All,
This is a minor little issue, but as I don't have direct ability to submit to the Intel ticketing system, I thought I'd whine here for hope of a fix in a future version.
Namely, due to my own idiocy recently with CMake and getting Fortran flags to match GNU Make, I encountered something interesting. Namely, when building GEOS (the model I work on with thousand+ Fortran files), every single compile spit out:
ifort: warning #10337: option '-fno-builtin' disables '-imf*' option
This was annoying because of the volume of messages. This was puzzling because I don't have '-fno-builtin' as a flag. Indeed, it does not appear in the man pages of Intel Fortran:
$ grep -ir no-builtin /usr/local/intel/2019/compilers_and_libraries_2019.2.187/linux/documentation/en/man/common/man1/*
$
So, a little investigation and it turns out it is '-nolib-inline' that causes it:
$ ifort -nolib-inline -fimf-arch-consistency=true PreciseSummationMod.F90 test.F90
ifort: warning #10337: option '-fno-builtin' disables '-imf*' option
ifort: warning #10337: option '-fno-builtin' disables '-imf*' option
$ ifort -fimf-arch-consistency=true PreciseSummationMod.F90 test.F90
$
My main appeal here would be to perhaps make that warning a bit more precise so that I'd know that I was (erroneously, it turns out) mixing these flags.
Matt
