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
|
/* { dg-require-effective-target vect_float } */
/* { dg-additional-options "-fdump-tree-optimized-details-blocks" } */
#include <stdarg.h>
#include "tree-vect.h"
#define N 128
float fa[N] __attribute__ ((__aligned__(__BIGGEST_ALIGNMENT__)));
float fb[N+4] __attribute__ ((__aligned__(__BIGGEST_ALIGNMENT__)));
float fc[N] __attribute__ ((__aligned__(__BIGGEST_ALIGNMENT__)));
/* Check handling of accesses for which the "initial condition" -
the expression that represents the first location accessed - is
more involved than just an ssa_name. */
__attribute__ ((noinline)) int
main1 (float * __restrict__ pa, float * __restrict__ pb, float *__restrict__ pc)
{
int i;
float *q = pb + 4;
for (i = 0; i < N; i++)
{
fb[i] = i;
fc[i] = 0.5+i;
asm volatile ("" ::: "memory");
}
for (; i < N+4; i++)
{
fb[i] = i;
asm volatile ("" ::: "memory");
}
for (i = 0; i < N; i++)
{
pa[i] = q[i] * pc[i];
}
#pragma GCC novector
for (i = 0; i < N; i++)
{
if (pa[i] != q[i] * pc[i])
abort ();
}
return 0;
}
int main (void)
{
check_vect ();
main1 (fa, fb, fc);
return 0;
}
/* For targets that don't support misaligned loads we version for the
all three accesses (peeling to align the store will not force the
two loads to be aligned). */
/* { dg-final { scan-tree-dump-times "vectorized 1 loops" 1 "vect" } } */
/* Uncomment when this testcase gets vectorized again:
dg-final { scan-tree-dump-times "Vectorizing an unaligned access" 2 "vect" { xfail vect_no_align } }
dg-final { scan-tree-dump-times "Alignment of access forced using peeling" 1 "vect" { xfail vect_no_align } }
dg-final { scan-tree-dump-times "Alignment of access forced using versioning." 3 "vect" { target vect_no_align } }
*/
/* { dg-final { scan-tree-dump-not "Invalid sum" "optimized" } } */
|
