If the -m64bit-doubles command-line option has been specified then then double directive generates double precision (64-bit) floating point constants, otherwise it generates single precision (32-bit) floating point constants. To force the generation of 64-bit floating point constants used the dc.d directive instead.