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
|
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
;=========================== begin_copyright_notice ============================
;
; Copyright (C) 2023 Intel Corporation
;
; SPDX-License-Identifier: MIT
;
;============================ end_copyright_notice =============================
; RUN: igc_opt %s --opaque-pointers -S -o - -types-legalization-pass | FileCheck %s
; REQUIRES: llvm-14-plus
define spir_kernel void @f0(ptr addrspace(1) %arg) {
; CHECK-LABEL: define spir_kernel void @f0(
; CHECK-SAME: ptr addrspace(1) [[ARG:%.*]]) {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: [[PTR:%.*]] = load ptr addrspace(1), ptr addrspace(1) [[ARG]], align 8
; CHECK-NEXT: call void @llvm.memset.p1.i64(ptr addrspace(1) align 8 [[PTR]], i8 0, i64 960032, i1 false)
; CHECK-NEXT: ret void
;
entry:
%ptr = load ptr addrspace(1), ptr addrspace(1) %arg, align 8
store [120004 x double] zeroinitializer, ptr addrspace(1) %ptr, align 8
ret void
}
define spir_kernel void @f1([10 x double] addrspace(1)* %arg) {
; CHECK-LABEL: define spir_kernel void @f1(
; CHECK-SAME: ptr addrspace(1) [[ARG:%.*]]) {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 0
; CHECK-NEXT: store double 1.000000e+00, ptr addrspace(1) [[TMP0]], align 8
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 1
; CHECK-NEXT: store double 2.000000e+00, ptr addrspace(1) [[TMP1]], align 8
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 2
; CHECK-NEXT: store double 3.000000e+00, ptr addrspace(1) [[TMP2]], align 8
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 3
; CHECK-NEXT: store double 4.000000e+00, ptr addrspace(1) [[TMP3]], align 8
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 4
; CHECK-NEXT: store double 5.000000e+00, ptr addrspace(1) [[TMP4]], align 8
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 5
; CHECK-NEXT: store double 6.000000e+00, ptr addrspace(1) [[TMP5]], align 8
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 6
; CHECK-NEXT: store double 7.000000e+00, ptr addrspace(1) [[TMP6]], align 8
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 7
; CHECK-NEXT: store double 8.000000e+00, ptr addrspace(1) [[TMP7]], align 8
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 8
; CHECK-NEXT: store double 9.000000e+00, ptr addrspace(1) [[TMP8]], align 8
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 9
; CHECK-NEXT: store double 1.000000e+01, ptr addrspace(1) [[TMP9]], align 8
; CHECK-NEXT: ret void
;
entry:
store [10 x double] [double 1.000000e+00, double 2.000000e+00, double 3.000000e+00, double 4.000000e+00, double 5.000000e+00, double 6.000000e+00, double 7.000000e+00, double 8.000000e+00, double 9.000000e+00, double 10.000000e+00], ptr addrspace(1) %arg, align 8
ret void
}
define spir_kernel void @f2(ptr addrspace(1) %arg) {
; CHECK-LABEL: define spir_kernel void @f2(
; CHECK-SAME: ptr addrspace(1) [[ARG:%.*]]) {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 0
; CHECK-NEXT: store double 1.000000e+00, ptr addrspace(1) [[TMP0]], align 8
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 1
; CHECK-NEXT: store double 1.000000e+00, ptr addrspace(1) [[TMP1]], align 8
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 2
; CHECK-NEXT: store double 1.000000e+00, ptr addrspace(1) [[TMP2]], align 8
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 3
; CHECK-NEXT: store double 1.000000e+00, ptr addrspace(1) [[TMP3]], align 8
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 4
; CHECK-NEXT: store double 1.000000e+00, ptr addrspace(1) [[TMP4]], align 8
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 5
; CHECK-NEXT: store double 1.000000e+00, ptr addrspace(1) [[TMP5]], align 8
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 6
; CHECK-NEXT: store double 1.000000e+00, ptr addrspace(1) [[TMP6]], align 8
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 7
; CHECK-NEXT: store double 1.000000e+00, ptr addrspace(1) [[TMP7]], align 8
