Debian Codegraph @

gcc/testsuite/gcc.target/aarch64/sme/acle-asm/ld1_hor_za8.c

raw
1 /* { dg-final { check-function-bodies "**" "" "-DCHECK_ASM" } } */
2
3 #include "test_sme_acle.h"
4
5 /*
6 ** ld1_za8_0_0:
7 ** mov (w1[2-5]), (?:wzr|#?0)
8 ** ld1b { za0h\.b\[\1, 0\] }, p0/z, \[x1\]
9 ** ret
10 */
11 TEST_LOAD_ZA (ld1_za8_0_0,
12 svld1_hor_za8 (0, 0, p0, x1),
13 svld1_hor_za8 (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 15. */
17 /*
18 ** ld1_za8_0_15:
19 ** mov (w1[2-5]), #?15
20 ** ld1b { za0h\.b\[\1, 0\] }, p0/z, \[x1\]
21 ** ret
22 */
23 TEST_LOAD_ZA (ld1_za8_0_15,
24 svld1_hor_za8 (0, 15, p0, x1),
25 svld1_hor_za8 (0, 15, p0, x1))
26
27 /*
28 ** ld1_za8_0_16:
29 ** mov (w1[2-5]), #?16
30 ** ld1b { za0h\.b\[\1, 0\] }, p0/z, \[x1\]
31 ** ret
32 */
33 TEST_LOAD_ZA (ld1_za8_0_16,
34 svld1_hor_za8 (0, 16, p0, x1),
35 svld1_hor_za8 (0, 16, p0, x1))
36
37 /*
38 ** ld1_za8_0_w0:
39 ** mov (w1[2-5]), w0
40 ** ld1b { za0h\.b\[\1, 0\] }, p0/z, \[x1\]
41 ** ret
42 */
43 TEST_LOAD_ZA (ld1_za8_0_w0,
44 svld1_hor_za8 (0, w0, p0, x1),
45 svld1_hor_za8 (0, w0, p0, x1))
46
47 /*
48 ** ld1_za8_0_w0_p1:
49 ** mov (w1[2-5]), w0
50 ** ld1b { za0h\.b\[\1, 1\] }, p0/z, \[x1\]
51 ** ret
52 */
53 TEST_LOAD_ZA (ld1_za8_0_w0_p1,
54 svld1_hor_za8 (0, w0 + 1, p0, x1),
55 svld1_hor_za8 (0, w0 + 1, p0, x1))
56
57 /*
58 ** ld1_za8_0_w0_p15:
59 ** mov (w1[2-5]), w0
60 ** ld1b { za0h\.b\[\1, 15\] }, p0/z, \[x1\]
61 ** ret
62 */
63 TEST_LOAD_ZA (ld1_za8_0_w0_p15,
64 svld1_hor_za8 (0, w0 + 15, p0, x1),
65 svld1_hor_za8 (0, w0 + 15, p0, x1))
66
67 /*
68 ** ld1_za8_0_w0_p13_index:
69 ** mov (w1[2-5]), w0
70 ** ld1b { za0h\.b\[\1, 15\] }, p0/z, \[x1, x2\]
71 ** ret
72 */
73 TEST_LOAD_ZA (ld1_za8_0_w0_p13_index,
74 svld1_hor_za8 (0, w0 + 15, p0, x1 + x2),
75 svld1_hor_za8 (0, w0 + 15, p0, x1 + x2))
76
77 /*
78 ** ld1_za8_0_w0_p16:
79 ** add (w1[2-5]), w0, #?16
80 ** ld1b { za0h\.b\[\1, 0\] }, p0/z, \[x1\]
81 ** ret
82 */
83 TEST_LOAD_ZA (ld1_za8_0_w0_p16,
84 svld1_hor_za8 (0, w0 + 16, p0, x1),
85 svld1_hor_za8 (0, w0 + 16, p0, x1))
86
87 /*
88 ** ld1_za8_0_w0_m1:
89 ** sub (w1[2-5]), w0, #?1
90 ** ld1b { za0h\.b\[\1, 0\] }, p0/z, \[x1\]
91 ** ret
92 */
93 TEST_LOAD_ZA (ld1_za8_0_w0_m1,
94 svld1_hor_za8 (0, w0 - 1, p0, x1),
95 svld1_hor_za8 (0, w0 - 1, p0, x1))