1 #include <arm_neon.h>
2 #include "arm-neon-ref.h"
3 #include "compute-ref-data.h"
4
5 /* Expected results for vst2, chunk 0. */
6 VECT_VAR_DECL(expected_st2_0,int,8,8) [] = { 0xf0, 0xf1, 0x0, 0x0,
7 0x0, 0x0, 0x0, 0x0 };
8 VECT_VAR_DECL(expected_st2_0,int,16,4) [] = { 0xfff0, 0xfff1, 0x0, 0x0 };
9 VECT_VAR_DECL(expected_st2_0,int,32,2) [] = { 0xfffffff0, 0xfffffff1 };
10 VECT_VAR_DECL(expected_st2_0,uint,8,8) [] = { 0xf0, 0xf1, 0x0, 0x0,
11 0x0, 0x0, 0x0, 0x0 };
12 VECT_VAR_DECL(expected_st2_0,uint,16,4) [] = { 0xfff0, 0xfff1, 0x0, 0x0 };
13 VECT_VAR_DECL(expected_st2_0,uint,32,2) [] = { 0xfffffff0, 0xfffffff1 };
14 VECT_VAR_DECL(expected_st2_0,poly,8,8) [] = { 0xf0, 0xf1, 0x0, 0x0,
15 0x0, 0x0, 0x0, 0x0 };
16 VECT_VAR_DECL(expected_st2_0,poly,16,4) [] = { 0xfff0, 0xfff1, 0x0, 0x0 };
17 VECT_VAR_DECL(expected_st2_0,hfloat,16,4) [] = { 0xcc00, 0xcb80, 0x0, 0x0 };
18 VECT_VAR_DECL(expected_st2_0,hfloat,32,2) [] = { 0xc1800000, 0xc1700000 };
19 VECT_VAR_DECL(expected_st2_0,int,16,8) [] = { 0xfff0, 0xfff1, 0x0, 0x0,
20 0x0, 0x0, 0x0, 0x0 };
21 VECT_VAR_DECL(expected_st2_0,int,32,4) [] = { 0xfffffff0, 0xfffffff1, 0x0, 0x0 };
22 VECT_VAR_DECL(expected_st2_0,uint,16,8) [] = { 0xfff0, 0xfff1, 0x0, 0x0,
23 0x0, 0x0, 0x0, 0x0 };
24 VECT_VAR_DECL(expected_st2_0,uint,32,4) [] = { 0xfffffff0, 0xfffffff1,
25 0x0, 0x0 };
26 VECT_VAR_DECL(expected_st2_0,poly,16,8) [] = { 0xfff0, 0xfff1, 0x0, 0x0,
27 0x0, 0x0, 0x0, 0x0 };
28 VECT_VAR_DECL(expected_st2_0,hfloat,16,8) [] = { 0xcc00, 0xcb80, 0x0, 0x0,
29 0x0, 0x0, 0x0, 0x0 };
30 VECT_VAR_DECL(expected_st2_0,hfloat,32,4) [] = { 0xc1800000, 0xc1700000,
31 0x0, 0x0 };
32
33 /* Expected results for vst2, chunk 1. */
34 VECT_VAR_DECL(expected_st2_1,int,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
35 0x0, 0x0, 0x0, 0x0 };
36 VECT_VAR_DECL(expected_st2_1,int,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
37 VECT_VAR_DECL(expected_st2_1,int,32,2) [] = { 0x0, 0x0 };
38 VECT_VAR_DECL(expected_st2_1,uint,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
39 0x0, 0x0, 0x0, 0x0 };
40 VECT_VAR_DECL(expected_st2_1,uint,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
41 VECT_VAR_DECL(expected_st2_1,uint,32,2) [] = { 0x0, 0x0 };
42 VECT_VAR_DECL(expected_st2_1,poly,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
43 0x0, 0x0, 0x0, 0x0 };
44 VECT_VAR_DECL(expected_st2_1,poly,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
45 VECT_VAR_DECL(expected_st2_1,hfloat,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
46 VECT_VAR_DECL(expected_st2_1,hfloat,32,2) [] = { 0x0, 0x0 };
47 VECT_VAR_DECL(expected_st2_1,int,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
48 0x0, 0x0, 0x0, 0x0 };
49 VECT_VAR_DECL(expected_st2_1,int,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
50 VECT_VAR_DECL(expected_st2_1,uint,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
51 0x0, 0x0, 0x0, 0x0 };
52 VECT_VAR_DECL(expected_st2_1,uint,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
53 VECT_VAR_DECL(expected_st2_1,poly,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
54 0x0, 0x0, 0x0, 0x0 };
55 VECT_VAR_DECL(expected_st2_1,hfloat,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
56 0x0, 0x0, 0x0, 0x0 };
57 VECT_VAR_DECL(expected_st2_1,hfloat,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
58
59 /* Expected results for vst3, chunk 0. */
60 VECT_VAR_DECL(expected_st3_0,int,8,8) [] = { 0xf0, 0xf1, 0xf2, 0x0,
61 0x0, 0x0, 0x0, 0x0 };
62 VECT_VAR_DECL(expected_st3_0,int,16,4) [] = { 0xfff0, 0xfff1, 0xfff2, 0x0 };
63 VECT_VAR_DECL(expected_st3_0,int,32,2) [] = { 0xfffffff0, 0xfffffff1 };
64 VECT_VAR_DECL(expected_st3_0,uint,8,8) [] = { 0xf0, 0xf1, 0xf2, 0x0,
65 0x0, 0x0, 0x0, 0x0 };
66 VECT_VAR_DECL(expected_st3_0,uint,16,4) [] = { 0xfff0, 0xfff1, 0xfff2, 0x0 };
67 VECT_VAR_DECL(expected_st3_0,uint,32,2) [] = { 0xfffffff0, 0xfffffff1 };
68 VECT_VAR_DECL(expected_st3_0,poly,8,8) [] = { 0xf0, 0xf1, 0xf2, 0x0,
69 0x0, 0x0, 0x0, 0x0 };
70 VECT_VAR_DECL(expected_st3_0,poly,16,4) [] = { 0xfff0, 0xfff1, 0xfff2, 0x0 };
71 VECT_VAR_DECL(expected_st3_0,hfloat,16,4) [] = { 0xcc00, 0xcb80, 0xcb00, 0x0 };
72 VECT_VAR_DECL(expected_st3_0,hfloat,32,2) [] = { 0xc1800000, 0xc1700000 };
73 VECT_VAR_DECL(expected_st3_0,int,16,8) [] = { 0xfff0, 0xfff1, 0xfff2, 0x0,
74 0x0, 0x0, 0x0, 0x0 };
75 VECT_VAR_DECL(expected_st3_0,int,32,4) [] = { 0xfffffff0, 0xfffffff1,
76 0xfffffff2, 0x0 };
