1; RUN: opt < %s  -loop-vectorize -force-vector-interleave=1 -force-vector-width=4 -dce -instcombine -S | FileCheck %s
2
3; int __attribute__((noinline)) sum_array(int *A, int n) {
4;  return std::accumulate(A, A + n, 0);
5; }
6
7target datalayout = "e-p:64:64:64-p1:16:16:16-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-n8:16:32:64-S128"
8
9;CHECK-LABEL: @sum_array(
10;CHECK: phi i64
11;CHECK: phi <4 x i32>
12;CHECK: load <4 x i32>
13;CHECK: add nsw <4 x i32>
14;CHECK: ret i32
15define i32 @sum_array(i32* %A, i32 %n) nounwind uwtable readonly noinline ssp {
16  %1 = sext i32 %n to i64
17  %2 = getelementptr inbounds i32, i32* %A, i64 %1
18  %3 = icmp eq i32 %n, 0
19  br i1 %3, label %_ZSt10accumulateIPiiET0_T_S2_S1_.exit, label %.lr.ph.i
20
21.lr.ph.i:                                         ; preds = %0, %.lr.ph.i
22  %.03.i = phi i32* [ %6, %.lr.ph.i ], [ %A, %0 ]
23  %.012.i = phi i32 [ %5, %.lr.ph.i ], [ 0, %0 ]
24  %4 = load i32, i32* %.03.i, align 4
25  %5 = add nsw i32 %4, %.012.i
26  %6 = getelementptr inbounds i32, i32* %.03.i, i64 1
27  %7 = icmp eq i32* %6, %2
28  br i1 %7, label %_ZSt10accumulateIPiiET0_T_S2_S1_.exit, label %.lr.ph.i
29
30_ZSt10accumulateIPiiET0_T_S2_S1_.exit:            ; preds = %.lr.ph.i, %0
31  %.01.lcssa.i = phi i32 [ 0, %0 ], [ %5, %.lr.ph.i ]
32  ret i32 %.01.lcssa.i
33}
34
35; Same, but use a pointer with a different size.
36;CHECK-LABEL: @sum_array_as1(
37;CHECK: phi i16
38;CHECK: phi <4 x i32>
39;CHECK: load <4 x i32>
40;CHECK: add nsw <4 x i32>
41;CHECK: ret i32
42define i32 @sum_array_as1(i32 addrspace(1)* %A, i32 %n) nounwind uwtable readonly noinline ssp {
43  %1 = sext i32 %n to i64
44  %2 = getelementptr inbounds i32, i32 addrspace(1)* %A, i64 %1
45  %3 = icmp eq i32 %n, 0
46  br i1 %3, label %_ZSt10accumulateIPiiET0_T_S2_S1_.exit, label %.lr.ph.i
47
48.lr.ph.i:                                         ; preds = %0, %.lr.ph.i
49  %.03.i = phi i32 addrspace(1)* [ %6, %.lr.ph.i ], [ %A, %0 ]
50  %.012.i = phi i32 [ %5, %.lr.ph.i ], [ 0, %0 ]
51  %4 = load i32, i32 addrspace(1)* %.03.i, align 4
52  %5 = add nsw i32 %4, %.012.i
53  %6 = getelementptr inbounds i32, i32 addrspace(1)* %.03.i, i64 1
54  %7 = icmp eq i32 addrspace(1)* %6, %2
55  br i1 %7, label %_ZSt10accumulateIPiiET0_T_S2_S1_.exit, label %.lr.ph.i
56
57_ZSt10accumulateIPiiET0_T_S2_S1_.exit:            ; preds = %.lr.ph.i, %0
58  %.01.lcssa.i = phi i32 [ 0, %0 ], [ %5, %.lr.ph.i ]
59  ret i32 %.01.lcssa.i
60}
61