1 ; RUN: llc -verify-machineinstrs < %s -mtriple=powerpc-unknown-linux-gnu -mcpu=g4 -disable-ppc-ilp-pref | FileCheck %s
2 ; RUN: llc -verify-machineinstrs < %s -mtriple=powerpc64-unknown-linux-gnu -mcpu=g4 -disable-ppc-ilp-pref | FileCheck %s
5 target datalayout = "E-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v128:128:128-n32:64"
6 target triple = "powerpc64-unknown-linux-gnu"
8 @a = common global [32000 x float] zeroinitializer, align 16
9 @b = common global [32000 x float] zeroinitializer, align 16
10 @c = common global [32000 x float] zeroinitializer, align 16
11 @d = common global [32000 x float] zeroinitializer, align 16
12 @e = common global [32000 x float] zeroinitializer, align 16
13 @aa = common global [256 x [256 x float]] zeroinitializer, align 16
14 @bb = common global [256 x [256 x float]] zeroinitializer, align 16
15 @cc = common global [256 x [256 x float]] zeroinitializer, align 16
17 @.str11 = private unnamed_addr constant [6 x i8] c"s122 \00", align 1
18 @.str152 = private unnamed_addr constant [14 x i8] c"S122\09 %.2f \09\09\00", align 1
20 declare i32 @printf(i8* nocapture, ...) nounwind
21 declare i32 @init(i8* %name) nounwind
22 declare i64 @clock() nounwind
23 declare i32 @dummy(float*, float*, float*, float*, float*, [256 x float]*, [256 x float]*, [256 x float]*, float)
24 declare void @check(i32 %name) nounwind
29 define i32 @s122(i32 %n1, i32 %n3) nounwind {
31 %call = tail call i32 @init(i8* getelementptr inbounds ([6 x i8], [6 x i8]* @.str11, i64 0, i64 0))
32 %call1 = tail call i64 @clock() nounwind
33 %sub = add nsw i32 %n1, -1
34 %cmp316 = icmp slt i32 %sub, 32000
35 br i1 %cmp316, label %entry.split.us, label %for.end.7
37 entry.split.us: ; preds = %entry
38 %0 = sext i32 %sub to i64
39 %1 = sext i32 %n3 to i64
40 br label %for.body4.lr.ph.us
42 for.body4.us: ; preds = %for.body4.lr.ph.us, %for.body4.us
43 %indvars.iv20 = phi i64 [ 0, %for.body4.lr.ph.us ], [ %indvars.iv.next21, %for.body4.us ]
44 %indvars.iv = phi i64 [ %0, %for.body4.lr.ph.us ], [ %indvars.iv.next, %for.body4.us ]
45 %indvars.iv.next21 = add i64 %indvars.iv20, 1
46 %sub5.us = sub i64 31999, %indvars.iv20
47 %sext = shl i64 %sub5.us, 32
48 %idxprom.us = ashr exact i64 %sext, 32
49 %arrayidx.us = getelementptr inbounds [32000 x float], [32000 x float]* @b, i64 0, i64 %idxprom.us
50 %2 = load float, float* %arrayidx.us, align 4
51 %arrayidx7.us = getelementptr inbounds [32000 x float], [32000 x float]* @a, i64 0, i64 %indvars.iv
52 %3 = load float, float* %arrayidx7.us, align 4
53 %add8.us = tail call float asm "fadd $0, $1, $2", "=f,f,f,~{cr2}"(float %3, float %2)
54 store float %add8.us, float* %arrayidx7.us, align 4
55 %indvars.iv.next = add i64 %indvars.iv, %1
56 %4 = trunc i64 %indvars.iv.next to i32
57 %cmp3.us = icmp slt i32 %4, 32000
58 br i1 %cmp3.us, label %for.body4.us, label %for.body4.lr.ph.us.1
60 for.body4.lr.ph.us: ; preds = %entry.split.us, %for.end.us.4
61 %nl.019.us = phi i32 [ 0, %entry.split.us ], [ %inc.us.4, %for.end.us.4 ]
62 br label %for.body4.us
64 for.end12: ; preds = %for.end.7, %for.end.us.4
65 %call13 = tail call i64 @clock() nounwind
66 %sub14 = sub nsw i64 %call13, %call1
67 %conv = sitofp i64 %sub14 to double
68 %div = fdiv double %conv, 1.000000e+06
69 %call15 = tail call i32 (i8*, ...) @printf(i8* getelementptr inbounds ([14 x i8], [14 x i8]* @.str152, i64 0, i64 0), double %div) nounwind
70 tail call void @check(i32 1)
73 for.body4.lr.ph.us.1: ; preds = %for.body4.us
74 %call10.us = tail call i32 @dummy(float* getelementptr inbounds ([32000 x float], [32000 x float]* @a, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @b, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @c, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @d, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @e, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @aa, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @bb, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @cc, i64 0, i64 0), float 0.000000e+00) nounwind
