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
|
;=========================== begin_copyright_notice ============================
;
; Copyright (C) 2022 Intel Corporation
;
; SPDX-License-Identifier: MIT
;
;============================ end_copyright_notice =============================
;
; REQUIRES: regkeys
; RUN: igc_opt --igc-loopcount --regkey=EnableKernelCostDebug=1 -S < %s 2>&1 | FileCheck %s
; ------------------------------------------------
; LoopCountAnalysis
; ------------------------------------------------
define spir_kernel void @test(i32 addrspace(1)* %d, i32 %N) {
;CHECK-LABEL: Loop at depth 1 with header for.body
;CHECK: Argument symbol found with index 1 isInDirect = 0
;CHECK; Total number of loops: 1
;
entry:
%cmp1 = icmp sgt i32 %N, 0
br i1 %cmp1, label %for.body.preheader, label %entry.for.end_crit_edge
entry.for.end_crit_edge: ; preds = %entry
br label %for.end
for.body.preheader: ; preds = %entry
br label %for.body
for.body: ; preds = %for.body.for.body_crit_edge, %for.body.preheader
%0 = phi i32 addrspace(1)* [ %3, %for.body.for.body_crit_edge ], [ %d, %for.body.preheader ]
%i.02 = phi i32 [ %inc, %for.body.for.body_crit_edge ], [ 0, %for.body.preheader ]
store i32 %i.02, i32 addrspace(1)* %0, align 4
%inc = add nuw nsw i32 %i.02, 1
%cmp = icmp slt i32 %inc, %N
%1 = ptrtoint i32 addrspace(1)* %0 to i64
%2 = add i64 %1, 4
%3 = inttoptr i64 %2 to i32 addrspace(1)*
br i1 %cmp, label %for.body.for.body_crit_edge, label %for.end.loopexit
for.body.for.body_crit_edge: ; preds = %for.body
br label %for.body
for.end.loopexit: ; preds = %for.body
br label %for.end
for.end: ; preds = %entry.for.end_crit_edge, %for.end.loopexit
ret void
}
define spir_kernel void @test_2(i32 addrspace(1)* %d, i32 addrspace(1)* %s){
;CHECK-LABEL: Loop at depth 1 with header for.body
;CHECK: Argument symbol found with index 1 isInDirect = 1
;CHECK; Total number of loops: 1
;
entry:
%0 = ptrtoint i32 addrspace(1)* %s to i64
%1 = add i64 %0, 4
%2 = inttoptr i64 %1 to i32 addrspace(1)*
%3 = load i32, i32 addrspace(1)* %2, align 4
%cmp1.not = icmp eq i32 %3, 0
br i1 %cmp1.not, label %entry.for.end_crit_edge, label %for.body.preheader
entry.for.end_crit_edge: ; preds = %entry
br label %for.end
for.body.preheader: ; preds = %entry
br label %for.body
for.body: ; preds = %for.body.for.body_crit_edge, %for.body.preheader
%4 = phi i32 addrspace(1)* [ %7, %for.body.for.body_crit_edge ], [ %d, %for.body.preheader ]
%i.02 = phi i32 [ %inc, %for.body.for.body_crit_edge ], [ 0, %for.body.preheader ]
store i32 %i.02, i32 addrspace(1)* %4, align 4
%inc = add nuw nsw i32 %i.02, 1
%cmp = icmp ult i32 %inc, %3
%5 = ptrtoint i32 addrspace(1)* %4 to i64
%6 = add i64 %5, 4
%7 = inttoptr i64 %6 to i32 addrspace(1)*
br i1 %cmp, label %for.body.for.body_crit_edge, label %for.end.loopexit
for.body.for.body_crit_edge: ; preds = %for.body
br label %for.body
for.end.loopexit: ; preds = %for.body
br label %for.end
for.end: ; preds = %entry.for.end_crit_edge, %for.end.loopexit
ret void
}
|
