Solved: New bug? Previous code, ifort

Solved: New bug? Previous code, ifort
Solved: New bug? Previous code, ifort


Code has been working properly for a really very long time (from Intel Visible Fortran Model 8 via oneAPI 2024.2). The code was even working positive in the present day, however one thing has occurred after a reboot.

The actual code in query, construct with ifort (will check ifx shortly), fails in a subroutine prefix (iow earlier than any statements are entered).

The code runs positive in single threaded mode in Debug construct (no optimizations)

However fails with optimizations disabled .AND. configured to generate OpenMP parallel code.

The affected subroutine doesn’t use OpenMP (no directives, however a module it makes use of accommodates USE OMP_LIB)

jimdempseyatthecove_0-1720195565320.png

The subroutine has no arguments, the meeting code on the left is preamble code when compiled with OpenMP.

One of many modules has ThreadPrivate information, and that is the requirement for the preamble code.

The subroutine is (on this name) referred to as exterior of a parallel area (later it is going to be referred to as inside a parallel area).

Register data stepping via code:

jimdempseyatthecove_1-1720196673291.png

On the …2326 location it will get an error studying location 3.

Apart from for a system reboot, I can’t account for this symptom.

 

Attempting ifx now…

 

Jim Dempsey

 

 

Leave a Reply

Your email address will not be published. Required fields are marked *