xref: /llvm-project/llvm/test/CodeGen/PowerPC/lsr-profitable-chain.ll (revision c4a60c9d34375e73fc2da5e02215eabe4bc90e8f) 
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
2; RUN: llc -verify-machineinstrs -mtriple=powerpc64le-unknown-linux-gnu \
3; RUN:   -mcpu=pwr9 < %s | FileCheck %s
4
5define void @foo(ptr readonly %0, ptr %1, i64 %2, i64 %3, i64 %4, i64 %5, i64 %6, i64 %7) {
6; CHECK-LABEL: foo:
7; CHECK:       # %bb.0:
8; CHECK-NEXT:    cmpd 5, 7
9; CHECK-NEXT:    bgelr 0
10; CHECK-NEXT:  # %bb.1: # %.preheader
11; CHECK-NEXT:    std 27, -40(1) # 8-byte Folded Spill
12; CHECK-NEXT:    addi 27, 5, 2
13; CHECK-NEXT:    std 28, -32(1) # 8-byte Folded Spill
14; CHECK-NEXT:    addi 28, 5, 3
15; CHECK-NEXT:    std 30, -16(1) # 8-byte Folded Spill
16; CHECK-NEXT:    addi 30, 5, 1
17; CHECK-NEXT:    mulld 12, 8, 5
18; CHECK-NEXT:    mulld 0, 9, 8
19; CHECK-NEXT:    std 29, -24(1) # 8-byte Folded Spill
20; CHECK-NEXT:    addi 29, 3, 16
21; CHECK-NEXT:    sldi 11, 10, 3
22; CHECK-NEXT:    std 22, -80(1) # 8-byte Folded Spill
23; CHECK-NEXT:    std 23, -72(1) # 8-byte Folded Spill
24; CHECK-NEXT:    std 24, -64(1) # 8-byte Folded Spill
25; CHECK-NEXT:    std 25, -56(1) # 8-byte Folded Spill
26; CHECK-NEXT:    std 26, -48(1) # 8-byte Folded Spill
27; CHECK-NEXT:    mulld 30, 8, 30
28; CHECK-NEXT:    mulld 28, 8, 28
29; CHECK-NEXT:    mulld 8, 8, 27
30; CHECK-NEXT:    b .LBB0_3
31; CHECK-NEXT:    .p2align 4
32; CHECK-NEXT:  .LBB0_2:
33; CHECK-NEXT:    add 5, 5, 9
34; CHECK-NEXT:    add 12, 12, 0
35; CHECK-NEXT:    add 30, 30, 0
36; CHECK-NEXT:    add 28, 28, 0
37; CHECK-NEXT:    add 8, 8, 0
38; CHECK-NEXT:    cmpd 5, 7
39; CHECK-NEXT:    bge 0, .LBB0_6
40; CHECK-NEXT:  .LBB0_3: # =>This Loop Header: Depth=1
41; CHECK-NEXT:    # Child Loop BB0_5 Depth 2
42; CHECK-NEXT:    sub 27, 5, 10
43; CHECK-NEXT:    cmpd 6, 27
44; CHECK-NEXT:    bge 0, .LBB0_2
45; CHECK-NEXT:  # %bb.4:
46; CHECK-NEXT:    add 25, 6, 12
47; CHECK-NEXT:    add 24, 6, 8
48; CHECK-NEXT:    sldi 26, 6, 3
49; CHECK-NEXT:    sldi 23, 25, 3
50; CHECK-NEXT:    add 25, 6, 30
51; CHECK-NEXT:    sldi 24, 24, 3
52; CHECK-NEXT:    add 26, 4, 26
53; CHECK-NEXT:    sldi 22, 25, 3
54; CHECK-NEXT:    add 25, 6, 28
55; CHECK-NEXT:    add 24, 29, 24
56; CHECK-NEXT:    add 23, 3, 23
57; CHECK-NEXT:    sldi 25, 25, 3
58; CHECK-NEXT:    add 22, 3, 22
59; CHECK-NEXT:    add 25, 29, 25
60; CHECK-NEXT:    .p2align 5
61; CHECK-NEXT:  .LBB0_5: # Parent Loop BB0_3 Depth=1
62; CHECK-NEXT:    # => This Inner Loop Header: Depth=2
63; CHECK-NEXT:    lfd 0, 0(26)
64; CHECK-NEXT:    lfd 1, 0(23)
65; CHECK-NEXT:    add 6, 6, 10
66; CHECK-NEXT:    cmpd 6, 27
67; CHECK-NEXT:    xsadddp 0, 0, 1
68; CHECK-NEXT:    lfd 1, 8(23)
69; CHECK-NEXT:    xsadddp 0, 0, 1
70; CHECK-NEXT:    lfd 1, 16(23)
71; CHECK-NEXT:    xsadddp 0, 0, 1
