pub fn f64x2_floor(a: v128) -> v128
simd128
Lane-wise rounding to the nearest integral value not greater than the input.