NEON的vsub方法溢出

關於NEON的vsub方法的溢出，結果以下：

1. vsub會產生溢出，根據數據bit表示規律，可知溢出結果和理論正確結果造成互補，好比249-(-7)=256
2. 使用類vreinterpretq_s16_u16方法能夠獲得正確的結果

void neontest()
{
    uchar v1[] = {0,1,2,3,4,5,6,7};
    uchar v2[] = {7,6,5,4,3, 2, 1, 7};
    uchar vr[8];
    short vr2[8];
    unsigned short vr3[8];
    uint8x8_t d1 = vld1_u8(v1);
    uint8x8_t d2 = vld1_u8(v2);

    uint8x8_t r1 = vsub_u8(d1, d2);
    vst1_u8(vr, r1);
    LOGD("%d %d %d %d %d %d %d %d", vr[0], vr[1], vr[2], vr[3], vr[4], vr[5], vr[6], vr[7]);

    int16x8_t r2 =  vreinterpretq_s16_u16(vsubl_u8(d1, d2));
    vst1q_s16(vr2, r2);
    LOGD("%d %d %d %d %d %d %d %d", vr2[0], vr2[1], vr2[2], vr2[3], vr2[4], vr2[5], vr2[6], vr2[7]);

    uint16x8_t r3 =  vsubl_u8(d1, d2);
    vst1q_u16(vr3, r3);
    LOGD("%d %d %d %d %d %d %d %d", vr3[0], vr3[1], vr3[2], vr3[3], vr3[4], vr3[5], vr3[6], vr3[7]);
}

///////////////out put
D/LOG:==========main start================
D/LOG:249 251 253 255 1 3 5 0
D/LOG:-7 -5 -3 -1 1 3 5 0
D/LOG:65529 65531 65533 65535 1 3 5 0
D/LOG:==========main end================