77 VECT_VAR_DECL(expected_st3_0,uint,16,8) [] = { 0xfff0, 0xfff1, 0xfff2, 0x0,
78 0x0, 0x0, 0x0, 0x0 };
79 VECT_VAR_DECL(expected_st3_0,uint,32,4) [] = { 0xfffffff0, 0xfffffff1,
80 0xfffffff2, 0x0 };
81 VECT_VAR_DECL(expected_st3_0,poly,16,8) [] = { 0xfff0, 0xfff1, 0xfff2, 0x0,
82 0x0, 0x0, 0x0, 0x0 };
83 VECT_VAR_DECL(expected_st3_0,hfloat,16,8) [] = { 0xcc00, 0xcb80, 0xcb00, 0x0,
84 0x0, 0x0, 0x0, 0x0 };
85 VECT_VAR_DECL(expected_st3_0,hfloat,32,4) [] = { 0xc1800000, 0xc1700000,
86 0xc1600000, 0x0 };
87
88 /* Expected results for vst3, chunk 1. */
89 VECT_VAR_DECL(expected_st3_1,int,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
90 0x0, 0x0, 0x0, 0x0 };
91 VECT_VAR_DECL(expected_st3_1,int,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
92 VECT_VAR_DECL(expected_st3_1,int,32,2) [] = { 0xfffffff2, 0x0 };
93 VECT_VAR_DECL(expected_st3_1,uint,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
94 0x0, 0x0, 0x0, 0x0 };
95 VECT_VAR_DECL(expected_st3_1,uint,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
96 VECT_VAR_DECL(expected_st3_1,uint,32,2) [] = { 0xfffffff2, 0x0 };
97 VECT_VAR_DECL(expected_st3_1,poly,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
98 0x0, 0x0, 0x0, 0x0 };
99 VECT_VAR_DECL(expected_st3_1,poly,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
100 VECT_VAR_DECL(expected_st3_1,hfloat,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
101 VECT_VAR_DECL(expected_st3_1,hfloat,32,2) [] = { 0xc1600000, 0x0 };
102 VECT_VAR_DECL(expected_st3_1,int,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
103 0x0, 0x0, 0x0, 0x0 };
104 VECT_VAR_DECL(expected_st3_1,int,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
105 VECT_VAR_DECL(expected_st3_1,uint,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
106 0x0, 0x0, 0x0, 0x0 };
107 VECT_VAR_DECL(expected_st3_1,uint,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
108 VECT_VAR_DECL(expected_st3_1,poly,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
109 0x0, 0x0, 0x0, 0x0 };
110 VECT_VAR_DECL(expected_st3_1,hfloat,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
111 0x0, 0x0, 0x0, 0x0 };
112 VECT_VAR_DECL(expected_st3_1,hfloat,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
113
114 /* Expected results for vst3, chunk 2. */
115 VECT_VAR_DECL(expected_st3_2,int,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
116 0x0, 0x0, 0x0, 0x0 };
117 VECT_VAR_DECL(expected_st3_2,int,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
118 VECT_VAR_DECL(expected_st3_2,int,32,2) [] = { 0x0, 0x0 };
119 VECT_VAR_DECL(expected_st3_2,uint,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
120 0x0, 0x0, 0x0, 0x0 };
121 VECT_VAR_DECL(expected_st3_2,uint,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
122 VECT_VAR_DECL(expected_st3_2,uint,32,2) [] = { 0x0, 0x0 };
123 VECT_VAR_DECL(expected_st3_2,poly,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
124 0x0, 0x0, 0x0, 0x0 };
125 VECT_VAR_DECL(expected_st3_2,poly,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
126 VECT_VAR_DECL(expected_st3_2,hfloat,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
127 VECT_VAR_DECL(expected_st3_2,hfloat,32,2) [] = { 0x0, 0x0 };
128 VECT_VAR_DECL(expected_st3_2,int,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
129 0x0, 0x0, 0x0, 0x0 };
130 VECT_VAR_DECL(expected_st3_2,int,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
131 VECT_VAR_DECL(expected_st3_2,uint,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
132 0x0, 0x0, 0x0, 0x0 };
133 VECT_VAR_DECL(expected_st3_2,uint,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
134 VECT_VAR_DECL(expected_st3_2,poly,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
135 0x0, 0x0, 0x0, 0x0 };
136 VECT_VAR_DECL(expected_st3_2,hfloat,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
137 0x0, 0x0, 0x0, 0x0 };
138 VECT_VAR_DECL(expected_st3_2,hfloat,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
139
140 /* Expected results for vst4, chunk 0. */
141 VECT_VAR_DECL(expected_st4_0,int,8,8) [] = { 0xf0, 0xf1, 0xf2, 0xf3,
142 0x0, 0x0, 0x0, 0x0 };
143 VECT_VAR_DECL(expected_st4_0,int,16,4) [] = { 0xfff0, 0xfff1, 0xfff2, 0xfff3 };
144 VECT_VAR_DECL(expected_st4_0,int,32,2) [] = { 0xfffffff0, 0xfffffff1 };
145 VECT_VAR_DECL(expected_st4_0,uint,8,8) [] = { 0xf0, 0xf1, 0xf2, 0xf3,
146 0x0, 0x0, 0x0, 0x0 };
147 VECT_VAR_DECL(expected_st4_0,uint,16,4) [] = { 0xfff0, 0xfff1, 0xfff2, 0xfff3 };
