This patch adds mid-end support for add/mul reduction operations Float16 in mainline JDK. It also includes backend aarch64 support for these operations.