82 lines
3.8 KiB
LLVM
82 lines
3.8 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
|
|
; RUN: opt -S < %s -load-store-vectorizer | FileCheck %s
|
|
; RUN: opt -S < %s -passes='function(load-store-vectorizer)' | FileCheck %s
|
|
|
|
declare void @llvm.sideeffect()
|
|
|
|
; load-store vectorization across a @llvm.sideeffect.
|
|
|
|
define void @test_sideeffect(float* %p) {
|
|
; CHECK-LABEL: @test_sideeffect(
|
|
; CHECK-NEXT: [[P0:%.*]] = getelementptr float, float* [[P:%.*]], i64 0
|
|
; CHECK-NEXT: [[TMP1:%.*]] = bitcast float* [[P0]] to <4 x float>*
|
|
; CHECK-NEXT: [[TMP2:%.*]] = load <4 x float>, <4 x float>* [[TMP1]], align 16
|
|
; CHECK-NEXT: [[L01:%.*]] = extractelement <4 x float> [[TMP2]], i32 0
|
|
; CHECK-NEXT: [[L12:%.*]] = extractelement <4 x float> [[TMP2]], i32 1
|
|
; CHECK-NEXT: [[L23:%.*]] = extractelement <4 x float> [[TMP2]], i32 2
|
|
; CHECK-NEXT: [[L34:%.*]] = extractelement <4 x float> [[TMP2]], i32 3
|
|
; CHECK-NEXT: call void @llvm.sideeffect()
|
|
; CHECK-NEXT: call void @llvm.sideeffect()
|
|
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <4 x float> poison, float [[L01]], i32 0
|
|
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x float> [[TMP3]], float [[L12]], i32 1
|
|
; CHECK-NEXT: [[TMP5:%.*]] = insertelement <4 x float> [[TMP4]], float [[L23]], i32 2
|
|
; CHECK-NEXT: [[TMP6:%.*]] = insertelement <4 x float> [[TMP5]], float [[L34]], i32 3
|
|
; CHECK-NEXT: [[TMP7:%.*]] = bitcast float* [[P0]] to <4 x float>*
|
|
; CHECK-NEXT: store <4 x float> [[TMP6]], <4 x float>* [[TMP7]], align 16
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
%p0 = getelementptr float, float* %p, i64 0
|
|
%p1 = getelementptr float, float* %p, i64 1
|
|
%p2 = getelementptr float, float* %p, i64 2
|
|
%p3 = getelementptr float, float* %p, i64 3
|
|
%l0 = load float, float* %p0, align 16
|
|
%l1 = load float, float* %p1
|
|
%l2 = load float, float* %p2
|
|
call void @llvm.sideeffect()
|
|
%l3 = load float, float* %p3
|
|
store float %l0, float* %p0, align 16
|
|
call void @llvm.sideeffect()
|
|
store float %l1, float* %p1
|
|
store float %l2, float* %p2
|
|
store float %l3, float* %p3
|
|
ret void
|
|
}
|
|
|
|
declare void @foo()
|
|
|
|
define void @test_inaccessiblememonly(float* %p) {
|
|
; CHECK-LABEL: @test_inaccessiblememonly(
|
|
; CHECK-NEXT: [[P0:%.*]] = getelementptr float, float* [[P:%.*]], i64 0
|
|
; CHECK-NEXT: [[TMP1:%.*]] = bitcast float* [[P0]] to <4 x float>*
|
|
; CHECK-NEXT: [[TMP2:%.*]] = load <4 x float>, <4 x float>* [[TMP1]], align 16
|
|
; CHECK-NEXT: [[L01:%.*]] = extractelement <4 x float> [[TMP2]], i32 0
|
|
; CHECK-NEXT: [[L12:%.*]] = extractelement <4 x float> [[TMP2]], i32 1
|
|
; CHECK-NEXT: [[L23:%.*]] = extractelement <4 x float> [[TMP2]], i32 2
|
|
; CHECK-NEXT: [[L34:%.*]] = extractelement <4 x float> [[TMP2]], i32 3
|
|
; CHECK-NEXT: call void @foo() #[[ATTR1:[0-9]+]]
|
|
; CHECK-NEXT: call void @foo() #[[ATTR1]]
|
|
; CHECK-NEXT: [[TMP3:%.*]] = insertelement <4 x float> poison, float [[L01]], i32 0
|
|
; CHECK-NEXT: [[TMP4:%.*]] = insertelement <4 x float> [[TMP3]], float [[L12]], i32 1
|
|
; CHECK-NEXT: [[TMP5:%.*]] = insertelement <4 x float> [[TMP4]], float [[L23]], i32 2
|
|
; CHECK-NEXT: [[TMP6:%.*]] = insertelement <4 x float> [[TMP5]], float [[L34]], i32 3
|
|
; CHECK-NEXT: [[TMP7:%.*]] = bitcast float* [[P0]] to <4 x float>*
|
|
; CHECK-NEXT: store <4 x float> [[TMP6]], <4 x float>* [[TMP7]], align 16
|
|
; CHECK-NEXT: ret void
|
|
;
|
|
%p0 = getelementptr float, float* %p, i64 0
|
|
%p1 = getelementptr float, float* %p, i64 1
|
|
%p2 = getelementptr float, float* %p, i64 2
|
|
%p3 = getelementptr float, float* %p, i64 3
|
|
%l0 = load float, float* %p0, align 16
|
|
%l1 = load float, float* %p1
|
|
%l2 = load float, float* %p2
|
|
call void @foo() inaccessiblememonly nounwind
|
|
%l3 = load float, float* %p3
|
|
store float %l0, float* %p0, align 16
|
|
call void @foo() inaccessiblememonly nounwind
|
|
store float %l1, float* %p1
|
|
store float %l2, float* %p2
|
|
store float %l3, float* %p3
|
|
ret void
|
|
}
|