Compilation aborted (code 1)

[Kaushik_Biswas]

I'm converting a fixed-format (.for) fortran program in free-format (.f90). The compiler aborts when I try to build the solution for the 'AtticSim_ASV_Version_III_DP.F90' file. The error code is 'Compilation aborted (code 1)'. I have attached both the 'f90' and '.for' files, and the build log for the '.f90' file. The build log suggests that the problem is with the subroutine 'VIEW2'.

I'm trying the compilation on a macbook using 'Parallels Desktop' and Windows 7. When I try to compile the same F90 source file on my PC desktop, also running Windows 7, it seems to work.

Any help or suggestions will be appreciated.

PS: I'm having network issues and may end with multiple posts. My apologies if that is the case.

Thanks,

Kaushik

[Steven_L_Intel1]

All I can suggest is trying the current compiler, Update 4. But if it fails only under Parallels I'm not sure what we can do about it. Still, we have fixed many bugs since the 12.0 compiler was initially released so try the current one and see what happens. You download updates from the Intel Registration Center.

[Kaushik_Biswas]

Hi Steve,

Thanks for the response. I tried Update 4, but still the same problem. I tried compilingthe older fixed format version ('.for'), but the compiler still aborted. A colleague of mine also uses Parallels Desktop > Windows 7 > Intel Fortran Compiler, and she was able to build and run the project, using the same '.F90' source code.

Any pointers for troubleshooting? I bascially reinstalled everything: Parallels Desktop > Windows 7 > Intel Fortran Compiler > Update 4. But still no luck.

Any help would be greatly appreciated.

Kaushik

[jimdempseyatthecove]

Some things to experiment with.

1) In your.F90 file, lines 537, 586, 588, ... (many other places) you have spaces following the line continuation character (&). This may be the (a)problem.

2) Your main program is using IMPLICIT NONE (good) but it is also using intrinsic functions (e.g. DSIN(x)) add USE IFPORT to the module USE list. This same issue resides in other functions and subroutines.

3) In your .F90 file, line 447 you have a unlabled CONTINUE statement. This continue statement starts at column 1. This should be innocuous, but you might try removing it.

5) You use mostly REAL(KIND=DBL) but often you use literals without suffix of _DBL or D0. Lack of proper suffix will affect accuracy in some places an performance in all places. Also with DBL consider using more accurate value for PI (and make it a parameter).

6) *** line 1854 and elsewheris (may)improperly continued and/or commented

ELSE IF (ABS(PHI) .GT. 1.D-3 .AND. & ! NEARLY HORIZONTAL SURFACES

The IVF reference guide shows an example like your codeabove, however, in source forms section for free form it states that the & must appear at the end of the source line (at least in my reference guide).

There may be some other things. Work through the above list and see what happens.

(bleep'n editor cann't properly drag and drop formatted text....)

Jim Dempsey


ABS(PHI) .LT. 2.) THEN ! UP TO 2 DEG TILT

[Steven_L_Intel1]

I don't think there's a problem with the formatting - the rule about continuation applies only to continues character constants. In any event, such a thing would not cause an internal compiler error.

I suggest you start cuttung down the source to see what the minimum it takes to show the error.

[Les_Neilson]

As matter of interest I downloaded the f90 file and it compiles & links with IVF 11.1.065 (we haven't yet upgraded to 12)

When I ran the program it stopped of course when trying to open the data files which I don't have.

I am running Windows Vista.

Les

[Kaushik_Biswas]

Jim,

I tried the steps you mentioned, but still getting the internal error. I've made further changes and additions to the F90 code and it successfully compiles on my desktop running Windows 7. The internal error only happens when I am using Parellels Desktop > Windows 7 on my macbook.

Steve, I'll try your suggestion about cutting down the source code to find out where the error originates.

But, any other suggestions? Could it be a 32- vs 64- bit installation issue? I believe I used the 64-bit installation. As mentioned earlier, a colleague of mine also using Parallels Desktop > Windows 7 could compile the program. Therefore I am thinking that it has to be some installation issue with my software(s). Anyone have any similar experience?

The latest F90 file is attached, if it helps.

Thanks.
Kaushik

[Steven_L_Intel1]

If those "err_read" meassages are from the compiler, and they may well be (I have never seen them before), it looks as if the compiler is having difficulty reading the .mod files that get generated. Hmm - what if you disable /warn:interface there? Or better, what if you use update 4?

[Santiago_S_]

I had the same problem when moving from CVF6.6 to IVF XE 12.0 update 4. The way I proceeded to find the culprit was:

1- Comment all lines of code, except:

* program start / end

* subroutine protocols (subroutine x(var1, ...))

* subroutine end

* implicit none

* variables declarations

2- Incrementally uncomment small pieces of code.

I isolated the issue. I was passing a string constant as an argument for a subroutine call. The corresponding parameter of the subroutine was

character*200 mystr

I replaced the string constant in the caller line of code by a variable name. That variable was also defined as character*200. That solved the problem.

[Steven_L_Intel1]

You don't know it's the same problem - only that it's the same symptom (internal compiler error.) This has many, many possible causes.

If you encounter this error with a current compiler release, please create a NEW thread here about the problem and provide a test case, or report it to Intel Premier Support. Do not not assume the cause of the error is the same as anyone else who had the same message.