The goal is to develop faster x86 intrinsics for floating point class checks in Float and Double class methods: isNaN(), isFinite() and isInfinite() using vfpclasss/d instructions.