148 VECT_VAR_DECL(expected_st4_0,uint,32,2) [] = { 0xfffffff0, 0xfffffff1 };
149 VECT_VAR_DECL(expected_st4_0,poly,8,8) [] = { 0xf0, 0xf1, 0xf2, 0xf3,
150 0x0, 0x0, 0x0, 0x0 };
151 VECT_VAR_DECL(expected_st4_0,poly,16,4) [] = { 0xfff0, 0xfff1, 0xfff2, 0xfff3 };
152 VECT_VAR_DECL(expected_st4_0,hfloat,16,4) [] = { 0xcc00, 0xcb80, 0xcb00, 0xca80 };
153 VECT_VAR_DECL(expected_st4_0,hfloat,32,2) [] = { 0xc1800000, 0xc1700000 };
154 VECT_VAR_DECL(expected_st4_0,int,16,8) [] = { 0xfff0, 0xfff1, 0xfff2, 0xfff3,
155 0x0, 0x0, 0x0, 0x0 };
156 VECT_VAR_DECL(expected_st4_0,int,32,4) [] = { 0xfffffff0, 0xfffffff1,
157 0xfffffff2, 0xfffffff3 };
158 VECT_VAR_DECL(expected_st4_0,uint,16,8) [] = { 0xfff0, 0xfff1, 0xfff2, 0xfff3,
159 0x0, 0x0, 0x0, 0x0 };
160 VECT_VAR_DECL(expected_st4_0,uint,32,4) [] = { 0xfffffff0, 0xfffffff1,
161 0xfffffff2, 0xfffffff3 };
162 VECT_VAR_DECL(expected_st4_0,poly,16,8) [] = { 0xfff0, 0xfff1, 0xfff2, 0xfff3,
163 0x0, 0x0, 0x0, 0x0 };
164 VECT_VAR_DECL(expected_st4_0,hfloat,16,8) [] = { 0xcc00, 0xcb80, 0xcb00, 0xca80,
165 0x0, 0x0, 0x0, 0x0 };
166 VECT_VAR_DECL(expected_st4_0,hfloat,32,4) [] = { 0xc1800000, 0xc1700000,
167 0xc1600000, 0xc1500000 };
168
169 /* Expected results for vst4, chunk 1. */
170 VECT_VAR_DECL(expected_st4_1,int,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
171 0x0, 0x0, 0x0, 0x0 };
172 VECT_VAR_DECL(expected_st4_1,int,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
173 VECT_VAR_DECL(expected_st4_1,int,32,2) [] = { 0xfffffff2, 0xfffffff3 };
174 VECT_VAR_DECL(expected_st4_1,uint,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
175 0x0, 0x0, 0x0, 0x0 };
176 VECT_VAR_DECL(expected_st4_1,uint,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
177 VECT_VAR_DECL(expected_st4_1,uint,32,2) [] = { 0xfffffff2, 0xfffffff3 };
178 VECT_VAR_DECL(expected_st4_1,poly,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
179 0x0, 0x0, 0x0, 0x0 };
180 VECT_VAR_DECL(expected_st4_1,poly,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
181 VECT_VAR_DECL(expected_st4_1,hfloat,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
182 VECT_VAR_DECL(expected_st4_1,hfloat,32,2) [] = { 0xc1600000, 0xc1500000 };
183 VECT_VAR_DECL(expected_st4_1,int,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
184 0x0, 0x0, 0x0, 0x0 };
185 VECT_VAR_DECL(expected_st4_1,int,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
186 VECT_VAR_DECL(expected_st4_1,uint,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
187 0x0, 0x0, 0x0, 0x0 };
188 VECT_VAR_DECL(expected_st4_1,uint,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
189 VECT_VAR_DECL(expected_st4_1,poly,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
190 0x0, 0x0, 0x0, 0x0 };
191 VECT_VAR_DECL(expected_st4_1,hfloat,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
192 0x0, 0x0, 0x0, 0x0 };
193 VECT_VAR_DECL(expected_st4_1,hfloat,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
194
195 /* Expected results for vst4, chunk 2. */
196 VECT_VAR_DECL(expected_st4_2,int,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
197 0x0, 0x0, 0x0, 0x0 };
198 VECT_VAR_DECL(expected_st4_2,int,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
199 VECT_VAR_DECL(expected_st4_2,int,32,2) [] = { 0x0, 0x0 };
200 VECT_VAR_DECL(expected_st4_2,uint,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
201 0x0, 0x0, 0x0, 0x0 };
202 VECT_VAR_DECL(expected_st4_2,uint,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
203 VECT_VAR_DECL(expected_st4_2,uint,32,2) [] = { 0x0, 0x0 };
204 VECT_VAR_DECL(expected_st4_2,poly,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
205 0x0, 0x0, 0x0, 0x0 };
206 VECT_VAR_DECL(expected_st4_2,poly,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
207 VECT_VAR_DECL(expected_st4_2,hfloat,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
208 VECT_VAR_DECL(expected_st4_2,hfloat,32,2) [] = { 0x0, 0x0 };
209 VECT_VAR_DECL(expected_st4_2,int,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
210 0x0, 0x0, 0x0, 0x0 };
211 VECT_VAR_DECL(expected_st4_2,int,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
212 VECT_VAR_DECL(expected_st4_2,uint,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
213 0x0, 0x0, 0x0, 0x0 };
214 VECT_VAR_DECL(expected_st4_2,uint,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
215 VECT_VAR_DECL(expected_st4_2,poly,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
216 0x0, 0x0, 0x0, 0x0 };
