Debian Codegraph @

gcc/testsuite/gcc.target/riscv/rvv/base/pr112431-34.c

raw
1 /* { dg-do compile } */
2 /* { dg-options "-march=rv64gcv -mabi=lp64d -O3" } */
3
4 #include "riscv_vector.h"
5
6 size_t __attribute__ ((noinline))
7 sumation (size_t sum0, size_t sum1, size_t sum2, size_t sum3, size_t sum4,
8 size_t sum5, size_t sum6, size_t sum7, size_t sum8, size_t sum9,
9 size_t sum10, size_t sum11, size_t sum12, size_t sum13, size_t sum14,
10 size_t sum15)
11 {
12 return sum0 + sum1 + sum2 + sum3 + sum4 + sum5 + sum6 + sum7 + sum8 + sum9
13 + sum10 + sum11 + sum12 + sum13 + sum14 + sum15;
14 }
15
16 size_t
17 foo (char const *buf, size_t len)
18 {
19 size_t sum = 0;
20 size_t vl = __riscv_vsetvlmax_e8m8 ();
21 size_t step = vl * 4;
22 const char *it = buf, *end = buf + len;
23 for (; it + step <= end;)
24 {
25 vuint8m1_t v0 = __riscv_vle8_v_u8m1 ((void *) it, vl);
26 it += vl;
27 vuint8m1_t v1 = __riscv_vle8_v_u8m1 ((void *) it, vl);
28 it += vl;
29 vuint8m1_t v2 = __riscv_vle8_v_u8m1 ((void *) it, vl);
30 it += vl;
31 vuint8m1_t v3 = __riscv_vle8_v_u8m1 ((void *) it, vl);
32 it += vl;
33 vuint8m1_t v4 = __riscv_vle8_v_u8m1 ((void *) it, vl);
34 it += vl;
35 vuint8m1_t v5 = __riscv_vle8_v_u8m1 ((void *) it, vl);
36 it += vl;
37 vuint8m1_t v6 = __riscv_vle8_v_u8m1 ((void *) it, vl);
38 it += vl;
39 vuint8m1_t v7 = __riscv_vle8_v_u8m1 ((void *) it, vl);
40 it += vl;
41 vuint8m1_t v8 = __riscv_vle8_v_u8m1 ((void *) it, vl);
42 it += vl;
43 vuint8m1_t v9 = __riscv_vle8_v_u8m1 ((void *) it, vl);
44 it += vl;
45 vuint8m1_t v10 = __riscv_vle8_v_u8m1 ((void *) it, vl);
46 it += vl;
47 vuint8m1_t v11 = __riscv_vle8_v_u8m1 ((void *) it, vl);
48 it += vl;
49 vuint8m1_t v12 = __riscv_vle8_v_u8m1 ((void *) it, vl);
50 it += vl;
51 vuint8m1_t v13 = __riscv_vle8_v_u8m1 ((void *) it, vl);
52 it += vl;
53 vuint8m1_t v14 = __riscv_vle8_v_u8m1 ((void *) it, vl);
54 it += vl;
55 vuint8m1_t v15 = __riscv_vle8_v_u8m1 ((void *) it, vl);
56 it += vl;
57
58 asm volatile("nop" ::: "memory");
59 vint16m2_t vw0 = __riscv_vluxei8_v_i16m2 ((void *) it, v0, vl);
60 vint16m2_t vw1 = __riscv_vluxei8_v_i16m2 ((void *) it, v1, vl);
61 vint16m2_t vw2 = __riscv_vluxei8_v_i16m2 ((void *) it, v2, vl);
62 vint16m2_t vw3 = __riscv_vluxei8_v_i16m2 ((void *) it, v3, vl);
63 vint16m2_t vw4 = __riscv_vluxei8_v_i16m2 ((void *) it, v4, vl);
64 vint16m2_t vw5 = __riscv_vluxei8_v_i16m2 ((void *) it, v5, vl);
65 vint16m2_t vw6 = __riscv_vluxei8_v_i16m2 ((void *) it, v6, vl);
66 vint16m2_t vw7 = __riscv_vluxei8_v_i16m2 ((void *) it, v7, vl);
67 vint16m2_t vw8 = __riscv_vluxei8_v_i16m2 ((void *) it, v8, vl);
68 vint16m2_t vw9 = __riscv_vluxei8_v_i16m2 ((void *) it, v9, vl);
69 vint16m2_t vw10 = __riscv_vluxei8_v_i16m2 ((void *) it, v10, vl);
70 vint16m2_t vw11 = __riscv_vluxei8_v_i16m2 ((void *) it, v11, vl);
71 vint16m2_t vw12 = __riscv_vluxei8_v_i16m2 ((void *) it, v12, vl);
72 vint16m2_t vw13 = __riscv_vluxei8_v_i16m2 ((void *) it, v13, vl);
73 vint16m2_t vw14 = __riscv_vluxei8_v_i16m2 ((void *) it, v14, vl);
74 vbool8_t mask = *(vbool8_t*)it;
75 vint16m2_t vw15 = __riscv_vluxei8_v_i16m2_m (mask, (void *) it, v15, vl);
76
77 asm volatile("nop" ::: "memory");
78 size_t sum0 = __riscv_vmv_x_s_i16m2_i16 (vw0);
79 size_t sum1 = __riscv_vmv_x_s_i16m2_i16 (vw1);
80 size_t sum2 = __riscv_vmv_x_s_i16m2_i16 (vw2);
81 size_t sum3 = __riscv_vmv_x_s_i16m2_i16 (vw3);
82 size_t sum4 = __riscv_vmv_x_s_i16m2_i16 (vw4);
83 size_t sum5 = __riscv_vmv_x_s_i16m2_i16 (vw5);
84 size_t sum6 = __riscv_vmv_x_s_i16m2_i16 (vw6);
85 size_t sum7 = __riscv_vmv_x_s_i16m2_i16 (vw7);
86 size_t sum8 = __riscv_vmv_x_s_i16m2_i16 (vw8);
87 size_t sum9 = __riscv_vmv_x_s_i16m2_i16 (vw9);
88 size_t sum10 = __riscv_vmv_x_s_i16m2_i16 (vw10);
89 size_t sum11 = __riscv_vmv_x_s_i16m2_i16 (vw11);
90 size_t sum12 = __riscv_vmv_x_s_i16m2_i16 (vw12);
91 size_t sum13 = __riscv_vmv_x_s_i16m2_i16 (vw13);
92 size_t sum14 = __riscv_vmv_x_s_i16m2_i16 (vw14);
93 size_t sum15 = __riscv_vmv_x_s_i16m2_i16 (vw15);
94
95 sum += sumation (sum0, sum1, sum2, sum3, sum4, sum5, sum6, sum7, sum8,
96 sum9, sum10, sum11, sum12, sum13, sum14, sum15);
97 }
98 return sum;
99 }
100
101 /* { dg-final { scan-assembler-not {vluxei8\.v\tv0,\s*\([a-x0-9]+\),\s*v[0-9]+,\s*v0.t} } } */