1 #include <arm_neon.h>
2 #include "arm-neon-ref.h"
3 #include "compute-ref-data.h"
4
5 /* Expected results. */
6 VECT_VAR_DECL(expected,int,8,8) [] = { 0xf7, 0x11, 0x11, 0x11,
7 0x11, 0x11, 0x11, 0x11 };
8 VECT_VAR_DECL(expected,int,16,4) [] = { 0xfff3, 0x22, 0x22, 0x22 };
9 VECT_VAR_DECL(expected,int,32,2) [] = { 0xfffffff1, 0x33 };
10 VECT_VAR_DECL(expected,int,64,1) [] = { 0xfffffffffffffff0 };
11 VECT_VAR_DECL(expected,uint,8,8) [] = { 0xf6, 0xf7, 0x55, 0x55,
12 0x55, 0x55, 0x55, 0x55 };
13 VECT_VAR_DECL(expected,uint,16,4) [] = { 0xfff2, 0xfff3, 0x66, 0x66 };
14 VECT_VAR_DECL(expected,uint,32,2) [] = { 0xfffffff1, 0x77 };
15 VECT_VAR_DECL(expected,uint,64,1) [] = { 0xfffffffffffffff0 };
16 VECT_VAR_DECL(expected,poly,8,8) [] = { 0xf6, 0xf7, 0x55, 0x55,
17 0x55, 0x55, 0x55, 0x55 };
18 VECT_VAR_DECL(expected,poly,16,4) [] = { 0xfff2, 0xfff3, 0x66, 0x66 };
19 #if defined (FP16_SUPPORTED)
20 VECT_VAR_DECL (expected, hfloat, 16, 4) [] = { 0xcb00, 0xca80,
21 0x4b4d, 0x4b4d };
22 #endif
23 VECT_VAR_DECL(expected,hfloat,32,2) [] = { 0xc1700000, 0x42066666 };
24 VECT_VAR_DECL(expected,int,8,16) [] = { 0xfe, 0xff, 0x11, 0x11,
25 0x11, 0x11, 0x11, 0x11,
26 0x11, 0x11, 0x11, 0x11,
27 0x11, 0x11, 0x11, 0x11 };
28 VECT_VAR_DECL(expected,int,16,8) [] = { 0xfff7, 0x22, 0x22, 0x22,
29 0x22, 0x22, 0x22, 0x22 };
30 VECT_VAR_DECL(expected,int,32,4) [] = { 0xfffffff3, 0x33, 0x33, 0x33 };
31 VECT_VAR_DECL(expected,int,64,2) [] = { 0xfffffffffffffff1, 0x44 };
32 VECT_VAR_DECL(expected,uint,8,16) [] = { 0xfc, 0xfd, 0xfe, 0xff,
33 0x55, 0x55, 0x55, 0x55,
34 0x55, 0x55, 0x55, 0x55,
35 0x55, 0x55, 0x55, 0x55 };
36 VECT_VAR_DECL(expected,uint,16,8) [] = { 0xfff6, 0xfff7, 0x66, 0x66,
37 0x66, 0x66, 0x66, 0x66 };
38 VECT_VAR_DECL(expected,uint,32,4) [] = { 0xfffffff3, 0x77, 0x77, 0x77 };
39 VECT_VAR_DECL(expected,uint,64,2) [] = { 0xfffffffffffffff1, 0x88 };
40 VECT_VAR_DECL(expected,poly,8,16) [] = { 0xfc, 0xfd, 0xfe, 0xff,
41 0x55, 0x55, 0x55, 0x55,
42 0x55, 0x55, 0x55, 0x55,
43 0x55, 0x55, 0x55, 0x55 };
44 VECT_VAR_DECL(expected,poly,16,8) [] = { 0xfff6, 0xfff7, 0x66, 0x66,
45 0x66, 0x66, 0x66, 0x66 };
46 #if defined (FP16_SUPPORTED)
47 VECT_VAR_DECL (expected, hfloat, 16, 8) [] = { 0xc880, 0x4b4d,
48 0x4b4d, 0x4b4d,
49 0x4b4d, 0x4b4d,
50 0x4b4d, 0x4b4d };
51 #endif
52 VECT_VAR_DECL(expected,hfloat,32,4) [] = { 0xc1500000, 0x4204cccd,
53 0x4204cccd, 0x4204cccd };
54
55 #define TEST_MSG "VEXT/VEXTQ"
56 void exec_vext (void)
57 {
58 /* vector_res = vext(vector1,vector2,offset), then store the result. */
59 #define TEST_VEXT(Q, T1, T2, W, N, V) \
60 VECT_VAR(vector_res, T1, W, N) = \
61 vext##Q##_##T2##W(VECT_VAR(vector1, T1, W, N), \
62 VECT_VAR(vector2, T1, W, N), \
63 V); \
64 vst1##Q##_##T2##W(VECT_VAR(result, T1, W, N), VECT_VAR(vector_res, T1, W, N))
65
66 DECL_VARIABLE_ALL_VARIANTS(vector1);
