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.
88 lines
2.3 KiB
LLVM
88 lines
2.3 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"
|
|
|
|
%struct.coordinate = type { i32, i32 }
|
|
|
|
; Make sure that we don't generate a wide load when accessing the struct.
|
|
; struct coordinate {
|
|
; int x;
|
|
; int y;
|
|
; };
|
|
;
|
|
;
|
|
; int foo(struct coordinate *A, int n) {
|
|
;
|
|
; int sum = 0;
|
|
; for (int i = 0; i < n; ++i)
|
|
; sum += A[i].x;
|
|
;
|
|
; return sum;
|
|
; }
|
|
|
|
;CHECK-LABEL: @foo(
|
|
;CHECK-NOT: load <4 x i32>
|
|
;CHECK: ret
|
|
define i32 @foo(ptr nocapture %A, i32 %n) readonly {
|
|
entry:
|
|
%cmp4 = icmp sgt i32 %n, 0
|
|
br i1 %cmp4, label %for.body, label %for.end
|
|
|
|
for.body:
|
|
%indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
|
|
%sum.05 = phi i32 [ %add, %for.body ], [ 0, %entry ]
|
|
%x = getelementptr inbounds %struct.coordinate, ptr %A, i64 %indvars.iv, i32 0
|
|
%0 = load i32, ptr %x, align 4
|
|
%add = add nsw i32 %0, %sum.05
|
|
%indvars.iv.next = add i64 %indvars.iv, 1
|
|
%lftr.wideiv = trunc i64 %indvars.iv.next to i32
|
|
%exitcond = icmp eq i32 %lftr.wideiv, %n
|
|
br i1 %exitcond, label %for.end, label %for.body
|
|
|
|
for.end:
|
|
%sum.0.lcssa = phi i32 [ 0, %entry ], [ %add, %for.body ]
|
|
ret i32 %sum.0.lcssa
|
|
}
|
|
|
|
%struct.lit = type { i32 }
|
|
|
|
; Verify that we still vectorize the access if the struct has the same size as
|
|
; the loaded element.
|
|
; struct lit {
|
|
; int x;
|
|
; };
|
|
;
|
|
;
|
|
; int bar(struct lit *A, int n) {
|
|
;
|
|
; int sum = 0;
|
|
; for (int i = 0; i < n; ++i)
|
|
; sum += A[i].x;
|
|
;
|
|
; return sum;
|
|
; }
|
|
|
|
;CHECK-LABEL: @bar(
|
|
;CHECK: load <4 x i32>
|
|
;CHECK: ret
|
|
define i32 @bar(ptr nocapture %A, i32 %n) readonly {
|
|
entry:
|
|
%cmp4 = icmp sgt i32 %n, 0
|
|
br i1 %cmp4, label %for.body, label %for.end
|
|
|
|
for.body:
|
|
%indvars.iv = phi i64 [ %indvars.iv.next, %for.body ], [ 0, %entry ]
|
|
%sum.05 = phi i32 [ %add, %for.body ], [ 0, %entry ]
|
|
%x = getelementptr inbounds %struct.lit, ptr %A, i64 %indvars.iv, i32 0
|
|
%0 = load i32, ptr %x, align 4
|
|
%add = add nsw i32 %0, %sum.05
|
|
%indvars.iv.next = add i64 %indvars.iv, 1
|
|
%lftr.wideiv = trunc i64 %indvars.iv.next to i32
|
|
%exitcond = icmp eq i32 %lftr.wideiv, %n
|
|
br i1 %exitcond, label %for.end, label %for.body
|
|
|
|
for.end:
|
|
%sum.0.lcssa = phi i32 [ 0, %entry ], [ %add, %for.body ]
|
|
ret i32 %sum.0.lcssa
|
|
}
|