a99dd8344e
Many tests were running extra passes after loop-vectorize, but they made no difference to the outcome. If tests don't require these extra passes, then we shouldn't run them because it increases the testing time.
44 lines
1.2 KiB
LLVM
44 lines
1.2 KiB
LLVM
; RUN: opt < %s -passes=loop-vectorize -force-vector-interleave=1 -force-vector-width=4 -S | FileCheck %s
|
|
|
|
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"
|
|
|
|
; PR15882: This test ensures that we do not produce wrapping arithmetic when
|
|
; creating constant reverse step vectors.
|
|
;
|
|
; int foo(int n, int *A) {
|
|
; int sum;
|
|
; for (int i=n; i > 0; i--)
|
|
; sum += A[i*2];
|
|
; return sum;
|
|
; }
|
|
;
|
|
|
|
;CHECK-LABEL: @foo(
|
|
;CHECK: sub <4 x i32> {{.*}}, <i32 0, i32 1, i32 2, i32 3>
|
|
;CHECK: ret
|
|
define i32 @foo(i32 %n, ptr nocapture %A) {
|
|
br label %.lr.ph
|
|
|
|
.lr.ph:
|
|
%2 = sext i32 %n to i64
|
|
br label %3
|
|
|
|
; <label>:
|
|
%indvars.iv = phi i64 [ %2, %.lr.ph ], [ %indvars.iv.next, %3 ]
|
|
%sum.01 = phi i32 [ 0, %.lr.ph ], [ %9, %3 ]
|
|
%4 = trunc i64 %indvars.iv to i32
|
|
%5 = shl nsw i32 %4, 1
|
|
%6 = sext i32 %5 to i64
|
|
%7 = getelementptr inbounds i32, ptr %A, i64 %6
|
|
%8 = load i32, ptr %7, align 4
|
|
%9 = add nsw i32 %8, %sum.01
|
|
%indvars.iv.next = add i64 %indvars.iv, -1
|
|
%10 = trunc i64 %indvars.iv.next to i32
|
|
%11 = icmp sgt i32 %10, 0
|
|
br i1 %11, label %3, label %._crit_edge
|
|
|
|
._crit_edge:
|
|
ret i32 %9
|
|
}
|
|
|