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
|
;=========================== begin_copyright_notice ============================
;
; Copyright (C) 2022-2024 Intel Corporation
;
; SPDX-License-Identifier: MIT
;
;============================ end_copyright_notice =============================
; REQUIRES: llvm-14-plus
; RUN: igc_opt --opaque-pointers %s -S -o - -opt-atomics-pass | FileCheck %s
declare i32 @llvm.genx.GenISA.intatomicrawA64.i32.p1.p1(ptr addrspace(1), ptr addrspace(1), i32, i32)
declare i32 @llvm.genx.GenISA.icmpxchgatomicrawA64.i32.p1.p1(ptr addrspace(1), ptr addrspace(1), i32, i32)
define spir_kernel void @kernel1(ptr addrspace(1) %src) {
entry:
br label %back
back:
%call1 = call i32 @llvm.genx.GenISA.intatomicrawA64.i32.p1.p1(ptr addrspace(1) %src, ptr addrspace(1) %src, i32 0, i32 9)
%bc1 = bitcast i32 %call1 to float
%operation = fadd fast float %bc1, -2.000000e+00
%bc2 = bitcast float %operation to i32
%call2 = call i32 @llvm.genx.GenISA.icmpxchgatomicrawA64.i32.p1.p1(ptr addrspace(1) %src, ptr addrspace(1) %src, i32 %call1, i32 %bc2)
%cmp = icmp eq i32 %call1, %call2
br i1 %cmp, label %exit, label %back
exit:
ret void
}
!igc.functions = !{!0}
!0 = !{ptr @kernel1, !1}
!1 = !{!2}
!2 = !{!"function_type", i32 0}
; CHECK-LABEL: @kernel1(
; CHECK-NEXT: entry:
; CHECK-NEXT: br label %back
; CHECK: back:
; CHECK-NEXT: [[TMP0:%.*]] = call i16 @llvm.genx.GenISA.simdLaneId()
; CHECK-NEXT: [[TMP1:%.*]] = zext i16 [[TMP0]] to i32
; CHECK-NEXT: [[TMP2:%.*]] = call float @llvm.genx.GenISA.WaveAll.f32(float -2.000000e+00, i8 9, i32 0)
; CHECK-NEXT: [[TMP3:%.*]] = icmp eq i32 [[TMP1]], 0
; CHECK-NEXT: br i1 [[TMP3]], label %[[TMP4:.*]], label [[EXIT:%.*]]
; CHECK: [[TMP4]]:
; CHECK-NEXT: [[CALL1:%.*]] = call i32 @llvm.genx.GenISA.intatomicrawA64.i32.p1.p1(ptr addrspace(1) [[SRC:%.*]], ptr addrspace(1) [[SRC]], i32 0, i32 9)
; CHECK-NEXT: [[BC1:%.*]] = bitcast i32 [[CALL1]] to float
; CHECK-NEXT: [[OPERATION:%.*]] = fadd fast float [[BC1]], [[TMP2]]
; CHECK-NEXT: [[BC2:%.*]] = bitcast float [[OPERATION]] to i32
; CHECK-NEXT: [[CALL2:%.*]] = call i32 @llvm.genx.GenISA.icmpxchgatomicrawA64.i32.p1.p1(ptr addrspace(1) [[SRC]], ptr addrspace(1) [[SRC]], i32 [[CALL1]], i32 [[BC2]])
; CHECK-NEXT: [[CMP:%.*]] = icmp eq i32 [[CALL1]], [[CALL2]]
; CHECK-NEXT: br i1 [[CMP]], label [[EXIT]], label [[TMP4:%.*]]
; CHECK: exit:
; CHECK-NEXT: ret void
;
define spir_kernel void @kernel2(ptr addrspace(1) %src) {
entry:
br label %back
back:
%call1 = call i32 @llvm.genx.GenISA.intatomicrawA64.i32.p1.p1(ptr addrspace(1) %src, ptr addrspace(1) %src, i32 0, i32 9)
%bc1 = bitcast i32 %call1 to float
%operation = fadd float %bc1, -2.000000e+00
%bc2 = bitcast float %operation to i32
%call2 = call i32 @llvm.genx.GenISA.icmpxchgatomicrawA64.i32.p1.p1(ptr addrspace(1) %src, ptr addrspace(1) %src, i32 %call1, i32 %bc2)
%cmp = icmp eq i32 %call1, %call2
br i1 %cmp, label %exit, label %back
exit:
ret void
}
!igc.functions = !{!3}
!3 = !{ptr @kernel2, !4}
!4 = !{!5}
!5 = !{!"function_type", i32 0}
; Optimization should not be applied in this case. Since the flag for fast math was not passed.
;
; CHECK-LABEL: define spir_kernel void @kernel2(ptr addrspace(1) %src) {
; CHECK-NEXT: entry:
; CHECK-NEXT: br label %back
; CHECK: back:
; CHECK-NEXT: %call1 = call i32 @llvm.genx.GenISA.intatomicrawA64.i32.p1.p1(ptr addrspace(1) %src, ptr addrspace(1) %src, i32 0, i32 9)
; CHECK-NEXT: %bc1 = bitcast i32 %call1 to float
; CHECK-NEXT: %operation = fadd float %bc1, -2.000000e+00
; CHECK-NEXT: %bc2 = bitcast float %operation to i32
; CHECK-NEXT: %call2 = call i32 @llvm.genx.GenISA.icmpxchgatomicrawA64.i32.p1.p1(ptr addrspace(1) %src, ptr addrspace(1) %src, i32 %call1, i32 %bc2)
; CHECK-NEXT: %cmp = icmp eq i32 %call1, %call2
; CHECK-NEXT: br i1 %cmp, label %exit, label %back
; CHECK: exit:
; CHECK-NEXT: ret void
; CHECK-NEXT: }
;
|
