1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
|
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
// RUN: %clang_cc1 -triple riscv64-none-linux-gnu -target-feature +f -target-feature +d -target-feature +zve64d -mvscale-min=4 -mvscale-max=4 -S -disable-llvm-passes -emit-llvm -o - %s | FileCheck %s
// REQUIRES: riscv-registered-target
#include <riscv_vector.h>
typedef __rvv_int8m1_t vint8m1_t;
typedef __rvv_uint8m1_t vuint8m1_t;
typedef __rvv_int16m1_t vint16m1_t;
typedef __rvv_uint16m1_t vuint16m1_t;
typedef __rvv_int32m1_t vint32m1_t;
typedef __rvv_uint32m1_t vuint32m1_t;
typedef __rvv_int64m1_t vint64m1_t;
typedef __rvv_uint64m1_t vuint64m1_t;
typedef __rvv_float32m1_t vfloat32m1_t;
typedef __rvv_float64m1_t vfloat64m1_t;
typedef __rvv_int8m2_t vint8m2_t;
typedef __rvv_uint8m2_t vuint8m2_t;
typedef __rvv_int16m2_t vint16m2_t;
typedef __rvv_uint16m2_t vuint16m2_t;
typedef __rvv_int32m2_t vint32m2_t;
typedef __rvv_uint32m2_t vuint32m2_t;
typedef __rvv_int64m2_t vint64m2_t;
typedef __rvv_uint64m2_t vuint64m2_t;
typedef __rvv_float32m2_t vfloat32m2_t;
typedef __rvv_float64m2_t vfloat64m2_t;
typedef vint32m1_t fixed_int32m1_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen)));
typedef vint32m2_t fixed_int32m2_t __attribute__((riscv_rvv_vector_bits(__riscv_v_fixed_vlen * 2)));
fixed_int32m1_t global_vec;
fixed_int32m2_t global_vec_m2;
// CHECK-LABEL: @test_ptr_to_global(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[RETVAL:%.*]] = alloca <8 x i32>, align 8
// CHECK-NEXT: [[GLOBAL_VEC_PTR:%.*]] = alloca ptr, align 8
// CHECK-NEXT: store ptr @global_vec, ptr [[GLOBAL_VEC_PTR]], align 8
// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[GLOBAL_VEC_PTR]], align 8
// CHECK-NEXT: [[TMP1:%.*]] = load <8 x i32>, ptr [[TMP0]], align 8
// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[RETVAL]], align 8
// CHECK-NEXT: [[TMP2:%.*]] = load <8 x i32>, ptr [[RETVAL]], align 8
// CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[TMP2]], i64 0)
// CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
//
fixed_int32m1_t test_ptr_to_global() {
fixed_int32m1_t *global_vec_ptr;
global_vec_ptr = &global_vec;
return *global_vec_ptr;
}
//
// Test casting pointer from fixed-length array to scalable vector.
// CHECK-LABEL: @array_arg(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[RETVAL:%.*]] = alloca <8 x i32>, align 8
// CHECK-NEXT: [[ARR_ADDR:%.*]] = alloca ptr, align 8
// CHECK-NEXT: store ptr [[ARR:%.*]], ptr [[ARR_ADDR]], align 8
// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[ARR_ADDR]], align 8
// CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds <8 x i32>, ptr [[TMP0]], i64 0
// CHECK-NEXT: [[TMP1:%.*]] = load <8 x i32>, ptr [[ARRAYIDX]], align 8
// CHECK-NEXT: store <8 x i32> [[TMP1]], ptr [[RETVAL]], align 8
// CHECK-NEXT: [[TMP2:%.*]] = load <8 x i32>, ptr [[RETVAL]], align 8
// CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[TMP2]], i64 0)
// CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE]]
//
fixed_int32m1_t array_arg(fixed_int32m1_t arr[]) {
return arr[0];
}
// CHECK-LABEL: @test_cast(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[RETVAL:%.*]] = alloca <8 x i32>, align 8
// CHECK-NEXT: [[VEC_ADDR:%.*]] = alloca <vscale x 2 x i32>, align 4
// CHECK-NEXT: store <vscale x 2 x i32> [[VEC:%.*]], ptr [[VEC_ADDR]], align 4
// CHECK-NEXT: [[TMP0:%.*]] = load <8 x i32>, ptr @global_vec, align 8
// CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[TMP0]], i64 0)
// CHECK-NEXT: [[TMP1:%.*]] = load <vscale x 2 x i32>, ptr [[VEC_ADDR]], align 4
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i32> @llvm.riscv.vadd.nxv2i32.nxv2i32.i64(<vscale x 2 x i32> poison, <vscale x 2 x i32> [[CAST_SCALABLE]], <vscale x 2 x i32> [[TMP1]], i64 8)
