1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
|
/* { dg-do run } */
/* { dg-require-effective-target avx2 } */
/* { dg-options "-O2 -mavx2" } */
#include <string.h>
#include "avx2-check.h"
#define msk0 0xC0
#define msk1 0x01
#define msk2 0xF2
#define msk3 0x03
#define msk4 0x84
#define msk5 0x05
#define msk6 0xE6
#define msk7 0x67
static void
compute_mpsadbw (int *i1, int *i2, int mask, int *r)
{
unsigned char s[4];
int i, j;
int offs1, offs2;
unsigned char *v1 = (char *) i1;
unsigned char *v2 = (char *) i2;
unsigned short *ret = (unsigned short *) r;
memset (ret, 0, 32);
/* Lower part */
offs2 = 4 * (mask & 3);
for (i = 0; i < 4; i++)
s[i] = v2[offs2 + i];
offs1 = 4 * ((mask & 4) >> 2);
for (j = 0; j < 8; j++)
for (i = 0; i < 4; i++)
ret[j] += abs (v1[offs1 + j + i] - s[i]);
/* Higher part */
offs2 = 4 * ((mask >> 3) & 3) + 16;
for (i = 0; i < 4; i++)
s[i] = v2[offs2 + i];
offs1 = 4 * ((mask & 0x20) >> 5) + 16;
for (j = 0; j < 8; j++)
for (i = 0; i < 4; i++)
ret[j + 8] += abs (v1[offs1 + j + i] - s[i]);
}
static void
avx2_test (void)
{
union256i_d val1, val2, val3[8], res[8];
int tmp[8];
unsigned char masks[8];
int i, j;
val1.a[0] = 0x35251505;
val1.a[1] = 0x75655545;
val1.a[2] = 0xB5A59585;
val1.a[3] = 0xF5E5D5C5;
val1.a[4] = 0x35251505;
val1.a[5] = 0x75655545;
val1.a[6] = 0xB5A59585;
val1.a[7] = 0xF5E5D5C5;
val2.a[0] = 0x31211101;
val2.a[1] = 0x71615141;
val2.a[2] = 0xB1A19181;
val2.a[3] = 0xF1E1D1C1;
val2.a[4] = 0x31211101;
val2.a[5] = 0x71615141;
val2.a[6] = 0xB1A19181;
val2.a[7] = 0xF1E1D1C1;
for (i = 0; i < 8; i++)
switch (i % 3)
{
case 1:
val3[i].a[0] = 0xF1E1D1C1;
val3[i].a[1] = 0xB1A19181;
val3[i].a[2] = 0x71615141;
val3[i].a[3] = 0x31211101;
break;
default:
val3[i].x = val2.x;
break;
}
/* Check mpsadbw imm8, ymm, ymm. */
res[0].x = _mm256_mpsadbw_epu8 (val1.x, val2.x, msk0);
res[1].x = _mm256_mpsadbw_epu8 (val1.x, val2.x, msk1);
res[2].x = _mm256_mpsadbw_epu8 (val1.x, val2.x, msk2);
res[3].x = _mm256_mpsadbw_epu8 (val1.x, val2.x, msk3);
res[4].x = _mm256_mpsadbw_epu8 (val1.x, val2.x, msk4);
res[5].x = _mm256_mpsadbw_epu8 (val1.x, val2.x, msk5);
res[6].x = _mm256_mpsadbw_epu8 (val1.x, val2.x, msk6);
res[7].x = _mm256_mpsadbw_epu8 (val1.x, val2.x, msk7);
masks[0] = msk0;
masks[1] = msk1;
masks[2] = msk2;
masks[3] = msk3;
masks[4] = msk4;
masks[5] = msk5;
masks[6] = msk6;
masks[7] = msk7;
for (i = 0; i < 8; i++)
{
compute_mpsadbw (val1.a, val2.a, masks[i], tmp);
if (check_union256i_d (res[i], tmp))
abort ();
}
/* Check mpsadbw imm8, m256, ymm. */
for (i = 0; i < 8; i++)
{
res[i].x = _mm256_mpsadbw_epu8 (val1.x, val3[i].x, msk4);
masks[i] = msk4;
}
for (i = 0; i < 8; i++)
{
compute_mpsadbw (val1.a, val3[i].a, masks[i], tmp);
if (check_union256i_d (res[i], tmp))
abort ();
}
}
|
