1 /*
2
3 Copyright (c) 2009, 2010, 2011 STMicroelectronics
4 Written by Christophe Lyon
5
6 Permission is hereby granted, free of charge, to any person obtaining a copy
7 of this software and associated documentation files (the "Software"), to deal
8 in the Software without restriction, including without limitation the rights
9 to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
10 copies of the Software, and to permit persons to whom the Software is
11 furnished to do so, subject to the following conditions:
12
13 The above copyright notice and this permission notice shall be included in
14 all copies or substantial portions of the Software.
15
16 THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
17 IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
18 FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
19 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
20 LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
21 OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
22 THE SOFTWARE.
23
24 */
25
26 #if defined(__arm__) || defined(__aarch64__)
27 #include <arm_neon.h>
28 #else
29 #include "stm-arm-neon.h"
30 #endif
31
32 #include "stm-arm-neon-ref.h"
33
34 #ifndef INSN_NAME
35 #define INSN_NAME vmla
36 #define TEST_MSG "VMLA_LANE"
37 #endif
38
39 #define FNNAME1(NAME) void exec_ ## NAME ##_lane (void)
40 #define FNNAME(NAME) FNNAME1(NAME)
41
FNNAME(INSN_NAME)42 FNNAME (INSN_NAME)
43 {
44 #define DECL_VMLX_LANE(VAR) \
45 DECL_VARIABLE(VAR, int, 16, 4); \
46 DECL_VARIABLE(VAR, int, 32, 2); \
47 DECL_VARIABLE(VAR, uint, 16, 4); \
48 DECL_VARIABLE(VAR, uint, 32, 2); \
49 DECL_VARIABLE(VAR, float, 32, 2); \
50 DECL_VARIABLE(VAR, int, 16, 8); \
51 DECL_VARIABLE(VAR, int, 32, 4); \
52 DECL_VARIABLE(VAR, uint, 16, 8); \
53 DECL_VARIABLE(VAR, uint, 32, 4); \
54 DECL_VARIABLE(VAR, float, 32, 4)
55
56 /* vector_res = vmlx_lane(vector, vector2, vector3, lane),
57 then store the result. */
58 #define TEST_VMLX_LANE1(INSN, Q, T1, T2, W, N, N2, L) \
59 VECT_VAR(vector_res, T1, W, N) = \
60 INSN##Q##_lane_##T2##W(VECT_VAR(vector, T1, W, N), \
61 VECT_VAR(vector2, T1, W, N), \
62 VECT_VAR(vector3, T1, W, N2), \
63 L); \
64 vst1##Q##_##T2##W(VECT_VAR(result, T1, W, N), \
65 VECT_VAR(vector_res, T1, W, N))
66
67 #define TEST_VMLX_LANE(INSN, Q, T1, T2, W, N, N2, V) \
68 TEST_VMLX_LANE1(INSN, Q, T1, T2, W, N, N2, V)
69
70 /* With ARM RVCT, we need to declare variables before any executable
71 statement */
72 DECL_VMLX_LANE(vector);
73 DECL_VMLX_LANE(vector2);
74 DECL_VMLX_LANE(vector_res);
75
76 DECL_VARIABLE(vector3, int, 16, 4);
77 DECL_VARIABLE(vector3, int, 32, 2);
78 DECL_VARIABLE(vector3, uint, 16, 4);
79 DECL_VARIABLE(vector3, uint, 32, 2);
80 DECL_VARIABLE(vector3, float, 32, 2);
81
82 clean_results ();
83
84 VLOAD(vector, buffer, , int, s, 16, 4);
85 VLOAD(vector, buffer, , int, s, 32, 2);
86 VLOAD(vector, buffer, , uint, u, 16, 4);
87 VLOAD(vector, buffer, , uint, u, 32, 2);
88 VLOAD(vector, buffer, q, int, s, 16, 8);
89 VLOAD(vector, buffer, q, int, s, 32, 4);
90 VLOAD(vector, buffer, q, uint, u, 16, 8);
91 VLOAD(vector, buffer, q, uint, u, 32, 4);
92 VLOAD(vector, buffer, , float, f, 32, 2);
93 VLOAD(vector, buffer, q, float, f, 32, 4);
94
95 VDUP(vector2, , int, s, 16, 4, 0x55);
96 VDUP(vector2, , int, s, 32, 2, 0x55);
97 VDUP(vector2, , uint, u, 16, 4, 0x55);
98 VDUP(vector2, , uint, u, 32, 2, 0x55);
99 VDUP(vector2, , float, f, 32, 2, 55.3f);
100 VDUP(vector2, q, int, s, 16, 8, 0x55);
101 VDUP(vector2, q, int, s, 32, 4, 0x55);
102 VDUP(vector2, q, uint, u, 16, 8, 0x55);
103 VDUP(vector2, q, uint, u, 32, 4, 0x55);
104 VDUP(vector2, q, float, f, 32, 4, 55.8f);
105
106 VDUP(vector3, , int, s, 16, 4, 0xBB);
107 VDUP(vector3, , int, s, 32, 2, 0xBB);
108 VDUP(vector3, , uint, u, 16, 4, 0xBB);
109 VDUP(vector3, , uint, u, 32, 2, 0xBB);
110 VDUP(vector3, , float, f, 32, 2, 11.34f);
111
112 /* Choose lane arbitrarily */
113 TEST_VMLX_LANE(INSN_NAME, , int, s, 16, 4, 4, 2);
114 TEST_VMLX_LANE(INSN_NAME, , int, s, 32, 2, 2, 1);
115 TEST_VMLX_LANE(INSN_NAME, , uint, u, 16, 4, 4, 2);
116 TEST_VMLX_LANE(INSN_NAME, , uint, u, 32, 2, 2, 1);
117 TEST_VMLX_LANE(INSN_NAME, , float, f, 32, 2, 2, 1);
118 TEST_VMLX_LANE(INSN_NAME, q, int, s, 16, 8, 4, 3);
119 TEST_VMLX_LANE(INSN_NAME, q, int, s, 32, 4, 2, 1);
120 TEST_VMLX_LANE(INSN_NAME, q, uint, u, 16, 8, 4, 2);
121 TEST_VMLX_LANE(INSN_NAME, q, uint, u, 32, 4, 2, 1);
122 TEST_VMLX_LANE(INSN_NAME, q, float, f, 32, 4, 2, 1);
123
124 dump_results_hex (TEST_MSG);
125 }
126