1 /* { dg-do run } */
2 /* { dg-require-effective-target sse3 } */
3 /* { dg-options "-O2 -msse3 -mfpmath=sse" } */
4 
5 #ifndef CHECK_H
6 #define CHECK_H "sse3-check.h"
7 #endif
8 
9 #ifndef TEST
10 #define TEST sse3_test
11 #endif
12 
13 #include CHECK_H
14 
15 #include <pmmintrin.h>
16 
17 static void
sse3_test_movsldup_reg(float * i1,float * r)18 sse3_test_movsldup_reg (float *i1, float *r)
19 {
20   __m128 t1 = _mm_loadu_ps (i1);
21   __m128 t2 = _mm_moveldup_ps (t1);
22 
23   _mm_storeu_ps (r, t2);
24 }
25 
26 static void
sse3_test_movsldup_reg_subsume(float * i1,float * r)27 sse3_test_movsldup_reg_subsume (float *i1, float *r)
28 {
29   __m128 t1 = _mm_load_ps (i1);
30   __m128 t2 = _mm_moveldup_ps (t1);
31 
32   _mm_storeu_ps (r, t2);
33 }
34 
35 static int
chk_ps(float * v1,float * v2)36 chk_ps (float *v1, float *v2)
37 {
38   int i;
39   int n_fails = 0;
40 
41   for (i = 0; i < 4; i++)
42     if (v1[i] != v2[i])
43       n_fails += 1;
44 
45   return n_fails;
46 }
47 
48 static float p1[4] __attribute__ ((aligned(16)));
49 static float p2[4];
50 static float ck[4];
51 
52 static float vals[80] =
53   {
54     100.0,  200.0, 300.0, 400.0, 5.0, -1.0, .345, -21.5,
55     1100.0, 0.235, 321.3, 53.40, 0.3, 10.0, 42.0, 32.52,
56     32.6,   123.3, 1.234, 2.156, 0.1, 3.25, 4.75, 32.44,
57     12.16,  52.34, 64.12, 71.13, -.1, 2.30, 5.12, 3.785,
58     541.3,  321.4, 231.4, 531.4, 71., 321., 231., -531.,
59     23.45,  23.45, 23.45, 23.45, 23.45, 23.45, 23.45, 23.45,
60     23.45,  -1.43, -6.74, 6.345, -20.1, -20.1, -40.1, -40.1,
61     1.234,  2.345, 3.456, 4.567, 5.678, 6.789, 7.891, 8.912,
62     -9.32,  -8.41, -7.50, -6.59, -5.68, -4.77, -3.86, -2.95,
63     9.32,  8.41, 7.50, 6.59, -5.68, -4.77, -3.86, -2.95
64   };
65 
66 static void
TEST(void)67 TEST (void)
68 {
69   int i;
70   int fail = 0;
71 
72   for (i = 0; i < 80; i += 2)
73     {
74       p1[0] = vals[i+0];
75       p1[1] = 0.0;
76       p1[2] = vals[i+1];
77       p1[3] = 1.0;
78 
79       ck[0] = p1[0];
80       ck[1] = p1[0];
81       ck[2] = p1[2];
82       ck[3] = p1[2];
83 
84       sse3_test_movsldup_reg (p1, p2);
85 
86       fail += chk_ps (ck, p2);
87 
88       sse3_test_movsldup_reg_subsume (p1, p2);
89 
90       fail += chk_ps (ck, p2);
91     }
92 
93   if (fail != 0)
94     abort ();
95 }
96