Re: what's the ifc corresponding argument for -bp in absoft f90

From: Richard Maine (nospam_at_see.signature)
Date: 10/20/03 

Date: 20 Oct 2003 11:42:14 -0700

"Shi Jin" <jinzishuai@hotmail.com> writes:

> program ifctest
> real*8::x
> real::y
>
> x=1.0d0
> y=1.0
> print *,dsin(x),dsin(y)
...
> The problem is dsin(y) where y is a single precision floating point while
> the expected argument of dsin is a double precision.
> Basically this is the code's problem.

Yes. Guess I have to ask why you don't just fix the code?
There is almost never any good reason to use specific intrinsics
like dsin in the first place. You really ought to remove all
references to them from the code.

> But we have a way to overcome this problem under absoft f90 compiler --
> using the -dp option.

Is this the *ONLY* reason you are using the -dp option? To make
illegal calls to specific intrinsics compile? The -dp option
has *FAR* more implications than that. If this was the only reason
you were using -dp, I'd have to expect that it would break
more things than it would fix. I'm guessing that the code needs
the -dp option for more reasons than this and that this is just the
first symptom that you noticed.

I generally advise people to fix codes that need things like -dp,
but I recognize that sometimes that can be too much work to be
practical. I'll give you that advice also, with the same caveat.
If your only problem is the specific intrinsics, then I have
difficulty in believing that it is impractical to fix them.
(Sometimes things that I have difficulty in believing turn out
to be true anyway). If the problem goes deeper, then your job
might be harder.

> But for the intel ifc compiler for linux, we are having trouble finding out
> what's the corresponding switch .

I don't off-hand know. I'm pretty sure that the compiler docs come
with documentation of all the switches, though if I recall correctly,
there are so many of them that it might well be hard to find the one
you need. I wouldn't guarantee that it necessarily has such a switch.
Not all compilers do. That's part of why I recommend fixing codes that
need such a switch. 

-- 
Richard Maine                       |  Good judgment comes from experience;
email: my first.last at org.domain  |  experience comes from bad judgment.
org: nasa, domain: gov              |        -- Mark Twain

Relevant Pages

  • Re: Messenger Pro on RO
    ... but what I have ain't broke so I have no reason to fix it or to ... switch to something else. ...
    (comp.sys.acorn.apps)
  • [git pull] Input updates for 2.6.33-rc0
    ... keyboard - fix theoretical race on vt switch ... DMI: allow omitting ident strings in DMI tables ... i8042 - remove identification strings from DMI tables ... wistron_btns - switch to using sparse keymap library ...
    (Linux-Kernel)
  • Re: HEADSUP usb2 (usb4bsd) to become default in 2 weeks.
    ... on Jan 3rd I will switch the GENERIC kernel to use ... is there any reason why this needs to be rushed in over the ... developer attention and updates (and yeah there might be painful ... and the milestones to take and open items. ...
    (freebsd-current)
  • Re: IMPLICIT NONE (F2k8+/-)
    ... The assertion that there is absolutely no good reason to continue with a convention that has served Fortran for 50 years, is itself, no good reason for change. ... Those of us who like implicit typing find that we can instantly recognize the type of a variable from its initial letter, provided that explicit typing has not been turned on. ... We have to conclude that only single variables are susceptible to mistyping and not being trapped. ... IMPLICIT NONE is the case that could be selected by a compiler switch, as the user of this option will have declared the type of every variable explicitly, and other than the case where the programmer has made an error, it is immaterial whether or not it is used. ...
    (comp.lang.fortran)
  • Re: Sen. Specter threatens to cut off funding for secret wiretaps
    ... Of course that's not Iraq either but you get my drift. ... the same reason Clinton wasn't convicted by the senate. ... fix it is to legislatively close that loophole. ... initiating impeachment proceedings would be political suicide. ...
    (rec.martial-arts)