217 VECT_VAR_DECL(expected_st4_2,hfloat,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
218 0x0, 0x0, 0x0, 0x0 };
219 VECT_VAR_DECL(expected_st4_2,hfloat,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
220
221 /* Expected results for vst4, chunk 3. */
222 VECT_VAR_DECL(expected_st4_3,int,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
223 0x0, 0x0, 0x0, 0x0 };
224 VECT_VAR_DECL(expected_st4_3,int,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
225 VECT_VAR_DECL(expected_st4_3,int,32,2) [] = { 0x0, 0x0 };
226 VECT_VAR_DECL(expected_st4_3,uint,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
227 0x0, 0x0, 0x0, 0x0 };
228 VECT_VAR_DECL(expected_st4_3,uint,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
229 VECT_VAR_DECL(expected_st4_3,uint,32,2) [] = { 0x0, 0x0 };
230 VECT_VAR_DECL(expected_st4_3,poly,8,8) [] = { 0x0, 0x0, 0x0, 0x0,
231 0x0, 0x0, 0x0, 0x0 };
232 VECT_VAR_DECL(expected_st4_3,poly,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
233 VECT_VAR_DECL(expected_st4_3,hfloat,16,4) [] = { 0x0, 0x0, 0x0, 0x0 };
234 VECT_VAR_DECL(expected_st4_3,hfloat,32,2) [] = { 0x0, 0x0 };
235 VECT_VAR_DECL(expected_st4_3,int,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
236 0x0, 0x0, 0x0, 0x0 };
237 VECT_VAR_DECL(expected_st4_3,int,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
238 VECT_VAR_DECL(expected_st4_3,uint,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
239 0x0, 0x0, 0x0, 0x0 };
240 VECT_VAR_DECL(expected_st4_3,uint,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
241 VECT_VAR_DECL(expected_st4_3,poly,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
242 0x0, 0x0, 0x0, 0x0 };
243 VECT_VAR_DECL(expected_st4_3,hfloat,16,8) [] = { 0x0, 0x0, 0x0, 0x0,
244 0x0, 0x0, 0x0, 0x0 };
245 VECT_VAR_DECL(expected_st4_3,hfloat,32,4) [] = { 0x0, 0x0, 0x0, 0x0 };
246
247 /* Declare additional input buffers as needed. */
248 /* Input buffers for vld2_lane. */
249 VECT_VAR_DECL_INIT(buffer_vld2_lane, int, 8, 2);
250 VECT_VAR_DECL_INIT(buffer_vld2_lane, int, 16, 2);
251 VECT_VAR_DECL_INIT(buffer_vld2_lane, int, 32, 2);
252 VECT_VAR_DECL_INIT(buffer_vld2_lane, int, 64, 2);
253 VECT_VAR_DECL_INIT(buffer_vld2_lane, uint, 8, 2);
254 VECT_VAR_DECL_INIT(buffer_vld2_lane, uint, 16, 2);
255 VECT_VAR_DECL_INIT(buffer_vld2_lane, uint, 32, 2);
256 VECT_VAR_DECL_INIT(buffer_vld2_lane, uint, 64, 2);
257 VECT_VAR_DECL_INIT(buffer_vld2_lane, poly, 8, 2);
258 VECT_VAR_DECL_INIT(buffer_vld2_lane, poly, 16, 2);
259 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
260 VECT_VAR_DECL_INIT(buffer_vld2_lane, float, 16, 2);
261 #endif
262 VECT_VAR_DECL_INIT(buffer_vld2_lane, float, 32, 2);
263
264 /* Input buffers for vld3_lane. */
265 VECT_VAR_DECL_INIT(buffer_vld3_lane, int, 8, 3);
266 VECT_VAR_DECL_INIT(buffer_vld3_lane, int, 16, 3);
267 VECT_VAR_DECL_INIT(buffer_vld3_lane, int, 32, 3);
268 VECT_VAR_DECL_INIT(buffer_vld3_lane, int, 64, 3);
269 VECT_VAR_DECL_INIT(buffer_vld3_lane, uint, 8, 3);
270 VECT_VAR_DECL_INIT(buffer_vld3_lane, uint, 16, 3);
271 VECT_VAR_DECL_INIT(buffer_vld3_lane, uint, 32, 3);
272 VECT_VAR_DECL_INIT(buffer_vld3_lane, uint, 64, 3);
273 VECT_VAR_DECL_INIT(buffer_vld3_lane, poly, 8, 3);
274 VECT_VAR_DECL_INIT(buffer_vld3_lane, poly, 16, 3);
275 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
276 VECT_VAR_DECL_INIT(buffer_vld3_lane, float, 16, 3);
277 #endif
278 VECT_VAR_DECL_INIT(buffer_vld3_lane, float, 32, 3);
279
280 /* Input buffers for vld4_lane. */
281 VECT_VAR_DECL_INIT(buffer_vld4_lane, int, 8, 4);
282 VECT_VAR_DECL_INIT(buffer_vld4_lane, int, 16, 4);
283 VECT_VAR_DECL_INIT(buffer_vld4_lane, int, 32, 4);
284 VECT_VAR_DECL_INIT(buffer_vld4_lane, int, 64, 4);
285 VECT_VAR_DECL_INIT(buffer_vld4_lane, uint, 8, 4);
286 VECT_VAR_DECL_INIT(buffer_vld4_lane, uint, 16, 4);
287 VECT_VAR_DECL_INIT(buffer_vld4_lane, uint, 32, 4);
288 VECT_VAR_DECL_INIT(buffer_vld4_lane, uint, 64, 4);
289 VECT_VAR_DECL_INIT(buffer_vld4_lane, poly, 8, 4);
290 VECT_VAR_DECL_INIT(buffer_vld4_lane, poly, 16, 4);
291 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
292 VECT_VAR_DECL_INIT(buffer_vld4_lane, float, 16, 4);
293 #endif
294 VECT_VAR_DECL_INIT(buffer_vld4_lane, float, 32, 4);
295
exec_vstX_lane(void)296 void exec_vstX_lane (void)
297 {
298 /* In this case, input variables are arrays of vectors. */
299 #define DECL_VSTX_LANE(T1, W, N, X) \
300 VECT_ARRAY_TYPE(T1, W, N, X) VECT_ARRAY_VAR(vector, T1, W, N, X); \
