brintos

brintos / llvm-project-archived public Read only

0
0
Text · 2.8 KiB · a56efe8 Raw
76 lines · plain
1; RUN: opt -passes='require<profile-summary>,function(codegenprepare)' -passes=load-store-vectorizer %s -S -o - | FileCheck %s2; RUN: opt                 -passes=load-store-vectorizer %s -S -o - | FileCheck %s3; RUN: opt -aa-pipeline=basic-aa -passes='function(load-store-vectorizer)' %s -S -o - | FileCheck %s4 5target triple = "x86_64--"6 7%union = type { { [4 x [4 x [4 x [16 x float]]]], [4 x [4 x [4 x [16 x float]]]], [10 x [10 x [4 x float]]] } }8 9@global_pointer = external unnamed_addr global { %union, [2000 x i8] }, align 410 11; Function Attrs: convergent nounwind12define void @test(i32 %base) #0 {13; CHECK-LABEL: @test(14; CHECK-NOT: load i3215; CHECK: load <2 x i32>16; CHECK-NOT: load i3217entry:18  %mul331 = and i32 %base, -419  %add350.4 = add i32 4, %mul33120  %idx351.4 = zext i32 %add350.4 to i6421  %arrayidx352.4 = getelementptr inbounds { %union, [2000 x i8] }, ptr @global_pointer, i64 0, i32 0, i32 0, i32 1, i64 0, i64 0, i64 0, i64 %idx351.422  %add350.5 = add i32 5, %mul33123  %idx351.5 = zext i32 %add350.5 to i6424  %arrayidx352.5 = getelementptr inbounds { %union, [2000 x i8] }, ptr @global_pointer, i64 0, i32 0, i32 0, i32 1, i64 0, i64 0, i64 0, i64 %idx351.525  %cnd = icmp ult i32 %base, 100026  br i1 %cnd, label %loads, label %exit27 28loads:29  ; If and only if the loads are in a different BB from the GEPs codegenprepare30  ; would try to turn the GEPs into math, which makes LoadStoreVectorizer's job31  ; harder32  %tmp297.4 = load i32, ptr %arrayidx352.4, align 4, !tbaa !033  %tmp297.5 = load i32, ptr %arrayidx352.5, align 4, !tbaa !034  br label %exit35 36exit:37  ret void38}39 40; Function Attrs: convergent nounwind41define void @test.codegenprepared(i32 %base) #0 {42; CHECK-LABEL: @test.codegenprepared(43; CHECK-NOT: load i3244; CHECK: load <2 x i32>45; CHECK-NOT: load i3246entry:47  %mul331 = and i32 %base, -448  %add350.4 = add i32 4, %mul33149  %idx351.4 = zext i32 %add350.4 to i6450  %add350.5 = add i32 5, %mul33151  %idx351.5 = zext i32 %add350.5 to i6452  %cnd = icmp ult i32 %base, 100053  br i1 %cnd, label %loads, label %exit54 55loads:                                            ; preds = %entry56  %sunkaddr = mul i64 %idx351.4, 457  %sunkaddr1 = getelementptr inbounds i8, ptr @global_pointer, i64 %sunkaddr58  %sunkaddr2 = getelementptr inbounds i8, ptr %sunkaddr1, i64 409659  %tmp297.4 = load i32, ptr %sunkaddr2, align 4, !tbaa !060  %sunkaddr3 = mul i64 %idx351.5, 461  %sunkaddr4 = getelementptr inbounds i8, ptr @global_pointer, i64 %sunkaddr362  %sunkaddr5 = getelementptr inbounds i8, ptr %sunkaddr4, i64 409663  %tmp297.5 = load i32, ptr %sunkaddr5, align 4, !tbaa !064  br label %exit65 66exit:                                             ; preds = %loads, %entry67  ret void68}69 70attributes #0 = { convergent nounwind }71 72!0 = !{!1, !1, i64 0}73!1 = !{!"float", !2, i64 0}74!2 = !{!"omnipotent char", !3, i64 0}75!3 = !{!"Simple C++ TBAA"}76