Adding -fdefault-double-8 to gfortran build flags

The -fdefault-real-8 flag (required by the atmosphere core) promotes
all `double precision` types to real(16). The -fdefault-double-8 keeps
`double precision` types the same as real(8).

When `double precision` is real(16) the MPI_WTime routines return
incorrect timer information, and cause the timers to all be zero.

Author: douglasjacobsen

Makefile

@@ -121,10 +121,10 @@ gfortran:
 	"CC_PARALLEL = mpicc" \
 	"FC_SERIAL = gfortran" \
 	"CC_SERIAL = gcc" \
-	"FFLAGS_OPT = -O3 -m64 -ffree-line-length-none -fdefault-real-8 -fconvert=big-endian -ffree-form" \
+	"FFLAGS_OPT = -O3 -m64 -ffree-line-length-none -fdefault-real-8 -fdefault-double-8 -fconvert=big-endian -ffree-form" \
 	"CFLAGS_OPT = -O3 -m64" \
 	"LDFLAGS_OPT = -O3 -m64" \
-	"FFLAGS_DEBUG = -g -m64 -ffree-line-length-none -fdefault-real-8 -fconvert=big-endian -ffree-form -fbounds-check -fbacktrace -ffpe-trap=invalid,zero,overflow" \
+	"FFLAGS_DEBUG = -g -m64 -ffree-line-length-none -fdefault-real-8 -fdefault-double-8 -fconvert=big-endian -ffree-form -fbounds-check -fbacktrace -ffpe-trap=invalid,zero,overflow" \
 	"CFLAGS_DEBUG = -g -m64" \
 	"LDFLAGS_DEBUG = -g -m64" \
 	"CORE = $(CORE)" \
@@ -138,10 +138,10 @@ gfortran-openmpi:
 	"CC_PARALLEL = openmpicc" \
 	"FC_SERIAL = gfortran" \
 	"CC_SERIAL = gcc" \
-	"FFLAGS_OPT = -O3 -m64 -ffree-line-length-none -fdefault-real-8 -fconvert=big-endian -ffree-form" \
+	"FFLAGS_OPT = -O3 -m64 -ffree-line-length-none -fdefault-real-8 -fdefault-double-8 -fconvert=big-endian -ffree-form" \
 	"CFLAGS_OPT = -O3 -m64" \
 	"LDFLAGS_OPT = -O3 -m64" \
-	"FFLAGS_DEBUG = -g -m64 -ffree-line-length-none -fdefault-real-8 -fconvert=big-endian -ffree-form -fbounds-check -fbacktrace -ffpe-trap=invalid,zero,overflow" \
+	"FFLAGS_DEBUG = -g -m64 -ffree-line-length-none -fdefault-real-8 -fdefault-double-8 -fconvert=big-endian -ffree-form -fbounds-check -fbacktrace -ffpe-trap=invalid,zero,overflow" \
 	"CFLAGS_DEBUG = -g -m64" \
 	"LDFLAGS_DEBUG = -g -m64" \
 	"CORE = $(CORE)" \