301 VECT_ARRAY_TYPE(T1, W, N, X) VECT_ARRAY_VAR(vector_src, T1, W, N, X); \
302 VECT_VAR_DECL(result_bis_##X, T1, W, N)[X * N]
303
304 /* We need to use a temporary result buffer (result_bis), because
305 the one used for other tests is not large enough. A subset of the
306 result data is moved from result_bis to result, and it is this
307 subset which is used to check the actual behavior. The next
308 macro enables to move another chunk of data from result_bis to
309 result. */
310 /* We also use another extra input buffer (buffer_src), which we
311 fill with 0xAA, and which it used to load a vector from which we
312 read a given lane. */
313 #define TEST_VSTX_LANE(Q, T1, T2, W, N, X, L) \
314 memset (VECT_VAR(buffer_src, T1, W, N), 0xAA, \
315 sizeof(VECT_VAR(buffer_src, T1, W, N))); \
316 memset (VECT_VAR(result_bis_##X, T1, W, N), 0, \
317 sizeof(VECT_VAR(result_bis_##X, T1, W, N))); \
318 \
319 VECT_ARRAY_VAR(vector_src, T1, W, N, X) = \
320 vld##X##Q##_##T2##W(VECT_VAR(buffer_src, T1, W, N)); \
321 \
322 VECT_ARRAY_VAR(vector, T1, W, N, X) = \
323 /* Use dedicated init buffer, of size X. */ \
324 vld##X##Q##_lane_##T2##W(VECT_VAR(buffer_vld##X##_lane, T1, W, X), \
325 VECT_ARRAY_VAR(vector_src, T1, W, N, X), \
326 L); \
327 vst##X##Q##_lane_##T2##W(VECT_VAR(result_bis_##X, T1, W, N), \
328 VECT_ARRAY_VAR(vector, T1, W, N, X), \
329 L); \
330 memcpy(VECT_VAR(result, T1, W, N), VECT_VAR(result_bis_##X, T1, W, N), \
331 sizeof(VECT_VAR(result, T1, W, N)));
332
333 /* Overwrite "result" with the contents of "result_bis"[Y]. */
334 #define TEST_EXTRA_CHUNK(T1, W, N, X, Y) \
335 memcpy(VECT_VAR(result, T1, W, N), \
336 &(VECT_VAR(result_bis_##X, T1, W, N)[Y*N]), \
337 sizeof(VECT_VAR(result, T1, W, N)));
338
339 /* We need all variants in 64 bits, but there is no 64x2 variant,
340 nor 128 bits vectors of int8/uint8/poly8. */
341 #define DECL_ALL_VSTX_LANE_NO_FP16(X) \
342 DECL_VSTX_LANE(int, 8, 8, X); \
343 DECL_VSTX_LANE(int, 16, 4, X); \
344 DECL_VSTX_LANE(int, 32, 2, X); \
345 DECL_VSTX_LANE(uint, 8, 8, X); \
346 DECL_VSTX_LANE(uint, 16, 4, X); \
347 DECL_VSTX_LANE(uint, 32, 2, X); \
348 DECL_VSTX_LANE(poly, 8, 8, X); \
349 DECL_VSTX_LANE(poly, 16, 4, X); \
350 DECL_VSTX_LANE(float, 32, 2, X); \
351 DECL_VSTX_LANE(int, 16, 8, X); \
352 DECL_VSTX_LANE(int, 32, 4, X); \
353 DECL_VSTX_LANE(uint, 16, 8, X); \
354 DECL_VSTX_LANE(uint, 32, 4, X); \
355 DECL_VSTX_LANE(poly, 16, 8, X); \
356 DECL_VSTX_LANE(float, 32, 4, X)
357
358 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
359 #define DECL_ALL_VSTX_LANE(X) \
360 DECL_ALL_VSTX_LANE_NO_FP16(X); \
361 DECL_VSTX_LANE(float, 16, 4, X); \
362 DECL_VSTX_LANE(float, 16, 8, X)
363 #else
364 #define DECL_ALL_VSTX_LANE(X) DECL_ALL_VSTX_LANE_NO_FP16(X)
365 #endif
366
367 #define DUMMY_ARRAY(V, T, W, N, L) VECT_VAR_DECL(V,T,W,N)[N*L]
368
369 /* Use the same lanes regardless of the size of the array (X), for
370 simplicity. */
371 #define TEST_ALL_VSTX_LANE_NO_FP16(X) \
372 TEST_VSTX_LANE(, int, s, 8, 8, X, 7); \
373 TEST_VSTX_LANE(, int, s, 16, 4, X, 2); \
374 TEST_VSTX_LANE(, int, s, 32, 2, X, 0); \
375 TEST_VSTX_LANE(, float, f, 32, 2, X, 0); \
376 TEST_VSTX_LANE(, uint, u, 8, 8, X, 4); \
377 TEST_VSTX_LANE(, uint, u, 16, 4, X, 3); \
378 TEST_VSTX_LANE(, uint, u, 32, 2, X, 1); \
379 TEST_VSTX_LANE(, poly, p, 8, 8, X, 4); \
380 TEST_VSTX_LANE(, poly, p, 16, 4, X, 3); \
381 TEST_VSTX_LANE(q, int, s, 16, 8, X, 6); \
382 TEST_VSTX_LANE(q, int, s, 32, 4, X, 2); \
383 TEST_VSTX_LANE(q, uint, u, 16, 8, X, 5); \
384 TEST_VSTX_LANE(q, uint, u, 32, 4, X, 0); \
385 TEST_VSTX_LANE(q, poly, p, 16, 8, X, 5); \
386 TEST_VSTX_LANE(q, float, f, 32, 4, X, 2)
387
388 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
389 #define TEST_ALL_VSTX_LANE(X) \
390 TEST_ALL_VSTX_LANE_NO_FP16(X); \
391 TEST_VSTX_LANE(, float, f, 16, 4, X, 2); \
392 TEST_VSTX_LANE(q, float, f, 16, 8, X, 6)
393 #else
394 #define TEST_ALL_VSTX_LANE(X) TEST_ALL_VSTX_LANE_NO_FP16(X)
395 #endif
396
397 #define TEST_ALL_EXTRA_CHUNKS_NO_FP16(X, Y) \
398 TEST_EXTRA_CHUNK(int, 8, 8, X, Y); \
399 TEST_EXTRA_CHUNK(int, 16, 4, X, Y); \
400 TEST_EXTRA_CHUNK(int, 32, 2, X, Y); \
401 TEST_EXTRA_CHUNK(uint, 8, 8, X, Y); \
402 TEST_EXTRA_CHUNK(uint, 16, 4, X, Y); \
403 TEST_EXTRA_CHUNK(uint, 32, 2, X, Y); \
404 TEST_EXTRA_CHUNK(poly, 8, 8, X, Y); \
405 TEST_EXTRA_CHUNK(poly, 16, 4, X, Y); \
406 TEST_EXTRA_CHUNK(float, 32, 2, X, Y); \
407 TEST_EXTRA_CHUNK(int, 16, 8, X, Y); \