72; CHECK-NEXT:    lfd 1, 24(23)
73; CHECK-NEXT:    add 23, 23, 11
74; CHECK-NEXT:    xsadddp 0, 0, 1
75; CHECK-NEXT:    lfd 1, 0(22)
76; CHECK-NEXT:    xsadddp 0, 0, 1
77; CHECK-NEXT:    lfd 1, 8(22)
78; CHECK-NEXT:    xsadddp 0, 0, 1
79; CHECK-NEXT:    lfd 1, 16(22)
80; CHECK-NEXT:    xsadddp 0, 0, 1
81; CHECK-NEXT:    lfd 1, 24(22)
82; CHECK-NEXT:    add 22, 22, 11
83; CHECK-NEXT:    xsadddp 0, 0, 1
84; CHECK-NEXT:    lfd 1, -16(24)
85; CHECK-NEXT:    xsadddp 0, 0, 1
86; CHECK-NEXT:    lfd 1, -8(24)
87; CHECK-NEXT:    xsadddp 0, 0, 1
88; CHECK-NEXT:    lfd 1, 0(24)
89; CHECK-NEXT:    xsadddp 0, 0, 1
90; CHECK-NEXT:    lfd 1, 8(24)
91; CHECK-NEXT:    add 24, 24, 11
92; CHECK-NEXT:    xsadddp 0, 0, 1
93; CHECK-NEXT:    lfd 1, -16(25)
94; CHECK-NEXT:    xsadddp 0, 0, 1
95; CHECK-NEXT:    lfd 1, -8(25)
96; CHECK-NEXT:    xsadddp 0, 0, 1
97; CHECK-NEXT:    lfd 1, 0(25)
98; CHECK-NEXT:    xsadddp 0, 0, 1
99; CHECK-NEXT:    lfd 1, 8(25)
100; CHECK-NEXT:    add 25, 25, 11
101; CHECK-NEXT:    xsadddp 0, 0, 1
102; CHECK-NEXT:    stfd 0, 0(26)
103; CHECK-NEXT:    add 26, 26, 11
104; CHECK-NEXT:    blt 0, .LBB0_5
105; CHECK-NEXT:    b .LBB0_2
106; CHECK-NEXT:  .LBB0_6:
107; CHECK-NEXT:    ld 30, -16(1) # 8-byte Folded Reload
108; CHECK-NEXT:    ld 29, -24(1) # 8-byte Folded Reload
109; CHECK-NEXT:    ld 28, -32(1) # 8-byte Folded Reload
110; CHECK-NEXT:    ld 27, -40(1) # 8-byte Folded Reload
111; CHECK-NEXT:    ld 26, -48(1) # 8-byte Folded Reload
112; CHECK-NEXT:    ld 25, -56(1) # 8-byte Folded Reload
113; CHECK-NEXT:    ld 24, -64(1) # 8-byte Folded Reload
114; CHECK-NEXT:    ld 23, -72(1) # 8-byte Folded Reload
115; CHECK-NEXT:    ld 22, -80(1) # 8-byte Folded Reload
116; CHECK-NEXT:    blr
117  %9 = icmp slt i64 %2, %4
118  br i1 %9, label %10, label %97
119
12010:                                               ; preds = %8, %93
121  %11 = phi i64 [ %95, %93 ], [ %2, %8 ]
122  %12 = phi i64 [ %94, %93 ], [ %3, %8 ]
123  %13 = sub nsw i64 %11, %7
124  %14 = icmp slt i64 %12, %13
125  br i1 %14, label %15, label %93
126
12715:                                               ; preds = %10
128  %16 = mul nsw i64 %11, %5
129  %17 = add nsw i64 %11, 1
130  %18 = mul nsw i64 %17, %5
131  %19 = add nsw i64 %11, 2
132  %20 = mul nsw i64 %19, %5
133  %21 = add nsw i64 %11, 3
134  %22 = mul nsw i64 %21, %5
135  br label %23
136
13723:                                               ; preds = %15, %23
138  %24 = phi i64 [ %12, %15 ], [ %91, %23 ]
139  %25 = getelementptr inbounds double, ptr %1, i64 %24
140  %26 = load double, ptr %25, align 8
141  %27 = add nsw i64 %24, %16
142  %28 = getelementptr inbounds double, ptr %0, i64 %27
143  %29 = load double, ptr %28, align 8
144  %30 = fadd double %26, %29
145  %31 = add nsw i64 %27, 1
146  %32 = getelementptr inbounds double, ptr %0, i64 %31
