2 /* { dg-require-effective-target ssse3 } */
3 /* { dg-options "-O2 -mssse3" } */
5 #include "ssse3-check.h"
6 #include "ssse3-vals.h"
10 /* Test the 64-bit form */
12 ssse3_test_pmulhrsw (int *i1, int *i2, int *r)
14 __m64 t1 = *(__m64 *) i1;
15 __m64 t2 = *(__m64 *) i2;
16 *(__m64 *) r = _mm_mulhrs_pi16 (t1, t2);
20 /* Test the 128-bit form */
22 ssse3_test_pmulhrsw128 (int *i1, int *i2, int *r)
24 /* Assumes incoming pointers are 16-byte aligned */
25 __m128i t1 = *(__m128i *) i1;
26 __m128i t2 = *(__m128i *) i2;
27 *(__m128i *) r = _mm_mulhrs_epi16 (t1, t2);
30 /* Routine to manually compute the results */
32 compute_correct_result (int *i1, int *i2, int *r)
34 short *s1 = (short *) i1;
35 short *s2 = (short *) i2;
36 short *sout = (short *) r;
40 for (i = 0; i < 8; i++)
42 t0 = (((int) s1[i] * (int) s2[i]) >> 14) + 1;
43 sout[i] = (short) (t0 >> 1);
51 int r [4] __attribute__ ((aligned(16)));
55 for (i = 0; i < 256; i += 8)
57 /* Manually compute the result */
58 compute_correct_result (&vals[i + 0], &vals[i + 4], ck);
60 /* Run the 64-bit tests */
61 ssse3_test_pmulhrsw (&vals[i + 0], &vals[i + 4], &r[0]);
62 ssse3_test_pmulhrsw (&vals[i + 2], &vals[i + 6], &r[2]);
63 fail += chk_128 (ck, r);
65 /* Run the 128-bit tests */
66 ssse3_test_pmulhrsw128 (&vals[i + 0], &vals[i + 4], r);
67 fail += chk_128 (ck, r);