408 TEST_EXTRA_CHUNK(int, 32, 4, X, Y); \
409 TEST_EXTRA_CHUNK(uint, 16, 8, X, Y); \
410 TEST_EXTRA_CHUNK(uint, 32, 4, X, Y); \
411 TEST_EXTRA_CHUNK(poly, 16, 8, X, Y); \
412 TEST_EXTRA_CHUNK(float, 32, 4, X, Y)
413
414 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
415 #define TEST_ALL_EXTRA_CHUNKS(X,Y) \
416 TEST_ALL_EXTRA_CHUNKS_NO_FP16(X, Y); \
417 TEST_EXTRA_CHUNK(float, 16, 4, X, Y); \
418 TEST_EXTRA_CHUNK(float, 16, 8, X, Y)
419 #else
420 #define TEST_ALL_EXTRA_CHUNKS(X,Y) TEST_ALL_EXTRA_CHUNKS_NO_FP16(X, Y)
421 #endif
422
423 /* Declare the temporary buffers / variables. */
424 DECL_ALL_VSTX_LANE(2);
425 DECL_ALL_VSTX_LANE(3);
426 DECL_ALL_VSTX_LANE(4);
427
428 /* Define dummy input arrays, large enough for x4 vectors. */
429 DUMMY_ARRAY(buffer_src, int, 8, 8, 4);
430 DUMMY_ARRAY(buffer_src, int, 16, 4, 4);
431 DUMMY_ARRAY(buffer_src, int, 32, 2, 4);
432 DUMMY_ARRAY(buffer_src, uint, 8, 8, 4);
433 DUMMY_ARRAY(buffer_src, uint, 16, 4, 4);
434 DUMMY_ARRAY(buffer_src, uint, 32, 2, 4);
435 DUMMY_ARRAY(buffer_src, poly, 8, 8, 4);
436 DUMMY_ARRAY(buffer_src, poly, 16, 4, 4);
437 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
438 DUMMY_ARRAY(buffer_src, float, 16, 4, 4);
439 #endif
440 DUMMY_ARRAY(buffer_src, float, 32, 2, 4);
441 DUMMY_ARRAY(buffer_src, int, 16, 8, 4);
442 DUMMY_ARRAY(buffer_src, int, 32, 4, 4);
443 DUMMY_ARRAY(buffer_src, uint, 16, 8, 4);
444 DUMMY_ARRAY(buffer_src, uint, 32, 4, 4);
445 DUMMY_ARRAY(buffer_src, poly, 16, 8, 4);
446 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
447 DUMMY_ARRAY(buffer_src, float, 16, 8, 4);
448 #endif
449 DUMMY_ARRAY(buffer_src, float, 32, 4, 4);
450
451 /* Check vst2_lane/vst2q_lane. */
452 clean_results ();
453 #define TEST_MSG "VST2_LANE/VST2Q_LANE"
454 TEST_ALL_VSTX_LANE(2);
455
456 #define CMT " (chunk 0)"
457 CHECK(TEST_MSG, int, 8, 8, PRIx8, expected_st2_0, CMT);
458 CHECK(TEST_MSG, int, 16, 4, PRIx16, expected_st2_0, CMT);
459 CHECK(TEST_MSG, int, 32, 2, PRIx32, expected_st2_0, CMT);
460 CHECK(TEST_MSG, uint, 8, 8, PRIx8, expected_st2_0, CMT);
461 CHECK(TEST_MSG, uint, 16, 4, PRIx16, expected_st2_0, CMT);
462 CHECK(TEST_MSG, uint, 32, 2, PRIx32, expected_st2_0, CMT);
463 CHECK_POLY(TEST_MSG, poly, 8, 8, PRIx8, expected_st2_0, CMT);
464 CHECK_POLY(TEST_MSG, poly, 16, 4, PRIx16, expected_st2_0, CMT);
465 CHECK_FP(TEST_MSG, float, 32, 2, PRIx32, expected_st2_0, CMT);
466 CHECK(TEST_MSG, int, 16, 8, PRIx16, expected_st2_0, CMT);
467 CHECK(TEST_MSG, int, 32, 4, PRIx32, expected_st2_0, CMT);
468 CHECK(TEST_MSG, uint, 16, 8, PRIx16, expected_st2_0, CMT);
469 CHECK(TEST_MSG, uint, 32, 4, PRIx32, expected_st2_0, CMT);
470 CHECK_POLY(TEST_MSG, poly, 16, 8, PRIx16, expected_st2_0, CMT);
471 CHECK_FP(TEST_MSG, float, 32, 4, PRIx32, expected_st2_0, CMT);
472 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
473 CHECK_FP(TEST_MSG, float, 16, 4, PRIx16, expected_st2_0, CMT);
474 CHECK_FP(TEST_MSG, float, 16, 8, PRIx16, expected_st2_0, CMT);
475 #endif
476
477 TEST_ALL_EXTRA_CHUNKS(2, 1);
478 #undef CMT
479 #define CMT " (chunk 1)"
480 CHECK(TEST_MSG, int, 8, 8, PRIx8, expected_st2_1, CMT);
481 CHECK(TEST_MSG, int, 16, 4, PRIx16, expected_st2_1, CMT);
482 CHECK(TEST_MSG, int, 32, 2, PRIx32, expected_st2_1, CMT);
483 CHECK(TEST_MSG, uint, 8, 8, PRIx8, expected_st2_1, CMT);
484 CHECK(TEST_MSG, uint, 16, 4, PRIx16, expected_st2_1, CMT);
485 CHECK(TEST_MSG, uint, 32, 2, PRIx32, expected_st2_1, CMT);
486 CHECK_POLY(TEST_MSG, poly, 8, 8, PRIx8, expected_st2_1, CMT);
487 CHECK_POLY(TEST_MSG, poly, 16, 4, PRIx16, expected_st2_1, CMT);
488 CHECK_FP(TEST_MSG, float, 32, 2, PRIx32, expected_st2_1, CMT);
489 CHECK(TEST_MSG, int, 16, 8, PRIx16, expected_st2_1, CMT);
490 CHECK(TEST_MSG, int, 32, 4, PRIx32, expected_st2_1, CMT);
491 CHECK(TEST_MSG, uint, 16, 8, PRIx16, expected_st2_1, CMT);
492 CHECK(TEST_MSG, uint, 32, 4, PRIx32, expected_st2_1, CMT);
493 CHECK_POLY(TEST_MSG, poly, 16, 8, PRIx16, expected_st2_1, CMT);
494 CHECK_FP(TEST_MSG, float, 32, 4, PRIx32, expected_st2_1, CMT);
495 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
496 CHECK_FP(TEST_MSG, float, 16, 4, PRIx16, expected_st2_1, CMT);
497 CHECK_FP(TEST_MSG, float, 16, 8, PRIx16, expected_st2_1, CMT);
498 #endif
499
500
501 /* Check vst3_lane/vst3q_lane. */
502 clean_results ();
503 #undef TEST_MSG
504 #define TEST_MSG "VST3_LANE/VST3Q_LANE"
505 TEST_ALL_VSTX_LANE(3);
506
507 #undef CMT
508 #define CMT " (chunk 0)"
