performance - Linux x86/x86-64 Fortran 90/95/2003/2008 compiler, which? and, why? -


well, i've been asking myself question quite long time now.

i started using gnu fortran compiler (gfortran) , quite happy. started hearing intel fortran compiler (ifort) better because optimized intel processors (and therefore optimized computers today consequence intel has greater share of market). after started using ifort, truth i'm not happy because not detect errors gfortran (i.e. question: "reference passing changing values of matrix").

it true there other fortran compilers but, actually, never use other gfortran , ifort.

so ask you: advantages , disadvantages of each one? best one, if optimization not important?

in opinion, best stick standard you're comfortable with, , write code conforming standard. then, choice of compilers dictated support said standard.

here extensive list of compiler support 2003 , 2008 standards.

i myself used g95 lot because liked debug messages. nowadays, commonly use fortran 2008 features, , use gfortran (because it's open source), , ifort (because it's free non-commercial use). g95's support fortran 2003 , 2008 quite limited.

i compile code both compilers , warnings turned on, , quite i'm surprised errors missed 1 of compilers. in terms of performance of compiled code, never noticed difference (excluding bundled math libraries).

most commercially available compilers have non-standard extensions, openacc (e.g. cray, pgi) or special features (e.g. pgi cuda fortran) , bundled libraries (e.g. intel mkl). these might influence choice. hardware might need / might profit special compilers (like cray compiler).

edit: here benchmark results different compilers , machines.


Comments

Popular posts from this blog

java.util.scanner - How to read and add only numbers to array from a text file -

rewrite - Trouble with Wordpress multiple custom querystrings -