75 br label %for.body4.us.1
77 for.body4.us.1: ; preds = %for.body4.us.1, %for.body4.lr.ph.us.1
78 %indvars.iv20.1 = phi i64 [ 0, %for.body4.lr.ph.us.1 ], [ %indvars.iv.next21.1, %for.body4.us.1 ]
79 %indvars.iv.1 = phi i64 [ %0, %for.body4.lr.ph.us.1 ], [ %indvars.iv.next.1, %for.body4.us.1 ]
80 %indvars.iv.next21.1 = add i64 %indvars.iv20.1, 1
81 %sub5.us.1 = sub i64 31999, %indvars.iv20.1
82 %sext23 = shl i64 %sub5.us.1, 32
83 %idxprom.us.1 = ashr exact i64 %sext23, 32
84 %arrayidx.us.1 = getelementptr inbounds [32000 x float], [32000 x float]* @b, i64 0, i64 %idxprom.us.1
85 %5 = load float, float* %arrayidx.us.1, align 4
86 %arrayidx7.us.1 = getelementptr inbounds [32000 x float], [32000 x float]* @a, i64 0, i64 %indvars.iv.1
87 %6 = load float, float* %arrayidx7.us.1, align 4
88 %add8.us.1 = fadd float %6, %5
89 store float %add8.us.1, float* %arrayidx7.us.1, align 4
90 %indvars.iv.next.1 = add i64 %indvars.iv.1, %1
91 %7 = trunc i64 %indvars.iv.next.1 to i32
92 %cmp3.us.1 = icmp slt i32 %7, 32000
93 br i1 %cmp3.us.1, label %for.body4.us.1, label %for.body4.lr.ph.us.2
95 for.body4.lr.ph.us.2: ; preds = %for.body4.us.1
96 %call10.us.1 = tail call i32 @dummy(float* getelementptr inbounds ([32000 x float], [32000 x float]* @a, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @b, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @c, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @d, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @e, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @aa, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @bb, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @cc, i64 0, i64 0), float 0.000000e+00) nounwind
97 br label %for.body4.us.2
99 for.body4.us.2: ; preds = %for.body4.us.2, %for.body4.lr.ph.us.2
100 %indvars.iv20.2 = phi i64 [ 0, %for.body4.lr.ph.us.2 ], [ %indvars.iv.next21.2, %for.body4.us.2 ]
101 %indvars.iv.2 = phi i64 [ %0, %for.body4.lr.ph.us.2 ], [ %indvars.iv.next.2, %for.body4.us.2 ]
102 %indvars.iv.next21.2 = add i64 %indvars.iv20.2, 1
103 %sub5.us.2 = sub i64 31999, %indvars.iv20.2
104 %sext24 = shl i64 %sub5.us.2, 32
105 %idxprom.us.2 = ashr exact i64 %sext24, 32
106 %arrayidx.us.2 = getelementptr inbounds [32000 x float], [32000 x float]* @b, i64 0, i64 %idxprom.us.2
107 %8 = load float, float* %arrayidx.us.2, align 4
108 %arrayidx7.us.2 = getelementptr inbounds [32000 x float], [32000 x float]* @a, i64 0, i64 %indvars.iv.2
109 %9 = load float, float* %arrayidx7.us.2, align 4
110 %add8.us.2 = fadd float %9, %8
111 store float %add8.us.2, float* %arrayidx7.us.2, align 4
112 %indvars.iv.next.2 = add i64 %indvars.iv.2, %1
113 %10 = trunc i64 %indvars.iv.next.2 to i32
114 %cmp3.us.2 = icmp slt i32 %10, 32000
115 br i1 %cmp3.us.2, label %for.body4.us.2, label %for.body4.lr.ph.us.3
117 for.body4.lr.ph.us.3: ; preds = %for.body4.us.2
118 %call10.us.2 = tail call i32 @dummy(float* getelementptr inbounds ([32000 x float], [32000 x float]* @a, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @b, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @c, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @d, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @e, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @aa, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @bb, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @cc, i64 0, i64 0), float 0.000000e+00) nounwind
119 br label %for.body4.us.3
121 for.body4.us.3: ; preds = %for.body4.us.3, %for.body4.lr.ph.us.3
122 %indvars.iv20.3 = phi i64 [ 0, %for.body4.lr.ph.us.3 ], [ %indvars.iv.next21.3, %for.body4.us.3 ]
123 %indvars.iv.3 = phi i64 [ %0, %for.body4.lr.ph.us.3 ], [ %indvars.iv.next.3, %for.body4.us.3 ]
124 %indvars.iv.next21.3 = add i64 %indvars.iv20.3, 1
125 %sub5.us.3 = sub i64 31999, %indvars.iv20.3
126 %sext25 = shl i64 %sub5.us.3, 32
127 %idxprom.us.3 = ashr exact i64 %sext25, 32