; CHECK-NEXT: [[TMP8:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 8
; CHECK-NEXT: store double 1.000000e+00, ptr addrspace(1) [[TMP8]], align 8
; CHECK-NEXT: [[TMP9:%.*]] = getelementptr [10 x double], ptr addrspace(1) [[ARG]], i32 0, i32 9
; CHECK-NEXT: store double 1.000000e+00, ptr addrspace(1) [[TMP9]], align 8
; CHECK-NEXT: ret void
;
entry:
store [10 x double] [double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00, double 1.000000e+00], ptr addrspace(1) %arg, align 8
ret void
}
define spir_kernel void @f3(ptr addrspace(1) %arg) {
; CHECK-LABEL: define spir_kernel void @f3(
; CHECK-SAME: ptr addrspace(1) [[ARG:%.*]]) {
; CHECK-NEXT: [[ENTRY:.*:]]
; CHECK-NEXT: [[TMP0:%.*]] = getelementptr [8 x <2 x double>], ptr addrspace(1) [[ARG]], i32 0, i32 0
; CHECK-NEXT: store <2 x double> zeroinitializer, ptr addrspace(1) [[TMP0]], align 16
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr [8 x <2 x double>], ptr addrspace(1) [[ARG]], i32 0, i32 1
; CHECK-NEXT: store <2 x double> zeroinitializer, ptr addrspace(1) [[TMP1]], align 16
; CHECK-NEXT: [[TMP2:%.*]] = getelementptr [8 x <2 x double>], ptr addrspace(1) [[ARG]], i32 0, i32 2
; CHECK-NEXT: store <2 x double> zeroinitializer, ptr addrspace(1) [[TMP2]], align 16
; CHECK-NEXT: [[TMP3:%.*]] = getelementptr [8 x <2 x double>], ptr addrspace(1) [[ARG]], i32 0, i32 3
; CHECK-NEXT: store <2 x double> zeroinitializer, ptr addrspace(1) [[TMP3]], align 16
; CHECK-NEXT: [[TMP4:%.*]] = getelementptr [8 x <2 x double>], ptr addrspace(1) [[ARG]], i32 0, i32 4
; CHECK-NEXT: store <2 x double> zeroinitializer, ptr addrspace(1) [[TMP4]], align 16
; CHECK-NEXT: [[TMP5:%.*]] = getelementptr [8 x <2 x double>], ptr addrspace(1) [[ARG]], i32 0, i32 5
; CHECK-NEXT: store <2 x double> zeroinitializer, ptr addrspace(1) [[TMP5]], align 16
; CHECK-NEXT: [[TMP6:%.*]] = getelementptr [8 x <2 x double>], ptr addrspace(1) [[ARG]], i32 0, i32 6
; CHECK-NEXT: store <2 x double> zeroinitializer, ptr addrspace(1) [[TMP6]], align 16
; CHECK-NEXT: [[TMP7:%.*]] = getelementptr [8 x <2 x double>], ptr addrspace(1) [[ARG]], i32 0, i32 7
; CHECK-NEXT: store <2 x double> zeroinitializer, ptr addrspace(1) [[TMP7]], align 16
; CHECK-NEXT: ret void
;
entry:
store [8 x <2 x double>] [<2 x double> <double 0.000000e+00, double 0.000000e+00>, <2 x double> <double 0.000000e+00, double 0.000000e+00>, <2 x double> <double 0.000000e+00, double 0.000000e+00>, <2 x double> <double 0.000000e+00, double 0.000000e+00>, <2 x double> <double 0.000000e+00, double 0.000000e+00>, <2 x double> <double 0.000000e+00, double 0.000000e+00>, <2 x double> <double 0.000000e+00, double 0.000000e+00>, <2 x double> <double 0.000000e+00, double 0.000000e+00>], ptr addrspace(1) %arg, align 8
ret void
}
%simple.struct.type = type { i64, float }
%nested.struct.type = type { i32, %simple.struct.type }
%double.nested.struct.type = type { i32, %nested.struct.type }
; shouldn't create dead loads
define float @cover.dead.loads(%double.nested.struct.type* byval(%double.nested.struct.type) %struct.instance) #0
; CHECK-LABEL: define float @cover.dead.loads(
; CHECK-SAME: ptr byval([[DOUBLE_NESTED_STRUCT_TYPE:%.*]]) [[STRUCT_INSTANCE:%.*]]) #[[ATTR0:[0-9]+]] {
; CHECK-NEXT: [[PTR_TO_INNER_STRUCT:%.*]] = getelementptr [[DOUBLE_NESTED_STRUCT_TYPE]], ptr [[STRUCT_INSTANCE]], i32 0, i32 1
; CHECK-NEXT: [[TMP1:%.*]] = getelementptr [[NESTLE_STRUCT_TYPE:%.*]], ptr [[PTR_TO_INNER_STRUCT]], i32 0, i32 1, i32 1
; CHECK-NEXT: [[TMP2:%.*]] = load float, ptr [[TMP1]], align 4
; CHECK-NEXT: [[RESULT:%.*]] = fmul float [[TMP2]], 4.000000e+00
; CHECK-NEXT: ret float [[RESULT]]
;
{
%ptr.to.inner.struct = getelementptr %double.nested.struct.type, %double.nested.struct.type* %struct.instance, i32 0, i32 1
%inner.struct.value = load %nested.struct.type, %nested.struct.type* %ptr.to.inner.struct
%inner.struct.value.float = extractvalue %nested.struct.type %inner.struct.value, 1, 1
%result = fmul float %inner.struct.value.float, 4.0
ret float %result
}
attributes #0 = { optnone noinline }
|
