Available on x86 and target feature
sse2 only.Expand description
Loads 128-bits (composed of 2 packed double-precision (64-bit)
floating-point elements) from memory into the returned vector.
mem_addr does not need to be aligned on any particular boundary.