128 %arrayidx.us.3 = getelementptr inbounds [32000 x float], [32000 x float]* @b, i64 0, i64 %idxprom.us.3
129 %11 = load float, float* %arrayidx.us.3, align 4
130 %arrayidx7.us.3 = getelementptr inbounds [32000 x float], [32000 x float]* @a, i64 0, i64 %indvars.iv.3
131 %12 = load float, float* %arrayidx7.us.3, align 4
132 %add8.us.3 = fadd float %12, %11
133 store float %add8.us.3, float* %arrayidx7.us.3, align 4
134 %indvars.iv.next.3 = add i64 %indvars.iv.3, %1
135 %13 = trunc i64 %indvars.iv.next.3 to i32
136 %cmp3.us.3 = icmp slt i32 %13, 32000
137 br i1 %cmp3.us.3, label %for.body4.us.3, label %for.body4.lr.ph.us.4
139 for.body4.lr.ph.us.4: ; preds = %for.body4.us.3
140 %call10.us.3 = tail call i32 @dummy(float* getelementptr inbounds ([32000 x float], [32000 x float]* @a, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @b, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @c, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @d, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @e, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @aa, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @bb, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @cc, i64 0, i64 0), float 0.000000e+00) nounwind
141 br label %for.body4.us.4
143 for.body4.us.4: ; preds = %for.body4.us.4, %for.body4.lr.ph.us.4
144 %indvars.iv20.4 = phi i64 [ 0, %for.body4.lr.ph.us.4 ], [ %indvars.iv.next21.4, %for.body4.us.4 ]
145 %indvars.iv.4 = phi i64 [ %0, %for.body4.lr.ph.us.4 ], [ %indvars.iv.next.4, %for.body4.us.4 ]
146 %indvars.iv.next21.4 = add i64 %indvars.iv20.4, 1
147 %sub5.us.4 = sub i64 31999, %indvars.iv20.4
148 %sext26 = shl i64 %sub5.us.4, 32
149 %idxprom.us.4 = ashr exact i64 %sext26, 32
150 %arrayidx.us.4 = getelementptr inbounds [32000 x float], [32000 x float]* @b, i64 0, i64 %idxprom.us.4
151 %14 = load float, float* %arrayidx.us.4, align 4
152 %arrayidx7.us.4 = getelementptr inbounds [32000 x float], [32000 x float]* @a, i64 0, i64 %indvars.iv.4
153 %15 = load float, float* %arrayidx7.us.4, align 4
154 %add8.us.4 = fadd float %15, %14
155 store float %add8.us.4, float* %arrayidx7.us.4, align 4
156 %indvars.iv.next.4 = add i64 %indvars.iv.4, %1
157 %16 = trunc i64 %indvars.iv.next.4 to i32
158 %cmp3.us.4 = icmp slt i32 %16, 32000
159 br i1 %cmp3.us.4, label %for.body4.us.4, label %for.end.us.4
161 for.end.us.4: ; preds = %for.body4.us.4
162 %call10.us.4 = tail call i32 @dummy(float* getelementptr inbounds ([32000 x float], [32000 x float]* @a, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @b, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @c, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @d, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @e, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @aa, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @bb, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @cc, i64 0, i64 0), float 0.000000e+00) nounwind
163 %inc.us.4 = add nsw i32 %nl.019.us, 5
164 %exitcond.4 = icmp eq i32 %inc.us.4, 200000
165 br i1 %exitcond.4, label %for.end12, label %for.body4.lr.ph.us
167 for.end.7: ; preds = %entry, %for.end.7
168 %nl.019 = phi i32 [ %inc.7, %for.end.7 ], [ 0, %entry ]
169 %call10 = tail call i32 @dummy(float* getelementptr inbounds ([32000 x float], [32000 x float]* @a, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @b, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @c, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @d, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @e, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @aa, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @bb, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @cc, i64 0, i64 0), float 0.000000e+00) nounwind
