A level of floating point accuracy that usually requires twice the space for each value than does single precision, but provides considerably more precision. For most systems running the SPEC CPU tests from the OSG (e.g. CPU2000), double precision implies a 64-bit value