147  %33 = load double, ptr %32, align 8
148  %34 = fadd double %30, %33
149  %35 = add nsw i64 %27, 2
150  %36 = getelementptr inbounds double, ptr %0, i64 %35
151  %37 = load double, ptr %36, align 8
152  %38 = fadd double %34, %37
153  %39 = add nsw i64 %27, 3
154  %40 = getelementptr inbounds double, ptr %0, i64 %39
155  %41 = load double, ptr %40, align 8
156  %42 = fadd double %38, %41
157  %43 = add nsw i64 %24, %18
158  %44 = getelementptr inbounds double, ptr %0, i64 %43
159  %45 = load double, ptr %44, align 8
160  %46 = fadd double %42, %45
161  %47 = add nsw i64 %43, 1
162  %48 = getelementptr inbounds double, ptr %0, i64 %47
163  %49 = load double, ptr %48, align 8
164  %50 = fadd double %46, %49
165  %51 = add nsw i64 %43, 2
166  %52 = getelementptr inbounds double, ptr %0, i64 %51
167  %53 = load double, ptr %52, align 8
168  %54 = fadd double %50, %53
169  %55 = add nsw i64 %43, 3
170  %56 = getelementptr inbounds double, ptr %0, i64 %55
171  %57 = load double, ptr %56, align 8
172  %58 = fadd double %54, %57
173  %59 = add nsw i64 %24, %20
174  %60 = getelementptr inbounds double, ptr %0, i64 %59
175  %61 = load double, ptr %60, align 8
176  %62 = fadd double %58, %61
177  %63 = add nsw i64 %59, 1
178  %64 = getelementptr inbounds double, ptr %0, i64 %63
179  %65 = load double, ptr %64, align 8
180  %66 = fadd double %62, %65
181  %67 = add nsw i64 %59, 2
182  %68 = getelementptr inbounds double, ptr %0, i64 %67
183  %69 = load double, ptr %68, align 8
184  %70 = fadd double %66, %69
185  %71 = add nsw i64 %59, 3
186  %72 = getelementptr inbounds double, ptr %0, i64 %71
187  %73 = load double, ptr %72, align 8
188  %74 = fadd double %70, %73
189  %75 = add nsw i64 %24, %22
190  %76 = getelementptr inbounds double, ptr %0, i64 %75
191  %77 = load double, ptr %76, align 8
192  %78 = fadd double %74, %77
193  %79 = add nsw i64 %75, 1
194  %80 = getelementptr inbounds double, ptr %0, i64 %79
195  %81 = load double, ptr %80, align 8
196  %82 = fadd double %78, %81
197  %83 = add nsw i64 %75, 2
198  %84 = getelementptr inbounds double, ptr %0, i64 %83
199  %85 = load double, ptr %84, align 8
200  %86 = fadd double %82, %85
201  %87 = add nsw i64 %75, 3
202  %88 = getelementptr inbounds double, ptr %0, i64 %87
203  %89 = load double, ptr %88, align 8
204  %90 = fadd double %86, %89
205  store double %90, ptr %25, align 8
206  %91 = add nsw i64 %24, %7
207  %92 = icmp slt i64 %91, %13
208  br i1 %92, label %23, label %93
209
21093:                                               ; preds = %23, %10
211  %94 = phi i64 [ %12, %10 ], [ %91, %23 ]
212  %95 = add nsw i64 %11, %6
213  %96 = icmp slt i64 %95, %4
214  br i1 %96, label %10, label %97
215
21697:                                               ; preds = %93, %8
217  ret void
218}
219