170 %call10.1 = tail call i32 @dummy(float* getelementptr inbounds ([32000 x float], [32000 x float]* @a, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @b, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @c, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @d, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @e, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @aa, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @bb, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @cc, i64 0, i64 0), float 0.000000e+00) nounwind
171 %call10.2 = tail call i32 @dummy(float* getelementptr inbounds ([32000 x float], [32000 x float]* @a, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @b, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @c, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @d, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @e, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @aa, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @bb, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @cc, i64 0, i64 0), float 0.000000e+00) nounwind
172 %call10.3 = tail call i32 @dummy(float* getelementptr inbounds ([32000 x float], [32000 x float]* @a, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @b, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @c, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @d, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @e, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @aa, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @bb, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @cc, i64 0, i64 0), float 0.000000e+00) nounwind
173 %call10.4 = tail call i32 @dummy(float* getelementptr inbounds ([32000 x float], [32000 x float]* @a, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @b, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @c, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @d, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @e, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @aa, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @bb, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @cc, i64 0, i64 0), float 0.000000e+00) nounwind
174 %call10.5 = tail call i32 @dummy(float* getelementptr inbounds ([32000 x float], [32000 x float]* @a, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @b, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @c, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @d, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @e, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @aa, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @bb, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @cc, i64 0, i64 0), float 0.000000e+00) nounwind
175 %call10.6 = tail call i32 @dummy(float* getelementptr inbounds ([32000 x float], [32000 x float]* @a, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @b, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @c, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @d, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @e, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @aa, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @bb, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @cc, i64 0, i64 0), float 0.000000e+00) nounwind
176 %call10.7 = tail call i32 @dummy(float* getelementptr inbounds ([32000 x float], [32000 x float]* @a, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @b, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @c, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @d, i64 0, i64 0), float* getelementptr inbounds ([32000 x float], [32000 x float]* @e, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @aa, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @bb, i64 0, i64 0), [256 x float]* getelementptr inbounds ([256 x [256 x float]], [256 x [256 x float]]* @cc, i64 0, i64 0), float 0.000000e+00) nounwind
177 %inc.7 = add nsw i32 %nl.019, 8
178 %exitcond.7 = icmp eq i32 %inc.7, 200000
179 br i1 %exitcond.7, label %for.end12, label %for.end.7
182 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i1) nounwind
184 declare i32 @puts(i8* nocapture) nounwind
186 !3 = !{!"branch_weights", i32 64, i32 4}