1 /* { dg-do compile } */
2 /* { dg-options "-O2 -ftree-vectorize -ffast-math" } */
3
4 #include <stdint.h>
5
6 #define NUM_ELEMS(TYPE) (320 / sizeof (TYPE))
7
8 #define DEF_MAXMIN(TYPE, NAME, CMP_OP) \
9 void __attribute__ ((noinline, noclone)) \
10 fun_##NAME##_##TYPE (TYPE *restrict r, TYPE *restrict a, \
11 TYPE *restrict b) \
12 { \
13 for (int i = 0; i < NUM_ELEMS (TYPE); i++) \
14 r[i] = a[i] CMP_OP b[i] ? a[i] : b[i]; \
15 }
16
17 #define TEST_ALL(T) \
18 T (int8_t, max, >) \
19 T (int16_t, max, >) \
20 T (int32_t, max, >) \
21 T (int64_t, max, >) \
22 T (uint8_t, max, >) \
23 T (uint16_t, max, >) \
24 T (uint32_t, max, >) \
25 T (uint64_t, max, >) \
26 T (_Float16, max, >) \
27 T (float, max, >) \
28 T (double, max, >) \
29 \
30 T (int8_t, min, <) \
31 T (int16_t, min, <) \
32 T (int32_t, min, <) \
33 T (int64_t, min, <) \
34 T (uint8_t, min, <) \
35 T (uint16_t, min, <) \
36 T (uint32_t, min, <) \
37 T (uint64_t, min, <) \
38 T (_Float16, min, <) \
39 T (float, min, <) \
40 T (double, min, <)
41
42 TEST_ALL (DEF_MAXMIN)
43
44 /* { dg-final { scan-assembler-times {\tsmax\tz[0-9]+\.b, p[0-7]/m, z[0-9]+\.b, z[0-9]+\.b\n} 1 } } */
45 /* { dg-final { scan-assembler-times {\tsmax\tz[0-9]+\.h, p[0-7]/m, z[0-9]+\.h, z[0-9]+\.h\n} 1 } } */
46 /* { dg-final { scan-assembler-times {\tsmax\tz[0-9]+\.s, p[0-7]/m, z[0-9]+\.s, z[0-9]+\.s\n} 1 } } */
47 /* { dg-final { scan-assembler-times {\tsmax\tz[0-9]+\.d, p[0-7]/m, z[0-9]+\.d, z[0-9]+\.d\n} 1 } } */
48
49 /* { dg-final { scan-assembler-times {\tumax\tz[0-9]+\.b, p[0-7]/m, z[0-9]+\.b, z[0-9]+\.b\n} 1 } } */
50 /* { dg-final { scan-assembler-times {\tumax\tz[0-9]+\.h, p[0-7]/m, z[0-9]+\.h, z[0-9]+\.h\n} 1 } } */
51 /* { dg-final { scan-assembler-times {\tumax\tz[0-9]+\.s, p[0-7]/m, z[0-9]+\.s, z[0-9]+\.s\n} 1 } } */
52 /* { dg-final { scan-assembler-times {\tumax\tz[0-9]+\.d, p[0-7]/m, z[0-9]+\.d, z[0-9]+\.d\n} 1 } } */
53
54 /* { dg-final { scan-assembler-times {\tfmaxnm\tz[0-9]+\.h, p[0-7]/m, z[0-9]+\.h, z[0-9]+\.h\n} 1 } } */
55 /* { dg-final { scan-assembler-times {\tfmaxnm\tz[0-9]+\.s, p[0-7]/m, z[0-9]+\.s, z[0-9]+\.s\n} 1 } } */
56 /* { dg-final { scan-assembler-times {\tfmaxnm\tz[0-9]+\.d, p[0-7]/m, z[0-9]+\.d, z[0-9]+\.d\n} 1 } } */
57
58 /* { dg-final { scan-assembler-times {\tsmin\tz[0-9]+\.b, p[0-7]/m, z[0-9]+\.b, z[0-9]+\.b\n} 1 } } */
59 /* { dg-final { scan-assembler-times {\tsmin\tz[0-9]+\.h, p[0-7]/m, z[0-9]+\.h, z[0-9]+\.h\n} 1 } } */
60 /* { dg-final { scan-assembler-times {\tsmin\tz[0-9]+\.s, p[0-7]/m, z[0-9]+\.s, z[0-9]+\.s\n} 1 } } */
61 /* { dg-final { scan-assembler-times {\tsmin\tz[0-9]+\.d, p[0-7]/m, z[0-9]+\.d, z[0-9]+\.d\n} 1 } } */
62
63 /* { dg-final { scan-assembler-times {\tumin\tz[0-9]+\.b, p[0-7]/m, z[0-9]+\.b, z[0-9]+\.b\n} 1 } } */
64 /* { dg-final { scan-assembler-times {\tumin\tz[0-9]+\.h, p[0-7]/m, z[0-9]+\.h, z[0-9]+\.h\n} 1 } } */
65 /* { dg-final { scan-assembler-times {\tumin\tz[0-9]+\.s, p[0-7]/m, z[0-9]+\.s, z[0-9]+\.s\n} 1 } } */
66 /* { dg-final { scan-assembler-times {\tumin\tz[0-9]+\.d, p[0-7]/m, z[0-9]+\.d, z[0-9]+\.d\n} 1 } } */
67
68 /* { dg-final { scan-assembler-times {\tfminnm\tz[0-9]+\.h, p[0-7]/m, z[0-9]+\.h, z[0-9]+\.h\n} 1 } } */
69 /* { dg-final { scan-assembler-times {\tfminnm\tz[0-9]+\.s, p[0-7]/m, z[0-9]+\.s, z[0-9]+\.s\n} 1 } } */
70 /* { dg-final { scan-assembler-times {\tfminnm\tz[0-9]+\.d, p[0-7]/m, z[0-9]+\.d, z[0-9]+\.d\n} 1 } } */