These terms and technologies are used on the CPU (the "host" in NVIDIA's lingo) when running GPU programs.