| Nadav Rotem | 93fa5ef | 2012-12-04 22:40:22 +0000 | [diff] [blame^] | 1 | ; RUN: opt < %s -loop-vectorize -force-vector-width=4 -enable-if-conversion -dce -instcombine -licm -S | FileCheck %s |
| 2 | |
| 3 | 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-S128" |
| 4 | target triple = "x86_64-apple-macosx10.9.0" |
| 5 | |
| 6 | ;CHECK: @reduction_func |
| 7 | ;CHECK-NOT: load <4 x i32> |
| 8 | ;CHECK: ret i32 |
| 9 | define i32 @reduction_func(i32* nocapture %A, i32 %n) nounwind uwtable readonly ssp { |
| 10 | entry: |
| 11 | %cmp10 = icmp sgt i32 %n, 0 |
| 12 | br i1 %cmp10, label %for.body, label %for.end |
| 13 | |
| 14 | for.body: ; preds = %entry, %for.inc |
| 15 | %indvars.iv = phi i64 [ %indvars.iv.next, %for.inc ], [ 0, %entry ] |
| 16 | %sum.011 = phi i32 [ %sum.1, %for.inc ], [ 0, %entry ] |
| 17 | %arrayidx = getelementptr inbounds i32* %A, i64 %indvars.iv |
| 18 | %0 = load i32* %arrayidx, align 4 |
| 19 | %cmp1 = icmp sgt i32 %0, 30 |
| 20 | br i1 %cmp1, label %if.then, label %for.inc |
| 21 | |
| 22 | if.then: ; preds = %for.body |
| 23 | %add = add i32 %sum.011, 2 |
| 24 | %add4 = add i32 %add, %0 |
| 25 | br label %for.inc |
| 26 | |
| 27 | for.inc: ; preds = %for.body, %if.then |
| 28 | %sum.1 = phi i32 [ %add4, %if.then ], [ %sum.011, %for.body ] |
| 29 | %indvars.iv.next = add i64 %indvars.iv, 1 |
| 30 | %lftr.wideiv = trunc i64 %indvars.iv.next to i32 |
| 31 | %exitcond = icmp eq i32 %lftr.wideiv, %n |
| 32 | br i1 %exitcond, label %for.end, label %for.body |
| 33 | |
| 34 | for.end: ; preds = %for.inc, %entry |
| 35 | %sum.0.lcssa = phi i32 [ 0, %entry ], [ 4, %for.inc ] |
| 36 | ret i32 %sum.0.lcssa |
| 37 | } |
| 38 | |