509 CHECK(TEST_MSG, int, 8, 8, PRIx8, expected_st3_0, CMT);
510 CHECK(TEST_MSG, int, 16, 4, PRIx16, expected_st3_0, CMT);
511 CHECK(TEST_MSG, int, 32, 2, PRIx32, expected_st3_0, CMT);
512 CHECK(TEST_MSG, uint, 8, 8, PRIx8, expected_st3_0, CMT);
513 CHECK(TEST_MSG, uint, 16, 4, PRIx16, expected_st3_0, CMT);
514 CHECK(TEST_MSG, uint, 32, 2, PRIx32, expected_st3_0, CMT);
515 CHECK_POLY(TEST_MSG, poly, 8, 8, PRIx8, expected_st3_0, CMT);
516 CHECK_POLY(TEST_MSG, poly, 16, 4, PRIx16, expected_st3_0, CMT);
517 CHECK_FP(TEST_MSG, float, 32, 2, PRIx32, expected_st3_0, CMT);
518 CHECK(TEST_MSG, int, 16, 8, PRIx16, expected_st3_0, CMT);
519 CHECK(TEST_MSG, int, 32, 4, PRIx32, expected_st3_0, CMT);
520 CHECK(TEST_MSG, uint, 16, 8, PRIx16, expected_st3_0, CMT);
521 CHECK(TEST_MSG, uint, 32, 4, PRIx32, expected_st3_0, CMT);
522 CHECK_POLY(TEST_MSG, poly, 16, 8, PRIx16, expected_st3_0, CMT);
523 CHECK_FP(TEST_MSG, float, 32, 4, PRIx32, expected_st3_0, CMT);
524 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
525 CHECK_FP(TEST_MSG, float, 16, 4, PRIx16, expected_st3_0, CMT);
526 CHECK_FP(TEST_MSG, float, 16, 8, PRIx16, expected_st3_0, CMT);
527 #endif
528
529 TEST_ALL_EXTRA_CHUNKS(3, 1);
530
531 #undef CMT
532 #define CMT " (chunk 1)"
533 CHECK(TEST_MSG, int, 8, 8, PRIx8, expected_st3_1, CMT);
534 CHECK(TEST_MSG, int, 16, 4, PRIx16, expected_st3_1, CMT);
535 CHECK(TEST_MSG, int, 32, 2, PRIx32, expected_st3_1, CMT);
536 CHECK(TEST_MSG, uint, 8, 8, PRIx8, expected_st3_1, CMT);
537 CHECK(TEST_MSG, uint, 16, 4, PRIx16, expected_st3_1, CMT);
538 CHECK(TEST_MSG, uint, 32, 2, PRIx32, expected_st3_1, CMT);
539 CHECK_POLY(TEST_MSG, poly, 8, 8, PRIx8, expected_st3_1, CMT);
540 CHECK_POLY(TEST_MSG, poly, 16, 4, PRIx16, expected_st3_1, CMT);
541 CHECK_FP(TEST_MSG, float, 32, 2, PRIx32, expected_st3_1, CMT);
542 CHECK(TEST_MSG, int, 16, 8, PRIx16, expected_st3_1, CMT);
543 CHECK(TEST_MSG, int, 32, 4, PRIx32, expected_st3_1, CMT);
544 CHECK(TEST_MSG, uint, 16, 8, PRIx16, expected_st3_1, CMT);
545 CHECK(TEST_MSG, uint, 32, 4, PRIx32, expected_st3_1, CMT);
546 CHECK_POLY(TEST_MSG, poly, 16, 8, PRIx16, expected_st3_1, CMT);
547 CHECK_FP(TEST_MSG, float, 32, 4, PRIx32, expected_st3_1, CMT);
548 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
549 CHECK_FP(TEST_MSG, float, 16, 4, PRIx16, expected_st3_1, CMT);
550 CHECK_FP(TEST_MSG, float, 16, 8, PRIx16, expected_st3_1, CMT);
551 #endif
552
553 TEST_ALL_EXTRA_CHUNKS(3, 2);
554
555 #undef CMT
556 #define CMT " (chunk 2)"
557 CHECK(TEST_MSG, int, 8, 8, PRIx8, expected_st3_2, CMT);
558 CHECK(TEST_MSG, int, 16, 4, PRIx16, expected_st3_2, CMT);
559 CHECK(TEST_MSG, int, 32, 2, PRIx32, expected_st3_2, CMT);
560 CHECK(TEST_MSG, uint, 8, 8, PRIx8, expected_st3_2, CMT);
561 CHECK(TEST_MSG, uint, 16, 4, PRIx16, expected_st3_2, CMT);
562 CHECK(TEST_MSG, uint, 32, 2, PRIx32, expected_st3_2, CMT);
563 CHECK_POLY(TEST_MSG, poly, 8, 8, PRIx8, expected_st3_2, CMT);
564 CHECK_POLY(TEST_MSG, poly, 16, 4, PRIx16, expected_st3_2, CMT);
565 CHECK_FP(TEST_MSG, float, 32, 2, PRIx32, expected_st3_2, CMT);
566 CHECK(TEST_MSG, int, 16, 8, PRIx16, expected_st3_2, CMT);
567 CHECK(TEST_MSG, int, 32, 4, PRIx32, expected_st3_2, CMT);
568 CHECK(TEST_MSG, uint, 16, 8, PRIx16, expected_st3_2, CMT);
569 CHECK(TEST_MSG, uint, 32, 4, PRIx32, expected_st3_2, CMT);
570 CHECK_POLY(TEST_MSG, poly, 16, 8, PRIx16, expected_st3_2, CMT);
571 CHECK_FP(TEST_MSG, float, 32, 4, PRIx32, expected_st3_2, CMT);
572 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
573 CHECK_FP(TEST_MSG, float, 16, 4, PRIx16, expected_st3_2, CMT);
574 CHECK_FP(TEST_MSG, float, 16, 8, PRIx16, expected_st3_2, CMT);
575 #endif
576
577
578 /* Check vst4_lane/vst4q_lane. */
579 clean_results ();
580 #undef TEST_MSG
581 #define TEST_MSG "VST4_LANE/VST4Q_LANE"
582 TEST_ALL_VSTX_LANE(4);
583
584 #undef CMT
585 #define CMT " (chunk 0)"
586 CHECK(TEST_MSG, int, 8, 8, PRIx8, expected_st4_0, CMT);
587 CHECK(TEST_MSG, int, 16, 4, PRIx16, expected_st4_0, CMT);
588 CHECK(TEST_MSG, int, 32, 2, PRIx32, expected_st4_0, CMT);
589 CHECK(TEST_MSG, uint, 8, 8, PRIx8, expected_st4_0, CMT);
590 CHECK(TEST_MSG, uint, 16, 4, PRIx16, expected_st4_0, CMT);
591 CHECK(TEST_MSG, uint, 32, 2, PRIx32, expected_st4_0, CMT);
592 CHECK_POLY(TEST_MSG, poly, 8, 8, PRIx8, expected_st4_0, CMT);
593 CHECK_POLY(TEST_MSG, poly, 16, 4, PRIx16, expected_st4_0, CMT);
594 CHECK_FP(TEST_MSG, float, 32, 2, PRIx32, expected_st4_0, CMT);
