1 /*
2
3 Copyright (c) 2009, 2010, 2011, 2012 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 #define INSN vqshlu
35 #define TEST_MSG "VQSHLU_N/VQSHLUQ_N"
36
37 #define FNNAME1(NAME) void exec_ ## NAME ## _n(void)
38 #define FNNAME(NAME) FNNAME1(NAME)
39
FNNAME(INSN)40 FNNAME (INSN)
41 {
42 /* Basic test: v2=vqshlu_n(v1,v), then store the result. */
43 #define TEST_VQSHLU_N2(INSN, Q, T1, T2, T3, T4, W, N, V) \
44 Set_Neon_Cumulative_Sat(0, VECT_VAR(vector_res, T3, W, N)); \
45 VECT_VAR(vector_res, T3, W, N) = \
46 INSN##Q##_n_##T2##W(VECT_VAR(vector, T1, W, N), \
47 V); \
48 vst1##Q##_##T4##W(VECT_VAR(result, T3, W, N), \
49 VECT_VAR(vector_res, T3, W, N)); \
50 dump_neon_cumulative_sat(TEST_MSG, xSTR(INSN##Q##_n_##T2##W), \
51 xSTR(T1), W, N)
52
53 /* Two auxliary macros are necessary to expand INSN */
54 #define TEST_VQSHLU_N1(INSN, Q, T1, T2, T3, T4, W, N, V) \
55 TEST_VQSHLU_N2(INSN, Q, T1, T2, T3, T4, W, N, V)
56
57 #define TEST_VQSHLU_N(Q, T1, T2, T3, T4, W, N, V) \
58 TEST_VQSHLU_N1(INSN, Q, T1, T2, T3, T4, W, N, V)
59
60
61 /* With ARM RVCT, we need to declare variables before any executable
62 statement */
63 DECL_VARIABLE_ALL_VARIANTS(vector);
64 DECL_VARIABLE_ALL_VARIANTS(vector_res);
65
66 clean_results ();
67
68 /* Fill input vector with negative values, to check saturation on limits */
69 VDUP(vector, , int, s, 8, 8, -1);
70 VDUP(vector, , int, s, 16, 4, -2);
71 VDUP(vector, , int, s, 32, 2, -3);
72 VDUP(vector, , int, s, 64, 1, -4);
73 VDUP(vector, q, int, s, 8, 16, -1);
74 VDUP(vector, q, int, s, 16, 8, -2);
75 VDUP(vector, q, int, s, 32, 4, -3);
76 VDUP(vector, q, int, s, 64, 2, -4);
77
78 /* Choose shift amount arbitrarily */
79 fprintf(ref_file, "\n%s cumulative saturation output:\n",
80 TEST_MSG " (negative input)");
81 TEST_VQSHLU_N(, int, s, uint, u, 8, 8, 2);
82 TEST_VQSHLU_N(, int, s, uint, u, 16, 4, 1);
83 TEST_VQSHLU_N(, int, s, uint, u, 32, 2, 1);
84 TEST_VQSHLU_N(, int, s, uint, u, 64, 1, 2);
85
86 TEST_VQSHLU_N(q, int, s, uint, u, 8, 16, 2);
87 TEST_VQSHLU_N(q, int, s, uint, u, 16, 8, 1);
88 TEST_VQSHLU_N(q, int, s, uint, u, 32, 4, 1);
89 TEST_VQSHLU_N(q, int, s, uint, u, 64, 2, 2);
90
91 /* FIXME: only a few result buffers are used, but we output all of them */
92 dump_results_hex2 (TEST_MSG, " (negative input)");
93
94 /* Fill input vector with max value, to check saturation on limits */
95 VDUP(vector, , int, s, 8, 8, 0x7F);
96 VDUP(vector, , int, s, 16, 4, 0x7FFF);
