+fltconst_strict HP Fortran option

[Michal_Kvasnicka]

What is the Intel compiler corresponding option to the HP Fortran compiler option "+fltconst_strict"?

+fltconst_strict: Uses a strict power table for converting floating constants with exponents.
Use of strict power table provides a more accurate floating point
representation.

Michal

[Steven_L_Intel1]

I don't know that we have an exact equivalent to that, but "-fp-model precise" is probably the closest. You might also look at the description of -fimf-precision.

[Michal_Kvasnicka]

Steve, thank you for your helpful answer.

I am trying to port obsolete HP Fortran code (about 100k line of source code),which works well onlyon PA-RISC CPUs with HPUX operating system. On Linux x86-64 + Intel fortran compiler 12 produce completely different (wrong)numerical results. I have no idea what is wrong, because program does not produce any warning and error messages.

Is there some fortran porting (HP-UX (PA-RISC)-> Linux (x86-64) + Intel fortran compiler) tutorial or white paper?

[Steven_L_Intel1]

I'm not aware of any such white paper. Have you first tried compiling without optimization (-O0) to see if you get better results? It may not be an option you need but something else. Perhaps you can add some code to see where in the program results start to differ. Sometimes implementation differences, such as the units for RECL=, can affect results.