Debian Codegraph @

gcc/testsuite/gcc.target/riscv/rvv/base/pr112431-20.c

raw
1 /* { dg-do compile } */
2 /* { dg-options "-march=rv64gcv_zvfh_zfh -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 __attribute__ ((noinline))
17 foo (float const *buf, size_t len)
18 {
19 size_t sum = 0;
20 size_t vl = 4;
21 const float *it = buf;
22 for (int i = 0; i < len; i++)
23 {
24 vfloat32m2_t v0 = __riscv_vle32_v_f32m2 (it, vl);
25 it += vl;
26 vfloat32m2_t v1 = __riscv_vle32_v_f32m2 (it, vl);
27 it += vl;
28 vfloat32m2_t v2 = __riscv_vle32_v_f32m2 (it, vl);
29 it += vl;
30 vfloat32m2_t v3 = __riscv_vle32_v_f32m2 (it, vl);
31 it += vl;
32 vfloat32m2_t v4 = __riscv_vle32_v_f32m2 (it, vl);
33 it += vl;
34 vfloat32m2_t v5 = __riscv_vle32_v_f32m2 (it, vl);
35 it += vl;
36 vfloat32m2_t v6 = __riscv_vle32_v_f32m2 (it, vl);
37 it += vl;
38 vfloat32m2_t v7 = __riscv_vle32_v_f32m2 (it, vl);
39 it += vl;
40 vfloat32m2_t v8 = __riscv_vle32_v_f32m2 (it, vl);
41 it += vl;
42 vfloat32m2_t v9 = __riscv_vle32_v_f32m2 (it, vl);
43 it += vl;
44 vfloat32m2_t v10 = __riscv_vle32_v_f32m2 (it, vl);
45 it += vl;
46 vfloat32m2_t v11 = __riscv_vle32_v_f32m2 (it, vl);
47 it += vl;
48 vfloat32m2_t v12 = __riscv_vle32_v_f32m2 (it, vl);
49 it += vl;
50 vfloat32m2_t v13 = __riscv_vle32_v_f32m2 (it, vl);
51 it += vl;
52 vfloat32m2_t v14 = __riscv_vle32_v_f32m2 (it, vl);
53 it += vl;
54 vfloat32m2_t v15 = __riscv_vle32_v_f32m2 (it, vl);
55 it += vl;
56
57 asm volatile("nop" ::: "memory");
58 vfloat32m2_t vw0 = __riscv_vfwadd_wf_f32m2 (v0, 55, vl);
59 vfloat32m2_t vw1 = __riscv_vfwadd_wf_f32m2 (v1, 55, vl);
60 vfloat32m2_t vw2 = __riscv_vfwadd_wf_f32m2 (v2, 55, vl);
61 vfloat32m2_t vw3 = __riscv_vfwadd_wf_f32m2 (v3, 55, vl);
62 vfloat32m2_t vw4 = __riscv_vfwadd_wf_f32m2 (v4, 55, vl);
63 vfloat32m2_t vw5 = __riscv_vfwadd_wf_f32m2 (v5, 55, vl);
64 vfloat32m2_t vw6 = __riscv_vfwadd_wf_f32m2 (v6, 55, vl);
65 vfloat32m2_t vw7 = __riscv_vfwadd_wf_f32m2 (v7, 55, vl);
66 vfloat32m2_t vw8 = __riscv_vfwadd_wf_f32m2 (v8, 55, vl);
67 vfloat32m2_t vw9 = __riscv_vfwadd_wf_f32m2 (v9, 55, vl);
68 vfloat32m2_t vw10 = __riscv_vfwadd_wf_f32m2 (v10, 55, vl);
69 vfloat32m2_t vw11 = __riscv_vfwadd_wf_f32m2 (v11, 55, vl);
70 vfloat32m2_t vw12 = __riscv_vfwadd_wf_f32m2 (v12, 55, vl);
71 vfloat32m2_t vw13 = __riscv_vfwadd_wf_f32m2 (v13, 55, vl);
72 vfloat32m2_t vw14 = __riscv_vfwadd_wf_f32m2 (v14, 55, vl);
73 vfloat32m2_t vw15 = __riscv_vfwadd_wf_f32m2 (v15, 55, vl);
74
75 asm volatile("nop" ::: "memory");
76 size_t sum0 = __riscv_vfmv_f_s_f32m2_f32 (vw0);
77 size_t sum1 = __riscv_vfmv_f_s_f32m2_f32 (vw1);
78 size_t sum2 = __riscv_vfmv_f_s_f32m2_f32 (vw2);
79 size_t sum3 = __riscv_vfmv_f_s_f32m2_f32 (vw3);
80 size_t sum4 = __riscv_vfmv_f_s_f32m2_f32 (vw4);
81 size_t sum5 = __riscv_vfmv_f_s_f32m2_f32 (vw5);
82 size_t sum6 = __riscv_vfmv_f_s_f32m2_f32 (vw6);
83 size_t sum7 = __riscv_vfmv_f_s_f32m2_f32 (vw7);
84 size_t sum8 = __riscv_vfmv_f_s_f32m2_f32 (vw8);
85 size_t sum9 = __riscv_vfmv_f_s_f32m2_f32 (vw9);
86 size_t sum10 = __riscv_vfmv_f_s_f32m2_f32 (vw10);
87 size_t sum11 = __riscv_vfmv_f_s_f32m2_f32 (vw11);
88 size_t sum12 = __riscv_vfmv_f_s_f32m2_f32 (vw12);
89 size_t sum13 = __riscv_vfmv_f_s_f32m2_f32 (vw13);
90 size_t sum14 = __riscv_vfmv_f_s_f32m2_f32 (vw14);
91 size_t sum15 = __riscv_vfmv_f_s_f32m2_f32 (vw15);
92
93 sum += sumation (sum0, sum1, sum2, sum3, sum4, sum5, sum6, sum7, sum8,
94 sum9, sum10, sum11, sum12, sum13, sum14, sum15);
95 }
96 return sum;
97 }
98
99 /* { dg-final { scan-assembler-not {vmv1r} } } */
100 /* { dg-final { scan-assembler-not {vmv2r} } } */
101 /* { dg-final { scan-assembler-not {vmv4r} } } */
102 /* { dg-final { scan-assembler-not {vmv8r} } } */
103 /* { dg-final { scan-assembler-not {csrr} } } */