97 VDUP(vector, , int, s, 32, 2, 0x7FFFFFFF);
98 VDUP(vector, , int, s, 64, 1, 0x7FFFFFFFFFFFFFFFLL);
99 VDUP(vector, q, int, s, 8, 16, 0x7F);
100 VDUP(vector, q, int, s, 16, 8, 0x7FFF);
101 VDUP(vector, q, int, s, 32, 4, 0x7FFFFFFF);
102 VDUP(vector, q, int, s, 64, 2, 0x7FFFFFFFFFFFFFFFULL);
103
104 /* shift by 1 */
105 fprintf(ref_file, "\n%s cumulative saturation output:\n",
106 TEST_MSG " (check cumulative saturation: shift by 1)");
107 TEST_VQSHLU_N(, int, s, uint, u, 8, 8, 1);
108 TEST_VQSHLU_N(, int, s, uint, u, 16, 4, 1);
109 TEST_VQSHLU_N(, int, s, uint, u, 32, 2, 1);
110 TEST_VQSHLU_N(, int, s, uint, u, 64, 1, 1);
111
112 TEST_VQSHLU_N(q, int, s, uint, u, 8, 16, 1);
113 TEST_VQSHLU_N(q, int, s, uint, u, 16, 8, 1);
114 TEST_VQSHLU_N(q, int, s, uint, u, 32, 4, 1);
115 TEST_VQSHLU_N(q, int, s, uint, u, 64, 2, 1);
116
117 dump_results_hex2 (TEST_MSG, " (check cumulative saturation: shift by 1)");
118
119 /* shift by 2 to force saturation */
120 fprintf(ref_file, "\n%s cumulative saturation output:\n",
121 TEST_MSG " (check cumulative saturation: shift by 2)");
122 TEST_VQSHLU_N(, int, s, uint, u, 8, 8, 2);
123 TEST_VQSHLU_N(, int, s, uint, u, 16, 4, 2);
124 TEST_VQSHLU_N(, int, s, uint, u, 32, 2, 2);
125 TEST_VQSHLU_N(, int, s, uint, u, 64, 1, 2);
126
127 TEST_VQSHLU_N(q, int, s, uint, u, 8, 16, 2);
128 TEST_VQSHLU_N(q, int, s, uint, u, 16, 8, 2);
129 TEST_VQSHLU_N(q, int, s, uint, u, 32, 4, 2);
130 TEST_VQSHLU_N(q, int, s, uint, u, 64, 2, 2);
131
132 dump_results_hex2 (TEST_MSG, " (check cumulative saturation: shift by 2)");
133
134 /* Fill input vector with positive values, to check normal case */
135 VDUP(vector, , int, s, 8, 8, 1);
136 VDUP(vector, , int, s, 16, 4, 2);
137 VDUP(vector, , int, s, 32, 2, 3);
138 VDUP(vector, , int, s, 64, 1, 4);
139 VDUP(vector, q, int, s, 8, 16, 5);
140 VDUP(vector, q, int, s, 16, 8, 6);
141 VDUP(vector, q, int, s, 32, 4, 7);
142 VDUP(vector, q, int, s, 64, 2, 8);
143
144 /* shift arbitrary amount */
145 fprintf(ref_file, "\n%s cumulative saturation output:\n", TEST_MSG);
146 TEST_VQSHLU_N(, int, s, uint, u, 8, 8, 1);
147 TEST_VQSHLU_N(, int, s, uint, u, 16, 4, 2);
148 TEST_VQSHLU_N(, int, s, uint, u, 32, 2, 3);
149 TEST_VQSHLU_N(, int, s, uint, u, 64, 1, 4);
150
151 TEST_VQSHLU_N(q, int, s, uint, u, 8, 16, 5);
152 TEST_VQSHLU_N(q, int, s, uint, u, 16, 8, 6);
153 TEST_VQSHLU_N(q, int, s, uint, u, 32, 4, 7);
154 TEST_VQSHLU_N(q, int, s, uint, u, 64, 2, 8);
155
156 dump_results_hex (TEST_MSG);
157 }
158