// CHECK-NEXT: [[CAST_FIXED:%.*]] = call <8 x i32> @llvm.vector.extract.v8i32.nxv2i32(<vscale x 2 x i32> [[TMP2]], i64 0)
// CHECK-NEXT: store <8 x i32> [[CAST_FIXED]], ptr [[RETVAL]], align 8
// CHECK-NEXT: [[TMP3:%.*]] = load <8 x i32>, ptr [[RETVAL]], align 8
// CHECK-NEXT: [[CAST_SCALABLE1:%.*]] = call <vscale x 2 x i32> @llvm.vector.insert.nxv2i32.v8i32(<vscale x 2 x i32> undef, <8 x i32> [[TMP3]], i64 0)
// CHECK-NEXT: ret <vscale x 2 x i32> [[CAST_SCALABLE1]]
//
fixed_int32m1_t test_cast(vint32m1_t vec) {
return __riscv_vadd(global_vec, vec, __riscv_v_fixed_vlen/32);
}
// CHECK-LABEL: @test_ptr_to_global_m2(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[RETVAL:%.*]] = alloca <16 x i32>, align 8
// CHECK-NEXT: [[GLOBAL_VEC_PTR:%.*]] = alloca ptr, align 8
// CHECK-NEXT: store ptr @global_vec_m2, ptr [[GLOBAL_VEC_PTR]], align 8
// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[GLOBAL_VEC_PTR]], align 8
// CHECK-NEXT: [[TMP1:%.*]] = load <16 x i32>, ptr [[TMP0]], align 8
// CHECK-NEXT: store <16 x i32> [[TMP1]], ptr [[RETVAL]], align 8
// CHECK-NEXT: [[TMP2:%.*]] = load <16 x i32>, ptr [[RETVAL]], align 8
// CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i32> @llvm.vector.insert.nxv4i32.v16i32(<vscale x 4 x i32> undef, <16 x i32> [[TMP2]], i64 0)
// CHECK-NEXT: ret <vscale x 4 x i32> [[CAST_SCALABLE]]
//
fixed_int32m2_t test_ptr_to_global_m2() {
fixed_int32m2_t *global_vec_ptr;
global_vec_ptr = &global_vec_m2;
return *global_vec_ptr;
}
//
// Test casting pointer from fixed-length array to scalable vector.
// CHECK-LABEL: @array_arg_m2(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[RETVAL:%.*]] = alloca <16 x i32>, align 8
// CHECK-NEXT: [[ARR_ADDR:%.*]] = alloca ptr, align 8
// CHECK-NEXT: store ptr [[ARR:%.*]], ptr [[ARR_ADDR]], align 8
// CHECK-NEXT: [[TMP0:%.*]] = load ptr, ptr [[ARR_ADDR]], align 8
// CHECK-NEXT: [[ARRAYIDX:%.*]] = getelementptr inbounds <16 x i32>, ptr [[TMP0]], i64 0
// CHECK-NEXT: [[TMP1:%.*]] = load <16 x i32>, ptr [[ARRAYIDX]], align 8
// CHECK-NEXT: store <16 x i32> [[TMP1]], ptr [[RETVAL]], align 8
// CHECK-NEXT: [[TMP2:%.*]] = load <16 x i32>, ptr [[RETVAL]], align 8
// CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i32> @llvm.vector.insert.nxv4i32.v16i32(<vscale x 4 x i32> undef, <16 x i32> [[TMP2]], i64 0)
// CHECK-NEXT: ret <vscale x 4 x i32> [[CAST_SCALABLE]]
//
fixed_int32m2_t array_arg_m2(fixed_int32m2_t arr[]) {
return arr[0];
}
// CHECK-LABEL: @test_cast_m2(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[RETVAL:%.*]] = alloca <16 x i32>, align 8
// CHECK-NEXT: [[VEC_ADDR:%.*]] = alloca <vscale x 4 x i32>, align 4
// CHECK-NEXT: store <vscale x 4 x i32> [[VEC:%.*]], ptr [[VEC_ADDR]], align 4
// CHECK-NEXT: [[TMP0:%.*]] = load <16 x i32>, ptr @global_vec_m2, align 8
// CHECK-NEXT: [[CAST_SCALABLE:%.*]] = call <vscale x 4 x i32> @llvm.vector.insert.nxv4i32.v16i32(<vscale x 4 x i32> undef, <16 x i32> [[TMP0]], i64 0)
// CHECK-NEXT: [[TMP1:%.*]] = load <vscale x 4 x i32>, ptr [[VEC_ADDR]], align 4
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.riscv.vadd.nxv4i32.nxv4i32.i64(<vscale x 4 x i32> poison, <vscale x 4 x i32> [[CAST_SCALABLE]], <vscale x 4 x i32> [[TMP1]], i64 16)
// CHECK-NEXT: [[CAST_FIXED:%.*]] = call <16 x i32> @llvm.vector.extract.v16i32.nxv4i32(<vscale x 4 x i32> [[TMP2]], i64 0)
// CHECK-NEXT: store <16 x i32> [[CAST_FIXED]], ptr [[RETVAL]], align 8
// CHECK-NEXT: [[TMP3:%.*]] = load <16 x i32>, ptr [[RETVAL]], align 8
// CHECK-NEXT: [[CAST_SCALABLE1:%.*]] = call <vscale x 4 x i32> @llvm.vector.insert.nxv4i32.v16i32(<vscale x 4 x i32> undef, <16 x i32> [[TMP3]], i64 0)
// CHECK-NEXT: ret <vscale x 4 x i32> [[CAST_SCALABLE1]]
//
fixed_int32m2_t test_cast_m2(vint32m2_t vec) {
return __riscv_vadd(global_vec_m2, vec, __riscv_v_fixed_vlen/16);
}
|
