Debian Codegraph @

gcc/testsuite/gcc.target/aarch64/sme/acle-asm/ld1_ver_za16.c

raw
1 /* { dg-final { check-function-bodies "**" "" "-DCHECK_ASM" } } */
2
3 #include "test_sme_acle.h"
4
5 /*
6 ** ld1_za16_0_0:
7 ** mov (w1[2-5]), (?:wzr|#?0)
8 ** ld1h { za0v\.h\[\1, 0\] }, p0/z, \[x1\]
9 ** ret
10 */
11 TEST_LOAD_ZA (ld1_za16_0_0,
12 svld1_ver_za16 (0, 0, p0, x1),
13 svld1_ver_za16 (0, 0, p0, x1))
14
15 /* It would also be OK (and perhaps better) to move 0 into a register
16 and use an offset of 7. */
17 /*
18 ** ld1_za16_0_7:
19 ** mov (w1[2-5]), #?7
20 ** ld1h { za0v\.h\[\1, 0\] }, p0/z, \[x1\]
21 ** ret
22 */
23 TEST_LOAD_ZA (ld1_za16_0_7,
24 svld1_ver_za16 (0, 7, p0, x1),
25 svld1_ver_za16 (0, 7, p0, x1))
26
27 /*
28 ** ld1_za16_0_8:
29 ** mov (w1[2-5]), #?8
30 ** ld1h { za0v\.h\[\1, 0\] }, p0/z, \[x1\]
31 ** ret
32 */
33 TEST_LOAD_ZA (ld1_za16_0_8,
34 svld1_ver_za16 (0, 8, p0, x1),
35 svld1_ver_za16 (0, 8, p0, x1))
36
37 /*
38 ** ld1_za16_0_w0:
39 ** mov (w1[2-5]), w0
40 ** ld1h { za0v\.h\[\1, 0\] }, p0/z, \[x1\]
41 ** ret
42 */
43 TEST_LOAD_ZA (ld1_za16_0_w0,
44 svld1_ver_za16 (0, w0, p0, x1),
45 svld1_ver_za16 (0, w0, p0, x1))
46
47 /*
48 ** ld1_za16_0_w0_p1:
49 ** mov (w1[2-5]), w0
50 ** ld1h { za0v\.h\[\1, 1\] }, p0/z, \[x1\]
51 ** ret
52 */
53 TEST_LOAD_ZA (ld1_za16_0_w0_p1,
54 svld1_ver_za16 (0, w0 + 1, p0, x1),
55 svld1_ver_za16 (0, w0 + 1, p0, x1))
56
57 /*
58 ** ld1_za16_0_w0_p7:
59 ** mov (w1[2-5]), w0
60 ** ld1h { za0v\.h\[\1, 7\] }, p0/z, \[x1\]
61 ** ret
62 */
63 TEST_LOAD_ZA (ld1_za16_0_w0_p7,
64 svld1_ver_za16 (0, w0 + 7, p0, x1),
65 svld1_ver_za16 (0, w0 + 7, p0, x1))
66
67 /*
68 ** ld1_za16_1_w0:
69 ** mov (w1[2-5]), w0
70 ** ld1h { za1v\.h\[\1, 0\] }, p0/z, \[x1\]
71 ** ret
72 */
73 TEST_LOAD_ZA (ld1_za16_1_w0,
74 svld1_ver_za16 (1, w0, p0, x1),
75 svld1_ver_za16 (1, w0, p0, x1))
76
77
78 /*
79 ** ld1_za16_1_w0_p1:
80 ** mov (w1[2-5]), w0
81 ** ld1h { za1v\.h\[\1, 1\] }, p0/z, \[x1\]
82 ** ret
83 */
84 TEST_LOAD_ZA (ld1_za16_1_w0_p1,
85 svld1_ver_za16 (1, w0 + 1, p0, x1),
86 svld1_ver_za16 (1, w0 + 1, p0, x1))
87
88 /*
89 ** ld1_za16_1_w0_p7:
90 ** mov (w1[2-5]), w0
91 ** ld1h { za1v\.h\[\1, 7\] }, p0/z, \[x1\]
92 ** ret
93 */
94 TEST_LOAD_ZA (ld1_za16_1_w0_p7,
95 svld1_ver_za16 (1, w0 + 7, p0, x1),
96 svld1_ver_za16 (1, w0 + 7, p0, x1))
97
98 /*
99 ** ld1_za16_1_w0_p5_index:
100 ** mov (w1[2-5]), w0
101 ** ld1h { za1v\.h\[\1, 5\] }, p0/z, \[x1, x2, lsl #?1\]
102 ** ret
103 */
104 TEST_LOAD_ZA (ld1_za16_1_w0_p5_index,
105 svld1_ver_za16 (1, w0 + 5, p0, x1 + x2 * 2),
106 svld1_ver_za16 (1, w0 + 5, p0, x1 + x2 * 2))
107
108 /*
109 ** ld1_za16_0_w0_p8:
110 ** add (w1[2-5]), w0, #?8
111 ** ld1h { za0v\.h\[\1, 0\] }, p0/z, \[x1\]
112 ** ret
113 */
114 TEST_LOAD_ZA (ld1_za16_0_w0_p8,
115 svld1_ver_za16 (0, w0 + 8, p0, x1),
116 svld1_ver_za16 (0, w0 + 8, p0, x1))
117
118 /*
119 ** ld1_za16_0_w0_m1:
120 ** sub (w1[2-5]), w0, #?1
121 ** ld1h { za0v\.h\[\1, 0\] }, p0/z, \[x1\]
122 ** ret
123 */
124 TEST_LOAD_ZA (ld1_za16_0_w0_m1,
125 svld1_ver_za16 (0, w0 - 1, p0, x1),
126 svld1_ver_za16 (0, w0 - 1, p0, x1))