topic Re: Double Precision in Software Archive

Double Precision

Intel_C_Intel — Tue, 12 Dec 2000 10:43:43 GMT

Hi all:
Is there any way to compile with global double precision, i.e., setting all constants,variables and intrinsics in double precision a la UNIX -r8 ?
Thanks,
Raul

Re: Double Precision

Steven_L_Intel1 — Wed, 13 Dec 2000 05:26:06 GMT

Yes. /real_size:64, though I think -r8 is also accepted. In DevStudio, its a setting under Fortran..Floating Point. Note that if you explicitly declare a variable REAL*4, it stays REAL*4. If you say just REAL, then it will get doubled.

Steve

Re: Double Precision

Intel_C_Intel — Wed, 13 Dec 2000 10:36:14 GMT

Steve: I added your compile option (it is not available as a DevStudio Fortran/Floating Point choice, and I can't find it anywhere in the documentation), ran a CFD code, and got slightly different results, so I take it that double precision "took". Much obliged by your timely help. Raul

Re: Double Precision

wkramer — Wed, 13 Dec 2000 17:17:40 GMT

From DevStudio go to Project Settings|Fortran|Fortran data. You can define a default real and integer kind there.

Walter Kramer