Incorrect handling of EQUIVALENCE in ifx

Re: Incorrect handling of EQUIVALENCE in ifx

Re: Incorrect handling of EQUIVALENCE in ifx

Re: Incorrect handling of EQUIVALENCE in ifx