595 CHECK(TEST_MSG, int, 16, 8, PRIx16, expected_st4_0, CMT);
596 CHECK(TEST_MSG, int, 32, 4, PRIx32, expected_st4_0, CMT);
597 CHECK(TEST_MSG, uint, 16, 8, PRIx16, expected_st4_0, CMT);
598 CHECK(TEST_MSG, uint, 32, 4, PRIx32, expected_st4_0, CMT);
599 CHECK_POLY(TEST_MSG, poly, 16, 8, PRIx16, expected_st4_0, CMT);
600 CHECK_FP(TEST_MSG, float, 32, 4, PRIx32, expected_st4_0, CMT);
601 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
602 CHECK_FP(TEST_MSG, float, 16, 4, PRIx16, expected_st4_0, CMT);
603 CHECK_FP(TEST_MSG, float, 16, 8, PRIx16, expected_st4_0, CMT);
604 #endif
605
606 TEST_ALL_EXTRA_CHUNKS(4, 1);
607
608 #undef CMT
609 #define CMT " (chunk 1)"
610 CHECK(TEST_MSG, int, 8, 8, PRIx8, expected_st4_1, CMT);
611 CHECK(TEST_MSG, int, 16, 4, PRIx16, expected_st4_1, CMT);
612 CHECK(TEST_MSG, int, 32, 2, PRIx32, expected_st4_1, CMT);
613 CHECK(TEST_MSG, uint, 8, 8, PRIx8, expected_st4_1, CMT);
614 CHECK(TEST_MSG, uint, 16, 4, PRIx16, expected_st4_1, CMT);
615 CHECK(TEST_MSG, uint, 32, 2, PRIx32, expected_st4_1, CMT);
616 CHECK_POLY(TEST_MSG, poly, 8, 8, PRIx8, expected_st4_1, CMT);
617 CHECK_POLY(TEST_MSG, poly, 16, 4, PRIx16, expected_st4_1, CMT);
618 CHECK_FP(TEST_MSG, float, 32, 2, PRIx32, expected_st4_1, CMT);
619 CHECK(TEST_MSG, int, 16, 8, PRIx16, expected_st4_1, CMT);
620 CHECK(TEST_MSG, int, 32, 4, PRIx32, expected_st4_1, CMT);
621 CHECK(TEST_MSG, uint, 16, 8, PRIx16, expected_st4_1, CMT);
622 CHECK(TEST_MSG, uint, 32, 4, PRIx32, expected_st4_1, CMT);
623 CHECK_POLY(TEST_MSG, poly, 16, 8, PRIx16, expected_st4_1, CMT);
624 CHECK_FP(TEST_MSG, float, 32, 4, PRIx32, expected_st4_1, CMT);
625 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
626 CHECK_FP(TEST_MSG, float, 16, 4, PRIx16, expected_st4_1, CMT);
627 CHECK_FP(TEST_MSG, float, 16, 8, PRIx16, expected_st4_1, CMT);
628 #endif
629
630 TEST_ALL_EXTRA_CHUNKS(4, 2);
631
632 #undef CMT
633 #define CMT " (chunk 2)"
634 CHECK(TEST_MSG, int, 8, 8, PRIx8, expected_st4_2, CMT);
635 CHECK(TEST_MSG, int, 16, 4, PRIx16, expected_st4_2, CMT);
636 CHECK(TEST_MSG, int, 32, 2, PRIx32, expected_st4_2, CMT);
637 CHECK(TEST_MSG, uint, 8, 8, PRIx8, expected_st4_2, CMT);
638 CHECK(TEST_MSG, uint, 16, 4, PRIx16, expected_st4_2, CMT);
639 CHECK(TEST_MSG, uint, 32, 2, PRIx32, expected_st4_2, CMT);
640 CHECK_POLY(TEST_MSG, poly, 8, 8, PRIx8, expected_st4_2, CMT);
641 CHECK_POLY(TEST_MSG, poly, 16, 4, PRIx16, expected_st4_2, CMT);
642 CHECK_FP(TEST_MSG, float, 32, 2, PRIx32, expected_st4_2, CMT);
643 CHECK(TEST_MSG, int, 16, 8, PRIx16, expected_st4_2, CMT);
644 CHECK(TEST_MSG, int, 32, 4, PRIx32, expected_st4_2, CMT);
645 CHECK(TEST_MSG, uint, 16, 8, PRIx16, expected_st4_2, CMT);
646 CHECK(TEST_MSG, uint, 32, 4, PRIx32, expected_st4_2, CMT);
647 CHECK_POLY(TEST_MSG, poly, 16, 8, PRIx16, expected_st4_2, CMT);
648 CHECK_FP(TEST_MSG, float, 32, 4, PRIx32, expected_st4_2, CMT);
649 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
650 CHECK_FP(TEST_MSG, float, 16, 4, PRIx16, expected_st4_2, CMT);
651 CHECK_FP(TEST_MSG, float, 16, 8, PRIx16, expected_st4_2, CMT);
652 #endif
653
654 TEST_ALL_EXTRA_CHUNKS(4, 3);
655
656 #undef CMT
657 #define CMT " (chunk 3)"
658 CHECK(TEST_MSG, int, 8, 8, PRIx8, expected_st4_3, CMT);
659 CHECK(TEST_MSG, int, 16, 4, PRIx16, expected_st4_3, CMT);
660 CHECK(TEST_MSG, int, 32, 2, PRIx32, expected_st4_3, CMT);
661 CHECK(TEST_MSG, uint, 8, 8, PRIx8, expected_st4_3, CMT);
662 CHECK(TEST_MSG, uint, 16, 4, PRIx16, expected_st4_3, CMT);
663 CHECK(TEST_MSG, uint, 32, 2, PRIx32, expected_st4_3, CMT);
664 CHECK_POLY(TEST_MSG, poly, 8, 8, PRIx8, expected_st4_3, CMT);
665 CHECK_POLY(TEST_MSG, poly, 16, 4, PRIx16, expected_st4_3, CMT);
666 CHECK_FP(TEST_MSG, float, 32, 2, PRIx32, expected_st4_3, CMT);
667 CHECK(TEST_MSG, int, 16, 8, PRIx16, expected_st4_3, CMT);
668 CHECK(TEST_MSG, int, 32, 4, PRIx32, expected_st4_3, CMT);
669 CHECK(TEST_MSG, uint, 16, 8, PRIx16, expected_st4_3, CMT);
670 CHECK(TEST_MSG, uint, 32, 4, PRIx32, expected_st4_3, CMT);
671 CHECK_POLY(TEST_MSG, poly, 16, 8, PRIx16, expected_st4_3, CMT);
672 CHECK_FP(TEST_MSG, float, 32, 4, PRIx32, expected_st4_3, CMT);
673 #if defined (__ARM_FP16_FORMAT_IEEE) || defined (__ARM_FP16_FORMAT_ALTERNATIVE)
674 CHECK_FP(TEST_MSG, float, 16, 4, PRIx16, expected_st4_3, CMT);
675 CHECK_FP(TEST_MSG, float, 16, 8, PRIx16, expected_st4_3, CMT);
676 #endif
677 }
678
main(void)679 int main (void)
680 {
681 exec_vstX_lane ();
682 return 0;
683 }
684