| 1 |
/* { dg-do compile } */ |
| 2 |
/* { dg-require-effective-target vect_early_break } */ |
| 3 |
/* { dg-require-effective-target arm_arch_v8a_hard_ok } */ |
| 4 |
/* { dg-options "-O3 -fno-schedule-insns -fno-reorder-blocks -fno-schedule-insns2" } */ |
| 5 |
/* { dg-add-options arm_arch_v8a_hard } */ |
| 6 |
/* { dg-final { check-function-bodies "**" "" "" } } */ |
| 7 |
|
| 8 |
#define N 640 |
| 9 |
int a[N] = {0}; |
| 10 |
int b[N] = {0}; |
| 11 |
|
| 12 |
/* |
| 13 |
** f1: |
| 14 |
** ... |
| 15 |
** vcgt.s32 q[0-9]+, q[0-9]+, #0 |
| 16 |
** vpmax.u32 d[0-9]+, d[0-9]+, d[0-9]+ |
| 17 |
** vpmax.u32 d[0-9]+, d[0-9]+, d[0-9]+ |
| 18 |
** vmov r[0-9]+, s[0-9]+ @ int |
| 19 |
** ( |
| 20 |
** cmp r[0-9]+, #0 |
| 21 |
** b(ne|eq) \.L[0-9]+ |
| 22 |
** | |
| 23 |
** cbn?z r[0-9]+, \.L.+ |
| 24 |
** ) |
| 25 |
** ... |
| 26 |
*/ |
| 27 |
void f1 () |
| 28 |
{ |
| 29 |
for (int i = 0; i < N; i++) |
| 30 |
{ |
| 31 |
b[i] += a[i]; |
| 32 |
if (a[i] > 0) |
| 33 |
break; |
| 34 |
} |
| 35 |
} |
| 36 |
|
| 37 |
/* |
| 38 |
** f2: |
| 39 |
** ... |
| 40 |
** vcge.s32 q[0-9]+, q[0-9]+, #0 |
| 41 |
** vpmax.u32 d[0-9]+, d[0-9]+, d[0-9]+ |
| 42 |
** vpmax.u32 d[0-9]+, d[0-9]+, d[0-9]+ |
| 43 |
** vmov r[0-9]+, s[0-9]+ @ int |
| 44 |
** ( |
| 45 |
** cmp r[0-9]+, #0 |
| 46 |
** b(ne|eq) \.L[0-9]+ |
| 47 |
** | |
| 48 |
** cbn?z r[0-9]+, \.L.+ |
| 49 |
** ) |
| 50 |
** ... |
| 51 |
*/ |
| 52 |
void f2 () |
| 53 |
{ |
| 54 |
for (int i = 0; i < N; i++) |
| 55 |
{ |
| 56 |
b[i] += a[i]; |
| 57 |
if (a[i] >= 0) |
| 58 |
break; |
| 59 |
} |
| 60 |
} |
| 61 |
|
| 62 |
/* |
| 63 |
** f3: |
| 64 |
** ... |
| 65 |
** vceq.i32 q[0-9]+, q[0-9]+, #0 |
| 66 |
** vpmax.u32 d[0-9]+, d[0-9]+, d[0-9]+ |
| 67 |
** vpmax.u32 d[0-9]+, d[0-9]+, d[0-9]+ |
| 68 |
** vmov r[0-9]+, s[0-9]+ @ int |
| 69 |
** ( |
| 70 |
** cmp r[0-9]+, #0 |
| 71 |
** b(ne|eq) \.L[0-9]+ |
| 72 |
** | |
| 73 |
** cbn?z r[0-9]+, \.L.+ |
| 74 |
** ) |
| 75 |
** ... |
| 76 |
*/ |
| 77 |
void f3 () |
| 78 |
{ |
| 79 |
for (int i = 0; i < N; i++) |
| 80 |
{ |
| 81 |
b[i] += a[i]; |
| 82 |
if (a[i] == 0) |
| 83 |
break; |
| 84 |
} |
| 85 |
} |
| 86 |
|
| 87 |
/* |
| 88 |
** f4: |
| 89 |
** ... |
| 90 |
** vceq.i32 q[0-9]+, q[0-9]+, #0 |
| 91 |
** vmvn q[0-9]+, q[0-9]+ |
| 92 |
** vpmax.u32 d[0-9]+, d[0-9]+, d[0-9]+ |
| 93 |
** vpmax.u32 d[0-9]+, d[0-9]+, d[0-9]+ |
| 94 |
** vmov r[0-9]+, s[0-9]+ @ int |
| 95 |
** ( |
| 96 |
** cmp r[0-9]+, #0 |
| 97 |
** b(ne|eq) \.L[0-9]+ |
| 98 |
** | |
| 99 |
** cbn?z r[0-9]+, \.L.+ |
| 100 |
** ) |
| 101 |
** ... |
| 102 |
*/ |
| 103 |
void f4 () |
| 104 |
{ |
| 105 |
for (int i = 0; i < N; i++) |
| 106 |
{ |
| 107 |
b[i] += a[i]; |
| 108 |
if (a[i] != 0) |
| 109 |
break; |
| 110 |
} |
| 111 |
} |
| 112 |
|
| 113 |
/* |
| 114 |
** f5: |
| 115 |
** ... |
| 116 |
** vclt.s32 q[0-9]+, q[0-9]+, #0 |
| 117 |
** vpmax.u32 d[0-9]+, d[0-9]+, d[0-9]+ |
| 118 |
** vpmax.u32 d[0-9]+, d[0-9]+, d[0-9]+ |
| 119 |
** vmov r[0-9]+, s[0-9]+ @ int |
| 120 |
** ( |
| 121 |
** cmp r[0-9]+, #0 |
| 122 |
** b(ne|eq) \.L[0-9]+ |
| 123 |
** | |
| 124 |
** cbn?z r[0-9]+, \.L.+ |
| 125 |
** ) |
| 126 |
** ... |
| 127 |
*/ |
| 128 |
void f5 () |
| 129 |
{ |
| 130 |
for (int i = 0; i < N; i++) |
| 131 |
{ |
| 132 |
b[i] += a[i]; |
| 133 |
if (a[i] < 0) |
| 134 |
break; |
| 135 |
} |
| 136 |
} |
| 137 |
|
| 138 |
/* |
| 139 |
** f6: |
| 140 |
** ... |
| 141 |
** vcle.s32 q[0-9]+, q[0-9]+, #0 |
| 142 |
** vpmax.u32 d[0-9]+, d[0-9]+, d[0-9]+ |
| 143 |
** vpmax.u32 d[0-9]+, d[0-9]+, d[0-9]+ |
| 144 |
** vmov r[0-9]+, s[0-9]+ @ int |
| 145 |
** ( |
| 146 |
** cmp r[0-9]+, #0 |
| 147 |
** b(ne|eq) \.L[0-9]+ |
| 148 |
** | |
| 149 |
** cbn?z r[0-9]+, \.L.+ |
| 150 |
** ) |
| 151 |
** ... |
| 152 |
*/ |
| 153 |
void f6 () |
| 154 |
{ |
| 155 |
for (int i = 0; i < N; i++) |
| 156 |
{ |
| 157 |
b[i] += a[i]; |
| 158 |
if (a[i] <= 0) |
| 159 |
break; |
| 160 |
} |
| 161 |
} |
| 162 |
|