1; NOTE: Assertions have been autogenerated by utils/update_test_checks.py 2; RUN: opt -loop-vectorize -force-vector-width=4 -S %s | FileCheck %s 3 4define void @test(float* %A, i32 %x) { 5; CHECK-LABEL: @test( 6; CHECK-NEXT: entry: 7; CHECK-NEXT: br i1 false, label [[SCALAR_PH:%.*]], label [[VECTOR_SCEVCHECK:%.*]] 8; CHECK: vector.scevcheck: 9; CHECK-NEXT: [[IDENT_CHECK:%.*]] = icmp ne i32 [[X:%.*]], 1 10; CHECK-NEXT: br i1 [[IDENT_CHECK]], label [[SCALAR_PH]], label [[VECTOR_PH:%.*]] 11; CHECK: vector.ph: 12; CHECK-NEXT: br label [[VECTOR_BODY:%.*]] 13; CHECK: vector.body: 14; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, [[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], [[VECTOR_BODY]] ] 15; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[INDEX]] to i32 16; CHECK-NEXT: [[TMP2:%.*]] = add i32 [[TMP1]], 0 17; CHECK-NEXT: [[TMP3:%.*]] = add i64 [[INDEX]], 0 18; CHECK-NEXT: [[TMP4:%.*]] = add nuw nsw i64 [[TMP3]], 1 19; CHECK-NEXT: [[TMP5:%.*]] = trunc i64 [[TMP4]] to i32 20; CHECK-NEXT: [[TMP6:%.*]] = mul i32 [[TMP5]], [[X]] 21; CHECK-NEXT: [[TMP7:%.*]] = zext i32 [[TMP6]] to i64 22; CHECK-NEXT: [[TMP8:%.*]] = getelementptr inbounds float, float* [[A:%.*]], i64 [[TMP7]] 23; CHECK-NEXT: [[TMP9:%.*]] = getelementptr inbounds float, float* [[TMP8]], i32 0 24; CHECK-NEXT: [[TMP10:%.*]] = bitcast float* [[TMP9]] to <4 x float>* 25; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x float>, <4 x float>* [[TMP10]], align 4 26; CHECK-NEXT: [[TMP11:%.*]] = mul i32 [[TMP2]], [[X]] 27; CHECK-NEXT: [[TMP12:%.*]] = zext i32 [[TMP11]] to i64 28; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds float, float* [[A]], i64 [[TMP12]] 29; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds float, float* [[TMP13]], i32 0 30; CHECK-NEXT: [[TMP15:%.*]] = bitcast float* [[TMP14]] to <4 x float>* 31; CHECK-NEXT: store <4 x float> [[WIDE_LOAD]], <4 x float>* [[TMP15]], align 4 32; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4 33; CHECK-NEXT: [[TMP16:%.*]] = icmp eq i64 [[INDEX_NEXT]], undef 34; CHECK-NEXT: br i1 [[TMP16]], label [[MIDDLE_BLOCK:%.*]], label [[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]] 35; CHECK: middle.block: 36; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 undef, undef 37; CHECK-NEXT: br i1 [[CMP_N]], label [[EXIT:%.*]], label [[SCALAR_PH]] 38; CHECK: scalar.ph: 39; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ undef, [[MIDDLE_BLOCK]] ], [ 0, [[ENTRY:%.*]] ], [ 0, [[VECTOR_SCEVCHECK]] ] 40; CHECK-NEXT: br label [[LOOP:%.*]] 41; CHECK: loop: 42; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[IV_NEXT:%.*]], [[LOOP]] ], [ [[BC_RESUME_VAL]], [[SCALAR_PH]] ] 43; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1 44; CHECK-NEXT: [[T_IV_NEXT:%.*]] = trunc i64 [[IV_NEXT]] to i32 45; CHECK-NEXT: [[MUL_IV_NEXT:%.*]] = mul i32 [[T_IV_NEXT]], [[X]] 46; CHECK-NEXT: [[IDX_1:%.*]] = zext i32 [[MUL_IV_NEXT]] to i64 47; CHECK-NEXT: [[ARRAYIDX1215:%.*]] = getelementptr inbounds float, float* [[A]], i64 [[IDX_1]] 48; CHECK-NEXT: [[LV:%.*]] = load float, float* [[ARRAYIDX1215]], align 4 49; CHECK-NEXT: [[T_IV:%.*]] = trunc i64 [[IV]] to i32 50; CHECK-NEXT: [[MUL_IV:%.*]] = mul i32 [[T_IV]], [[X]] 51; CHECK-NEXT: [[IDX_2:%.*]] = zext i32 [[MUL_IV]] to i64 52; CHECK-NEXT: [[ARRAYIDX1209:%.*]] = getelementptr inbounds float, float* [[A]], i64 [[IDX_2]] 53; CHECK-NEXT: store float [[LV]], float* [[ARRAYIDX1209]], align 4 54; CHECK-NEXT: [[EC:%.*]] = icmp eq i64 [[IV_NEXT]], undef 55; CHECK-NEXT: br i1 [[EC]], label [[EXIT]], label [[LOOP]], !llvm.loop [[LOOP2:![0-9]+]] 56; CHECK: exit: 57; CHECK-NEXT: ret void 58; 59entry: 60 br label %loop 61 62loop: ; preds = %loop, %entry 63 %iv = phi i64 [ %iv.next, %loop ], [ 0, %entry ] 64 %iv.next = add nuw nsw i64 %iv, 1 65 %t.iv.next = trunc i64 %iv.next to i32 66 %mul.iv.next = mul i32 %t.iv.next, %x 67 %idx.1 = zext i32 %mul.iv.next to i64 68 %arrayidx1215 = getelementptr inbounds float, float* %A, i64 %idx.1 69 %lv = load float, float* %arrayidx1215, align 4 70 71 %t.iv = trunc i64 %iv to i32 72 %mul.iv = mul i32 %t.iv, %x 73 %idx.2 = zext i32 %mul.iv to i64 74 %arrayidx1209 = getelementptr inbounds float, float* %A, i64 %idx.2 75 store float %lv, float* %arrayidx1209, align 4 76 %ec = icmp eq i64 %iv.next, undef 77 br i1 %ec, label %exit, label %loop 78 79exit: ; preds = %loop 80 ret void 81} 82