1 /* { dg-do compile } */
2 /* { dg-options "-O -msve-vector-bits=256" } */
3 
4 #ifndef BIAS
5 #define BIAS 0
6 #endif
7 
8 #include <stdint.h>
9 
10 typedef int64_t vnx2di __attribute__((vector_size (32)));
11 typedef int32_t vnx4si __attribute__((vector_size (32)));
12 typedef int16_t vnx8hi __attribute__((vector_size (32)));
13 typedef int8_t vnx16qi __attribute__((vector_size (32)));
14 typedef double vnx2df __attribute__((vector_size (32)));
15 typedef float vnx4sf __attribute__((vector_size (32)));
16 typedef _Float16 vnx8hf __attribute__((vector_size (32)));
17 
18 #define MASK_2(X, Y) X, Y + X
19 #define MASK_4(X, Y) MASK_2 (X, Y), MASK_2 (X + 1, Y)
20 #define MASK_8(X, Y) MASK_4 (X, Y), MASK_4 (X + 2, Y)
21 #define MASK_16(X, Y) MASK_8 (X, Y), MASK_8 (X + 4, Y)
22 #define MASK_32(X, Y) MASK_16 (X, Y), MASK_16 (X + 8, Y)
23 
24 #define INDEX_4 vnx2di
25 #define INDEX_8 vnx4si
26 #define INDEX_16 vnx8hi
27 #define INDEX_32 vnx16qi
28 
29 #define PERMUTE(TYPE, NUNITS)					\
30   TYPE permute_##TYPE (TYPE values1, TYPE values2)		\
31   {								\
32     return __builtin_shuffle					\
33       (values1, values2,					\
34        ((INDEX_##NUNITS) { MASK_##NUNITS (BIAS * (NUNITS / 2),	\
35 					  NUNITS) }));		\
36   }
37 
38 #define TEST_ALL(T)				\
39   T (vnx2di, 4)					\
40   T (vnx4si, 8)					\
41   T (vnx8hi, 16)				\
42   T (vnx16qi, 32)				\
43   T (vnx2df, 4)					\
44   T (vnx4sf, 8)					\
45   T (vnx8hf, 16)
46 
47 TEST_ALL (PERMUTE)
48 
49 /* { dg-final { scan-assembler-not {\ttbl\t} } } */
50 
51 /* { dg-final { scan-assembler-times {\tzip1\tz[0-9]+\.d, z[0-9]+\.d, z[0-9]+\.d} 2 } } */
52 /* { dg-final { scan-assembler-times {\tzip1\tz[0-9]+\.s, z[0-9]+\.s, z[0-9]+\.s} 2 } } */
53 /* { dg-final { scan-assembler-times {\tzip1\tz[0-9]+\.h, z[0-9]+\.h, z[0-9]+\.h} 2 } } */
54 /* { dg-final { scan-assembler-times {\tzip1\tz[0-9]+\.b, z[0-9]+\.b, z[0-9]+\.b} 1 } } */
55