1 ; RUN: opt < %s -loop-vectorize -mtriple=x86_64-unknown-linux -S -pass-remarks='loop-vectorize' -pass-remarks-missed='loop-vectorize' -pass-remarks-analysis='loop-vectorize' 2>&1 | FileCheck %s 2 3 ; CHECK: remark: no_fpmath.c:6:11: loop not vectorized: cannot prove it is safe to reorder floating-point operations 4 ; CHECK: remark: no_fpmath.c:6:14: loop not vectorized: 5 ; CHECK: remark: no_fpmath.c:17:14: vectorized loop (vectorization width: 2, interleaved count: 2) 6 7 target datalayout = "e-m:o-i64:64-f80:128-n8:16:32:64-S128" 8 target triple = "x86_64-apple-macosx10.10.0" 9 10 ; Function Attrs: nounwind readonly ssp uwtable 11 define double @cond_sum(i32* nocapture readonly %v, i32 %n) #0 !dbg !4 { 12 entry: 13 %cmp.7 = icmp sgt i32 %n, 0, !dbg !3 14 br i1 %cmp.7, label %for.body.preheader, label %for.cond.cleanup, !dbg !8 15 16 for.body.preheader: ; preds = %entry 17 br label %for.body, !dbg !9 18 19 for.cond.cleanup.loopexit: ; preds = %for.body 20 %add.lcssa = phi double [ %add, %for.body ] 21 br label %for.cond.cleanup, !dbg !10 22 23 for.cond.cleanup: ; preds = %for.cond.cleanup.loopexit, %entry 24 %a.0.lcssa = phi double [ 0.000000e+00, %entry ], [ %add.lcssa, %for.cond.cleanup.loopexit ] 25 ret double %a.0.lcssa, !dbg !10 26 27 for.body: ; preds = %for.body.preheader, %for.body 28 %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %for.body.preheader ] 29 %a.08 = phi double [ %add, %for.body ], [ 0.000000e+00, %for.body.preheader ] 30 %arrayidx = getelementptr inbounds i32, i32* %v, i64 %indvars.iv, !dbg !9 31 %0 = load i32, i32* %arrayidx, align 4, !dbg !9, !tbaa !11 32 %cmp1 = icmp eq i32 %0, 0, !dbg !15 33 %cond = select i1 %cmp1, double 3.400000e+00, double 1.150000e+00, !dbg !9 34 %add = fadd double %a.08, %cond, !dbg !16 35 %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1, !dbg !8 36 %lftr.wideiv = trunc i64 %indvars.iv.next to i32, !dbg !8 37 %exitcond = icmp eq i32 %lftr.wideiv, %n, !dbg !8 38 br i1 %exitcond, label %for.cond.cleanup.loopexit, label %for.body, !dbg !8, !llvm.loop !17 39 } 40 41 ; Function Attrs: nounwind readonly ssp uwtable 42 define double @cond_sum_loop_hint(i32* nocapture readonly %v, i32 %n) #0 !dbg !20 { 43 entry: 44 %cmp.7 = icmp sgt i32 %n, 0, !dbg !19 45 br i1 %cmp.7, label %for.body.preheader, label %for.cond.cleanup, !dbg !21 46 47 for.body.preheader: ; preds = %entry 48 br label %for.body, !dbg !22 49 50 for.cond.cleanup.loopexit: ; preds = %for.body 51 %add.lcssa = phi double [ %add, %for.body ] 52 br label %for.cond.cleanup, !dbg !23 53 54 for.cond.cleanup: ; preds = %for.cond.cleanup.loopexit, %entry 55 %a.0.lcssa = phi double [ 0.000000e+00, %entry ], [ %add.lcssa, %for.cond.cleanup.loopexit ] 56 ret double %a.0.lcssa, !dbg !23 57 58 for.body: ; preds = %for.body.preheader, %for.body 59 %indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %for.body.preheader ] 60 %a.08 = phi double [ %add, %for.body ], [ 0.000000e+00, %for.body.preheader ] 61 %arrayidx = getelementptr inbounds i32, i32* %v, i64 %indvars.iv, !dbg !22 62 %0 = load i32, i32* %arrayidx, align 4, !dbg !22, !tbaa !11 63 %cmp1 = icmp eq i32 %0, 0, !dbg !24 64 %cond = select i1 %cmp1, double 3.400000e+00, double 1.150000e+00, !dbg !22 65 %add = fadd double %a.08, %cond, !dbg !25 66 %indvars.iv.next = add nuw nsw i64 %indvars.iv, 1, !dbg !21 67 %lftr.wideiv = trunc i64 %indvars.iv.next to i32, !dbg !21 68 %exitcond = icmp eq i32 %lftr.wideiv, %n, !dbg !21 69 br i1 %exitcond, label %for.cond.cleanup.loopexit, label %for.body, !dbg !21, !llvm.loop !26 70 } 71 72 attributes #0 = { nounwind } 73 74 !llvm.module.flags = !{!0, !1} 75 !llvm.ident = !{!2} 76 77 !0 = !{i32 2, !"Debug Info Version", i32 3} 78 !1 = !{i32 1, !"PIC Level", i32 2} 79 !2 = !{!"clang version 3.7.0"} 80 !3 = !DILocation(line: 5, column: 20, scope: !4) 81 !4 = distinct !DISubprogram(name: "cond_sum", scope: !5, file: !5, line: 1, type: !6, isLocal: false, isDefinition: true, scopeLine: 1, flags: DIFlagPrototyped, isOptimized: true, variables: !7) 82 !5 = !DIFile(filename: "no_fpmath.c", directory: "") 83 !6 = !DISubroutineType(types: !7) 84 !7 = !{} 85 !8 = !DILocation(line: 5, column: 3, scope: !4) 86 !9 = !DILocation(line: 6, column: 14, scope: !4) 87 !10 = !DILocation(line: 9, column: 3, scope: !4) 88 !11 = !{!12, !12, i64 0} 89 !12 = !{!"int", !13, i64 0} 90 !13 = !{!"omnipotent char", !14, i64 0} 91 !14 = !{!"Simple C/C++ TBAA"} 92 !15 = !DILocation(line: 6, column: 19, scope: !4) 93 !16 = !DILocation(line: 6, column: 11, scope: !4) 94 !17 = distinct !{!17, !18} 95 !18 = !{!"llvm.loop.unroll.disable"} 96 !19 = !DILocation(line: 16, column: 20, scope: !20) 97 !20 = distinct !DISubprogram(name: "cond_sum_loop_hint", scope: !5, file: !5, line: 12, type: !6, isLocal: false, isDefinition: true, scopeLine: 12, flags: DIFlagPrototyped, isOptimized: true, variables: !7) 98 !21 = !DILocation(line: 16, column: 3, scope: !20) 99 !22 = !DILocation(line: 17, column: 14, scope: !20) 100 !23 = !DILocation(line: 20, column: 3, scope: !20) 101 !24 = !DILocation(line: 17, column: 19, scope: !20) 102 !25 = !DILocation(line: 17, column: 11, scope: !20) 103 !26 = distinct !{!26, !27, !18} 104 !27 = !{!"llvm.loop.vectorize.enable", i1 true} 105
