Hi,
On Mon, Apr 05, 2010 at 01:33:12PM -0700, Ross Walker wrote:
> I decided to try compiling AMBER Tools and AMBER with PGF90. Granted my
> pgf90 is very old v6.1.2 but this is all I have access to.
That one is too old. There are some known issues.
See this thread and earlier ones:
Subject: Re: [AMBER-Developers] Second release candidate for AmberTools 1.3;
numvec
> pgf90 -c -O1 -Mfree -o add_pdb.o _add_pdb.f90
> PGF90-S-0038-Symbol, command_argument_count, has not been explicitly
> declared (_add_pdb.f90: 1)
> 0 inform, 0 warnings, 1 severes, 0 fatal for add_pdb
>
> It looks like command_argument_count is a F2003 intrinsic. Are we allowing
> F2003 specifics in the code now or should this code be rewritten to be
> F90/95 compliant?
Short answer is yes for some since this and some others are in all tested
recent f95's.
> PGF90-S-0000-Internal compiler error. size_ast_of: unexpected ast type
> 4 (_xray_utils.f: 497)
> 0 inform, 0 warnings, 1 severes, 0 fatal for write_header
> make[1]: *** [xray_utils.o] Error 2
>
> Looking at the code this refers to:
>
> h_map_arch_stamp = transfer( CHAR(native_FT*16+native_FT) &
> // CHAR(native_IT*16+1) &
> // CHAR(0) // CHAR(0), &
> h_map_arch_stamp)
>
> = yuck! No idea what is going on here.
Joe and i had a discussion on this and others in circa may.
More or less we are punting on old pgi compilers.
> Does anyone have a more up to date PGF90 to test with?
Yes, Dave and i are testing with pgi 10 and 9; more will
be posted on the wiki soon.
scott
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Mon Apr 05 2010 - 14:30:02 PDT
