Re: compilers: Compaq, Lahey, Intel

allelopath wrote:
This last year, I 'ported' a fortran program that had been compiled
(for years) on Compaq Visual Fortran to the Intel Fortran compiler and
it was kindofa headache. The Compaq compiler seemed sort of 'loose'.
It may have simply been a matter of compiler options selected, but
there were a couple of things that no compiler should let by that the
compaq did (imho).

Now, a guy is writing some new fortran code using the Lahey compiler.
Would a transition from Lahey to Intel be difficult? Are there
specific steps (compiler options? other?) to choose with the Lahey to
make that transition easier?

I would prefer if he use the Intel compiler outright (it is ultimately
what will be used when running). Are there compelling reasons for
doing so at the outset?

Hello,

I have the Lahey, PGI, g95, and gfortran compilers on my linux box. I have the Intel and g95 compilers on my macbook. I also use the IBM AIX xlf95 compiler every day. All the code I write runs on all those different compilers and platforms. And some other folks have built various bits of my code using the Absoft f95 compiler on linux, and on SGIs, Suns. Even Windows!

Lest you think I'm all puffed up with my own importance, the point I'm trying to make is that if all the interested parties agree to a set of coding guidelines that includes writing standard Fortran95 code (maybe including or not the usual TR), then -- apart from issues like added-value debuggers, profilers, optimisation, etc -- the actual compiler shouldn't matter (I can hear everyone in clf-land laughing as I type that :o)

In addition, set up a bunch of tests that you run regularly (daily, weekly, whatever) using the different platforms and/or compilers. That will highlight quickly when code written by the lahey-using fellow breaks, say, the Intel build. Or vice-versa.

I do this sort of thing by hand, but it can be (usually is?) automated. Bil Kleb posted about this sort of thing not so long ago -- I guess it's pretty common where more rigorous software QA is done. For lots of developers, automated is the way to go (of course).

cheers,

paulv

--
Paul van Delst Ride lots.
CIMSS @ NOAA/NCEP/EMC Eddy Merckx
.

Relevant Pages

  • linker mistakes entry point in dll
    ... Microsoft Program Maintenance Utility Version 6.00.8168.0 ... IntelFortran Compiler for 32-bit applications, ... Copyright 1985-2004 Intel Corporation. ...
    (comp.lang.fortran)
  • Re: who can help me? about OpenMP
    ... I know nothing about how Intel has implemented their compiler on ... Your best source of info on that is the Intel Fortran ... Fortran Compiler for Windows & Visual Fortran ...
    (comp.parallel)
  • Re: Intel Fortran Compiler 11.0 Now Available
    ... compiler knowing from the start what the language requires. ... Developer Products Division ... User communities for Intel Software Development Products ... Intel Fortran Support ...
    (comp.lang.fortran)
  • Re: Using IMSL libraries with Lahey/Intel fortran compilers
    ... > trying to compile a program that use them with the Lahey and Intel ... > fortran compilers. ... You need to get IMSL libraries that are built for the ... compiler you are using. ...
    (comp.lang.fortran)
  • Re: [SLE] Portland Group Fortran Linux compiler
    ... >> You may consider to download and install the free unsupported version of Intel ... >> I'm currently using the intel fortran compiler for all my fortran program and ... > using the GNU compiler. ...
    (SuSE)