Compiler error in fft3dlib.f90

Message

bendix · #1 Post by **bendix** » Mon Jan 08, 2007 2:41 am

hello~
I complier vasp 4.6 with IFC 9.1.040 but get some errors.
These errors constrain me to stop complier...
Please tell me what to do!
Thanks~

my machines
compiler:IFC 9.1.040
lib.:MKL 8.1
CPU: Intel core 2 duo E6600
distr.:Fedora core 6

see below errors:
fft3dlib.f90(1538) : (col. 10) remark: loop was not vectorized: not inner loop.
fft3dlib.f90(1539) : (col. 30) remark: loop was not vectorized: unsupported loop structure.
fortcom: Error: fft3dlib.f90, line 1627: Sharing of a DO termination statement by more than one DO statement is an obsolescent feature in Fortran 95. Use an END DO or CONTINUE statement for each DO statement. [20]
20 CONTINUE
---^
fortcom: Error: fft3dlib.f90, line 1704: The computed GOTO statement is an obsolescent feature in Fortran 95.
GOTO (10,50,90,130,170,210,250),IGO
------^
fortcom: Error: fft3dlib.f90, line 2625: The computed GOTO statement is an obsolescent feature in Fortran 95.
GOTO (10,50,90,130,170,210,250),IGO
------^
fortcom: Error: fft3dlib.f90, line 3531: The computed GOTO statement is an obsolescent feature in Fortran 95.
GOTO (10,50,90,130,170,210,250),IGO
------^
fortcom: Error: fft3dlib.f90, line 4064: The computed GOTO statement is an obsolescent feature in Fortran 95.
GOTO (1010,1050,1090,1130,1170,1210,1250),IGO
------^
compilation aborted for fft3dlib.f90 (code 1)
make: *** [fft3dlib.o] Error 1

#2 Post by **admin** » Mon Jan 08, 2007 12:19 pm

these statements are standard fortran 77, however fortran 90 or 95 compilers should be backwards compatible and therefore be able to compile these statements.

bendix · #3 Post by **bendix** » Mon Jan 08, 2007 1:01 pm

I know these statements are standard fortran 77~
At the line 1704 "20 CONTINUE" loss a loop number, so i give it a loop number to fix this!
NOw, these "GOTO" errors...i don't what to do, because F95 doesn't support this "GOTO" which jump to many plates at the same time!

Please teach me how can neglect these errors?

Thanks a lot~~

tjf · #4 Post by **tjf** » Tue Jan 09, 2007 9:48 am

man ifort

Look for options regarding standards conformance, language levels and error/warning levels.

bendix · #5 Post by **bendix** » Sun Jan 14, 2007 8:03 am

I fix the line 343 to del "-e95" in makefile.linux_ifc_P4 and successd to compile, finally.

See below to fix
old:
$(FC) -FR -lowercase -O1 -tpp7 -xW -prefetch- -unroll0 -e95 -vec_report3 -c $*$(SUFFIX)
new:
$(FC) -FR -lowercase -O1 -tpp7 -xW -prefetch- -unroll0 -vec_report3 -c $*$(SUFFIX)

lahaye · #6 Post by **lahaye** » Tue May 22, 2007 9:17 am

Once again I ran into issue when compiling the upgrade 4.6.31.

The IFC makefiles still contain the "-e95" flag, but the fft3dlib is still written
in old F77 language style. Although 'admin' is right, that F95 compilers
should be able to handle F77 style (to some extend), there is an intrinsic
conflict here:

1) By default, the "-e95" flag is in the makefiles, which terminates
the compilation with error when non-standard F95 code is found
(read: old F77 code is found).

2) Keep the old F77 code in the respective fortran file.

This combination is doomed to fail and each time users (like me) have
to dig for a while to find out what to do about this.

I'd suggest to fix this, by either making fft3dlib.F complaint to F95, or
modify the makefiles provided by VASP.

Cheers,
Rob.

#7 Post by **admin** » Wed May 23, 2007 12:16 pm

please make the adjustments in the makefile yourself. This compiler flag ONLY causes conflicts with intel frortan compiler release 9.1, all previous releases compiled vasp successfully.

job · #8 Post by **job** » Wed May 23, 2007 1:15 pm

To be somewhat pedantic, there were a small number of features that were deleted in the Fortran 95 standard, so it's not true that all valid F77 is also valid F95. See e.g http://www.nsc.liu.se/~boein/f77to90/f95.html