| 1 |
/* { dg-do compile } */ |
| 2 |
/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect -fdump-tree-vect-details" } */ |
| 3 |
|
| 4 |
typedef unsigned char uint8_t; |
| 5 |
typedef unsigned int uint32_t; |
| 6 |
|
| 7 |
#define HADAMARD4(d0, d1, d2, d3, s0, s1, s2, s3) {\ |
| 8 |
int t0 = s0 + s1;\ |
| 9 |
int t1 = s0 - s1;\ |
| 10 |
int t2 = s2 + s3;\ |
| 11 |
int t3 = s2 - s3;\ |
| 12 |
d0 = t0 + t2;\ |
| 13 |
d1 = t1 + t3;\ |
| 14 |
d2 = t0 - t2;\ |
| 15 |
d3 = t1 - t3;\ |
| 16 |
} |
| 17 |
|
| 18 |
void sink(uint32_t tmp[4][4]); |
| 19 |
|
| 20 |
int x264_pixel_satd_8x4( uint8_t *pix1, int i_pix1, uint8_t *pix2, int i_pix2 ) |
| 21 |
{ |
| 22 |
uint32_t tmp[4][4]; |
| 23 |
int sum = 0; |
| 24 |
for( int i = 0; i < 4; i++, pix1 += i_pix1, pix2 += i_pix2 ) |
| 25 |
{ |
| 26 |
uint32_t a0 = (pix1[0] - pix2[0]) + ((pix1[4] - pix2[4]) << 16); |
| 27 |
uint32_t a1 = (pix1[1] - pix2[1]) + ((pix1[5] - pix2[5]) << 16); |
| 28 |
uint32_t a2 = (pix1[2] - pix2[2]) + ((pix1[6] - pix2[6]) << 16); |
| 29 |
uint32_t a3 = (pix1[3] - pix2[3]) + ((pix1[7] - pix2[7]) << 16); |
| 30 |
HADAMARD4( tmp[i][0], tmp[i][1], tmp[i][2], tmp[i][3], a0,a1,a2,a3 ); |
| 31 |
} |
| 32 |
sink(tmp); |
| 33 |
} |
| 34 |
|
| 35 |
/* { dg-final { scan-tree-dump "vectorizing stmts using SLP" "vect" } } */ |
| 36 |
/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */ |