1 ; RUN: opt %loadPolly %defaultOpts -polly-codegen -enable-polly-vector -dce -S %s | FileCheck %s
2 ; Obtained from C source as:
3 ; clang -S -emit-llvm -O0 no-fast-math-vect16.c
4 ; opt -correlated-propagation -mem2reg -instcombine -loop-simplify -indvars -instnamer no-fast-math-vect16.s > no-fast-math-vect16.ll
6 ; ModuleID = 'no-fast-math-vect16.s'
7 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-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
8 target triple = "x86_64-unknown-linux-gnu"
10 @main1.b = internal unnamed_addr constant [16 x float] [float 0.000000e+00, float 3.000000e+00, float 6.000000e+00, float 9.000000e+00, float 1.200000e+01, float 1.500000e+01, float 1.800000e+01, float 2.100000e+01, float 2.400000e+01, float 2.700000e+01, float 3.000000e+01, float 3.300000e+01, float 3.600000e+01, float 3.900000e+01, float 4.200000e+01, float 4.500000e+01], align 16
11 @main1.c = internal unnamed_addr constant [16 x float] [float 0.000000e+00, float 1.000000e+00, float 2.000000e+00, float 3.000000e+00, float 4.000000e+00, float 5.000000e+00, float 6.000000e+00, float 7.000000e+00, float 8.000000e+00, float 9.000000e+00, float 1.000000e+01, float 1.100000e+01, float 1.200000e+01, float 1.300000e+01, float 1.400000e+01, float 1.500000e+01], align 16
13 define i32 @main1() nounwind uwtable noinline {
15 %b = alloca [16 x float], align 16
16 %c = alloca [16 x float], align 16
17 %tmp = bitcast [16 x float]* %b to i8*
18 call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp, i8* bitcast ([16 x float]* @main1.b to i8*), i64 64, i32 16, i1 false)
19 %tmp2 = bitcast [16 x float]* %c to i8*
20 call void @llvm.memcpy.p0i8.p0i8.i64(i8* %tmp2, i8* bitcast ([16 x float]* @main1.c to i8*), i64 64, i32 16, i1 false)
23 for.cond: ; preds = %for.inc, %entry
24 %indvar = phi i64 [ %indvar.next, %for.inc ], [ 0, %entry ]
25 %diff.0 = phi float [ 0.000000e+00, %entry ], [ %add, %for.inc ]
26 %arrayidx9 = getelementptr [16 x float]* %c, i64 0, i64 %indvar
27 %arrayidx = getelementptr [16 x float]* %b, i64 0, i64 %indvar
28 %exitcond = icmp ne i64 %indvar, 16
29 br i1 %exitcond, label %for.body, label %for.end
31 for.body: ; preds = %for.cond
34 for.inc: ; preds = %for.body
35 %tmp10 = load float* %arrayidx9, align 4
36 %tmp6 = load float* %arrayidx, align 4
37 %sub = fsub float %tmp6, %tmp10
38 %add = fadd float %diff.0, %sub
39 %indvar.next = add i64 %indvar, 1
42 for.end: ; preds = %for.cond
43 %diff.0.lcssa = phi float [ %diff.0, %for.cond ]
44 %cmp14 = fcmp une float %diff.0.lcssa, 2.400000e+02
45 br i1 %cmp14, label %if.then, label %if.end
47 if.then: ; preds = %for.end
48 call void @abort() noreturn nounwind
51 if.end: ; preds = %for.end
55 declare void @llvm.memcpy.p0i8.p0i8.i64(i8* nocapture, i8* nocapture, i64, i32, i1) nounwind
57 declare void @abort() noreturn
59 define i32 @main() nounwind uwtable {
61 call void @check_vect()
62 %call = call i32 @main1()
66 define internal void @check_vect() nounwind uwtable noinline {
68 %a = alloca i32, align 4
69 %b = alloca i32, align 4
70 %c = alloca i32, align 4
71 %d = alloca i32, align 4
72 %call = call void (i32)* (i32, void (i32)*)* @signal(i32 4, void (i32)* @sig_ill_handler) nounwind
73 %call1 = call i32 @__get_cpuid(i32 1, i32* %a, i32* %b, i32* %c, i32* %d)
74 %tobool = icmp eq i32 %call1, 0
75 br i1 %tobool, label %if.then, label %lor.lhs.false
77 lor.lhs.false: ; preds = %entry
78 %tmp4 = load i32* %d, align 4
79 %and6 = and i32 %tmp4, 67108864
80 %cmp = icmp eq i32 %and6, 0
81 br i1 %cmp, label %if.then, label %if.end
83 if.then: ; preds = %entry, %lor.lhs.false
84 call void @exit(i32 0) noreturn nounwind
87 if.end: ; preds = %lor.lhs.false
88 %call7 = call void (i32)* (i32, void (i32)*)* @signal(i32 4, void (i32)* null) nounwind
92 declare void (i32)* @signal(i32, void (i32)*) nounwind
94 define internal void @sig_ill_handler(i32 %sig) nounwind uwtable {
96 call void @exit(i32 0) noreturn nounwind
99 return: ; No predecessors!
103 define internal i32 @__get_cpuid(i32 %__level, i32* %__eax, i32* %__ebx, i32* %__ecx, i32* %__edx) nounwind uwtable inlinehint {
105 %and = and i32 %__level, -2147483648
106 %call = call i32 @__get_cpuid_max(i32 %and, i32* null)
107 %cmp = icmp ult i32 %call, %__level
108 br i1 %cmp, label %if.then, label %if.end
110 if.then: ; preds = %entry
113 if.end: ; preds = %entry
114 %tmp = call { i32, i32, i32, i32 } asm "cpuid\0A\09", "={ax},={bx},={cx},={dx},0,~{dirflag},~{fpsr},~{flags}"(i32 %__level) nounwind, !srcloc !0
115 %asmresult = extractvalue { i32, i32, i32, i32 } %tmp, 0
116 %asmresult8 = extractvalue { i32, i32, i32, i32 } %tmp, 1
117 %asmresult9 = extractvalue { i32, i32, i32, i32 } %tmp, 2
118 %asmresult10 = extractvalue { i32, i32, i32, i32 } %tmp, 3
119 store i32 %asmresult, i32* %__eax, align 4
120 store i32 %asmresult8, i32* %__ebx, align 4
121 store i32 %asmresult9, i32* %__ecx, align 4
122 store i32 %asmresult10, i32* %__edx, align 4
125 return: ; preds = %if.end, %if.then
126 %retval.0 = phi i32 [ 0, %if.then ], [ 1, %if.end ]
130 declare void @exit(i32) noreturn
132 define internal i32 @__get_cpuid_max(i32 %__ext, i32* %__sig) nounwind uwtable inlinehint {
134 %tmp = call { i32, i32, i32, i32 } asm "cpuid\0A\09", "={ax},={bx},={cx},={dx},0,~{dirflag},~{fpsr},~{flags}"(i32 %__ext) nounwind, !srcloc !1
135 %asmresult = extractvalue { i32, i32, i32, i32 } %tmp, 0
136 %tobool = icmp eq i32* %__sig, null
137 br i1 %tobool, label %if.end, label %if.then
139 if.then: ; preds = %entry
140 %asmresult1 = extractvalue { i32, i32, i32, i32 } %tmp, 1
141 store i32 %asmresult1, i32* %__sig, align 4
144 if.end: ; preds = %entry, %if.then
148 !0 = metadata !{i32 -2147343641, i32 -2147343633}
149 !1 = metadata !{i32 -2147343764, i32 -2147343756}