Debian Codegraph @

gcc/testsuite/gcc.target/aarch64/sme/acle-asm/ld1_ver_za32.c

raw
1 /* { dg-final { check-function-bodies "**" "" "-DCHECK_ASM" } } */
2
3 #include "test_sme_acle.h"
4
5 /*
6 ** ld1_za32_0_0:
7 ** mov (w1[2-5]), (?:w0|#?0)
8 ** ld1w { za0v\.s\[\1, 0\] }, p0/z, \[x1\]
9 ** ret
10 */
11 TEST_LOAD_ZA (ld1_za32_0_0,
12 svld1_ver_za32 (0, 0, p0, x1),
13 svld1_ver_za32 (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 3. */
17 /*
18 ** ld1_za32_0_3:
19 ** mov (w1[2-5]), #?3
20 ** ld1w { za0v\.s\[\1, 0\] }, p0/z, \[x1\]
21 ** ret
22 */
23 TEST_LOAD_ZA (ld1_za32_0_3,
24 svld1_ver_za32 (0, 3, p0, x1),
25 svld1_ver_za32 (0, 3, p0, x1))
26
27 /*
28 ** ld1_za32_0_4:
29 ** mov (w1[2-5]), #?4
30 ** ld1w { za0v\.s\[\1, 0\] }, p0/z, \[x1\]
31 ** ret
32 */
33 TEST_LOAD_ZA (ld1_za32_0_4,
34 svld1_ver_za32 (0, 4, p0, x1),
35 svld1_ver_za32 (0, 4, p0, x1))
36
37 /*
38 ** ld1_za32_0_w0:
39 ** mov (w1[2-5]), w0
40 ** ld1w { za0v\.s\[\1, 0\] }, p0/z, \[x1\]
41 ** ret
42 */
43 TEST_LOAD_ZA (ld1_za32_0_w0,
44 svld1_ver_za32 (0, w0, p0, x1),
45 svld1_ver_za32 (0, w0, p0, x1))
46
47 /*
48 ** ld1_za32_0_w0_p1:
49 ** mov (w1[2-5]), w0
50 ** ld1w { za0v\.s\[\1, 1\] }, p0/z, \[x1\]
51 ** ret
52 */
53 TEST_LOAD_ZA (ld1_za32_0_w0_p1,
54 svld1_ver_za32 (0, w0 + 1, p0, x1),
55 svld1_ver_za32 (0, w0 + 1, p0, x1))
56
57 /*
58 ** ld1_za32_0_w0_p3:
59 ** mov (w1[2-5]), w0
60 ** ld1w { za0v\.s\[\1, 3\] }, p0/z, \[x1\]
61 ** ret
62 */
63 TEST_LOAD_ZA (ld1_za32_0_w0_p3,
64 svld1_ver_za32 (0, w0 + 3, p0, x1),
65 svld1_ver_za32 (0, w0 + 3, p0, x1))
66
67 /*
68 ** ld1_za32_3_w0:
69 ** mov (w1[2-5]), w0
70 ** ld1w { za3v\.s\[\1, 0\] }, p0/z, \[x1\]
71 ** ret
72 */
73 TEST_LOAD_ZA (ld1_za32_3_w0,
74 svld1_ver_za32 (3, w0, p0, x1),
75 svld1_ver_za32 (3, w0, p0, x1))
76
77 /*
78 ** ld1_za32_3_w0_p1:
79 ** mov (w1[2-5]), w0
80 ** ld1w { za3v\.s\[\1, 1\] }, p0/z, \[x1\]
81 ** ret
82 */
83 TEST_LOAD_ZA (ld1_za32_3_w0_p1,
84 svld1_ver_za32 (3, w0 + 1, p0, x1),
85 svld1_ver_za32 (3, w0 + 1, p0, x1))
86
87 /*
88 ** ld1_za32_3_w0_p3:
89 ** mov (w1[2-5]), w0
90 ** ld1w { za3v\.s\[\1, 3\] }, p0/z, \[x1\]
91 ** ret
92 */
93 TEST_LOAD_ZA (ld1_za32_3_w0_p3,
94 svld1_ver_za32 (3, w0 + 3, p0, x1),
95 svld1_ver_za32 (3, w0 + 3, p0, x1))
96
97 /*
98 ** ld1_za32_1_w0_p2_index:
99 ** mov (w1[2-5]), w0
100 ** ld1w { za1v\.s\[\1, 2\] }, p0/z, \[x1, x2, lsl #?2\]
101 ** ret
102 */
103 TEST_LOAD_ZA (ld1_za32_1_w0_p2_index,
104 svld1_ver_za32 (1, w0 + 2, p0, x1 + x2 * 4),
105 svld1_ver_za32 (1, w0 + 2, p0, x1 + x2 * 4))
106
107 /*
108 ** ld1_za32_0_w0_p4:
109 ** add (w1[2-5]), w0, #?4
110 ** ld1w { za0v\.s\[\1, 0\] }, p0/z, \[x1\]
111 ** ret
112 */
113 TEST_LOAD_ZA (ld1_za32_0_w0_p4,
114 svld1_ver_za32 (0, w0 + 4, p0, x1),
115 svld1_ver_za32 (0, w0 + 4, p0, x1))
116
117 /*
118 ** ld1_za32_0_w0_m1:
119 ** sub (w1[2-5]), w0, #?1
120 ** ld1w { za0v\.s\[\1, 0\] }, p0/z, \[x1\]
121 ** ret
122 */
123 TEST_LOAD_ZA (ld1_za32_0_w0_m1,
124 svld1_ver_za32 (0, w0 - 1, p0, x1),
125 svld1_ver_za32 (0, w0 - 1, p0, x1))