Better precision than the regular mul opcode.

v2: handle big endian properly.
Reviewed-by: NMichel Dänzer <michel.daenzer@amd.com>
Signed-off-by: NAlex Deucher <alexander.deucher@amd.com>