67 DECL_VARIABLE_ALL_VARIANTS(vector2);
68 DECL_VARIABLE_ALL_VARIANTS(vector_res);
69
70 clean_results ();
71
72 TEST_MACRO_ALL_VARIANTS_2_5(VLOAD, vector1, buffer);
73 #ifdef FP16_SUPPORTED
74 VLOAD(vector1, buffer, , float, f, 16, 4);
75 VLOAD(vector1, buffer, q, float, f, 16, 8);
76 #endif
77 VLOAD(vector1, buffer, , float, f, 32, 2);
78 VLOAD(vector1, buffer, q, float, f, 32, 4);
79
80 /* Choose arbitrary initialization values. */
81 VDUP(vector2, , int, s, 8, 8, 0x11);
82 VDUP(vector2, , int, s, 16, 4, 0x22);
83 VDUP(vector2, , int, s, 32, 2, 0x33);
84 VDUP(vector2, , int, s, 64, 1, 0x44);
85 VDUP(vector2, , uint, u, 8, 8, 0x55);
86 VDUP(vector2, , uint, u, 16, 4, 0x66);
87 VDUP(vector2, , uint, u, 32, 2, 0x77);
88 VDUP(vector2, , uint, u, 64, 1, 0x88);
89 VDUP(vector2, , poly, p, 8, 8, 0x55);
90 VDUP(vector2, , poly, p, 16, 4, 0x66);
91 #if defined (FP16_SUPPORTED)
92 VDUP (vector2, , float, f, 16, 4, 14.6f); /* 14.6f is 0x4b4d. */
93 #endif
94 VDUP(vector2, , float, f, 32, 2, 33.6f);
95
96 VDUP(vector2, q, int, s, 8, 16, 0x11);
97 VDUP(vector2, q, int, s, 16, 8, 0x22);
98 VDUP(vector2, q, int, s, 32, 4, 0x33);
99 VDUP(vector2, q, int, s, 64, 2, 0x44);
100 VDUP(vector2, q, uint, u, 8, 16, 0x55);
101 VDUP(vector2, q, uint, u, 16, 8, 0x66);
102 VDUP(vector2, q, uint, u, 32, 4, 0x77);
103 VDUP(vector2, q, uint, u, 64, 2, 0x88);
104 VDUP(vector2, q, poly, p, 8, 16, 0x55);
105 VDUP(vector2, q, poly, p, 16, 8, 0x66);
106 #if defined (FP16_SUPPORTED)
107 VDUP (vector2, q, float, f, 16, 8, 14.6f);
108 #endif
109 VDUP(vector2, q, float, f, 32, 4, 33.2f);
110
111 /* Choose arbitrary extract offsets. */
112 TEST_VEXT(, int, s, 8, 8, 7);
113 TEST_VEXT(, int, s, 16, 4, 3);
114 TEST_VEXT(, int, s, 32, 2, 1);
115 TEST_VEXT(, int, s, 64, 1, 0);
116 TEST_VEXT(, uint, u, 8, 8, 6);
117 TEST_VEXT(, uint, u, 16, 4, 2);
118 TEST_VEXT(, uint, u, 32, 2, 1);
119 TEST_VEXT(, uint, u, 64, 1, 0);
120 TEST_VEXT(, poly, p, 8, 8, 6);
121 TEST_VEXT(, poly, p, 16, 4, 2);
122 #if defined (FP16_SUPPORTED)
123 TEST_VEXT(, float, f, 16, 4, 2);
124 #endif
125 TEST_VEXT(, float, f, 32, 2, 1);
126
127 TEST_VEXT(q, int, s, 8, 16, 14);
128 TEST_VEXT(q, int, s, 16, 8, 7);
129 TEST_VEXT(q, int, s, 32, 4, 3);
130 TEST_VEXT(q, int, s, 64, 2, 1);
131 TEST_VEXT(q, uint, u, 8, 16, 12);
132 TEST_VEXT(q, uint, u, 16, 8, 6);
133 TEST_VEXT(q, uint, u, 32, 4, 3);
134 TEST_VEXT(q, uint, u, 64, 2, 1);
135 TEST_VEXT(q, poly, p, 8, 16, 12);
136 TEST_VEXT(q, poly, p, 16, 8, 6);
137 #if defined (FP16_SUPPORTED)
138 TEST_VEXT(q, float, f, 16, 8, 7);
139 #endif
140 TEST_VEXT(q, float, f, 32, 4, 3);
141
142 #if defined (FP16_SUPPORTED)
143 CHECK_RESULTS (TEST_MSG, "");
144 #else
145 CHECK_RESULTS_NO_FP16 (TEST_MSG, "");
146 #endif
147 }
148
149 int main (void)
150 {
151 exec_vext ();
152 return 0;
153 }