Subtracts the four single-precision, floating-point values of a and b (a - b) (see _mm_sub_ps)
在文件 AkSimd.h 第 344 行定义.