Store the lower double-precision (64-bit) floating-point element from a into 2 contiguous elements in memory. mem_addr must be aligned on a 16-byte boundary or a general-protection exception may be generated.