| ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 6 |
| ; RUN: opt -passes=loop-vectorize -force-vector-width=4 -S %s | FileCheck %s |
| |
| define void @replace_first_order_recurrence_with_versioned_iv_for_pointer_use(ptr %y, ptr %x, i32 %n) { |
| ; CHECK-LABEL: define void @replace_first_order_recurrence_with_versioned_iv_for_pointer_use( |
| ; CHECK-SAME: ptr [[Y:%.*]], ptr [[X:%.*]], i32 [[N:%.*]]) { |
| ; CHECK-NEXT: [[ENTRY:.*]]: |
| ; CHECK-NEXT: [[X2:%.*]] = ptrtoaddr ptr [[X]] to i64 |
| ; CHECK-NEXT: [[Y1:%.*]] = ptrtoaddr ptr [[Y]] to i64 |
| ; CHECK-NEXT: [[SMAX:%.*]] = call i32 @llvm.smax.i32(i32 [[N]], i32 1) |
| ; CHECK-NEXT: [[TMP0:%.*]] = zext nneg i32 [[SMAX]] to i64 |
| ; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], 4 |
| ; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]] |
| ; CHECK: [[VECTOR_MEMCHECK]]: |
| ; CHECK-NEXT: [[TMP8:%.*]] = add i64 [[Y1]], -8 |
| ; CHECK-NEXT: [[TMP9:%.*]] = sub i64 [[TMP8]], [[X2]] |
| ; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP9]], 32 |
| ; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]] |
| ; CHECK: [[VECTOR_PH]]: |
| ; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], 4 |
| ; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]] |
| ; CHECK-NEXT: [[IND_END:%.*]] = trunc i64 [[N_VEC]] to i32 |
| ; CHECK-NEXT: br label %[[VECTOR_BODY:.*]] |
| ; CHECK: [[VECTOR_BODY]]: |
| ; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[OFFSET_IDX:%.*]] = trunc i64 [[INDEX]] to i32 |
| ; CHECK-NEXT: [[TMP5:%.*]] = add i32 [[OFFSET_IDX]], 1 |
| ; CHECK-NEXT: [[TMP22:%.*]] = zext i32 [[TMP5]] to i64 |
| ; CHECK-NEXT: [[TMP29:%.*]] = getelementptr inbounds double, ptr [[X]], i64 [[TMP22]] |
| ; CHECK-NEXT: [[TMP18:%.*]] = getelementptr inbounds double, ptr [[Y]], i64 [[INDEX]] |
| ; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x double>, ptr [[TMP29]], align 8 |
| ; CHECK-NEXT: store <4 x double> [[WIDE_LOAD]], ptr [[TMP18]], align 8 |
| ; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4 |
| ; CHECK-NEXT: [[TMP11:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[TMP11]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]] |
| ; CHECK: [[MIDDLE_BLOCK]]: |
| ; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[CMP_N]], label %[[FOR_END:.*]], label %[[SCALAR_PH]] |
| ; CHECK: [[SCALAR_PH]]: |
| ; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[IND_END]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_MEMCHECK]] ] |
| ; CHECK-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_MEMCHECK]] ] |
| ; CHECK-NEXT: br label %[[FOR_BODY:.*]] |
| ; CHECK: [[FOR_BODY]]: |
| ; CHECK-NEXT: [[PHI32:%.*]] = phi i32 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[I32NEXT:%.*]], %[[FOR_BODY]] ] |
| ; CHECK-NEXT: [[SCALAR_RECUR:%.*]] = phi i64 [ [[BC_RESUME_VAL3]], %[[SCALAR_PH]] ], [ [[I64NEXT:%.*]], %[[FOR_BODY]] ] |
| ; CHECK-NEXT: [[I32NEXT]] = add i32 [[PHI32]], 1 |
| ; CHECK-NEXT: [[I64NEXT]] = zext i32 [[I32NEXT]] to i64 |
| ; CHECK-NEXT: [[XIP:%.*]] = getelementptr inbounds double, ptr [[X]], i64 [[I64NEXT]] |
| ; CHECK-NEXT: [[YIP:%.*]] = getelementptr inbounds double, ptr [[Y]], i64 [[SCALAR_RECUR]] |
| ; CHECK-NEXT: [[XI:%.*]] = load double, ptr [[XIP]], align 8 |
| ; CHECK-NEXT: store double [[XI]], ptr [[YIP]], align 8 |
| ; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[I32NEXT]], [[N]] |
| ; CHECK-NEXT: br i1 [[CMP]], label %[[FOR_BODY]], label %[[FOR_END]], !llvm.loop [[LOOP3:![0-9]+]] |
| ; CHECK: [[FOR_END]]: |
| ; CHECK-NEXT: ret void |
| ; |
| entry: |
| br label %for.body |
| |
| for.body: |
| %phi32 = phi i32 [ 0, %entry ], [ %i32next, %for.body ] |
| %phi64 = phi i64 [ 0, %entry ], [ %i64next, %for.body ] |
| %i32next = add i32 %phi32, 1 |
| %i64next = zext i32 %i32next to i64 |
| %xip = getelementptr inbounds double, ptr %x, i64 %i64next |
| %yip = getelementptr inbounds double, ptr %y, i64 %phi64 |
| %xi = load double, ptr %xip, align 8 |
| store double %xi, ptr %yip, align 8 |
| %cmp = icmp slt i32 %i32next, %n |
| br i1 %cmp, label %for.body, label %for.end |
| |
| for.end: |
| ret void |
| } |
| |
| define void @do_not_replace_first_order_recurrence_with_versioned_iv_for_pointer_use_btc_versioned(ptr %y, ptr %x, i64 %n) { |
| ; CHECK-LABEL: define void @do_not_replace_first_order_recurrence_with_versioned_iv_for_pointer_use_btc_versioned( |
| ; CHECK-SAME: ptr [[Y:%.*]], ptr [[X:%.*]], i64 [[N:%.*]]) { |
| ; CHECK-NEXT: [[ENTRY:.*]]: |
| ; CHECK-NEXT: [[X2:%.*]] = ptrtoaddr ptr [[X]] to i64 |
| ; CHECK-NEXT: [[Y1:%.*]] = ptrtoaddr ptr [[Y]] to i64 |
| ; CHECK-NEXT: [[SMAX3:%.*]] = call i64 @llvm.smax.i64(i64 [[N]], i64 1) |
| ; CHECK-NEXT: [[TMP0:%.*]] = trunc i64 [[SMAX3]] to i32 |
| ; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 4 |
| ; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_SCEVCHECK:.*]] |
| ; CHECK: [[VECTOR_SCEVCHECK]]: |
| ; CHECK-NEXT: [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 [[N]], i64 1) |
| ; CHECK-NEXT: [[TMP1:%.*]] = add nsw i64 [[SMAX]], -1 |
| ; CHECK-NEXT: [[TMP2:%.*]] = trunc i64 [[TMP1]] to i32 |
| ; CHECK-NEXT: [[TMP3:%.*]] = add i32 1, [[TMP2]] |
| ; CHECK-NEXT: [[TMP4:%.*]] = icmp ult i32 [[TMP3]], 1 |
| ; CHECK-NEXT: [[TMP5:%.*]] = icmp ugt i64 [[TMP1]], 4294967295 |
| ; CHECK-NEXT: [[TMP6:%.*]] = or i1 [[TMP4]], [[TMP5]] |
| ; CHECK-NEXT: br i1 [[TMP6]], label %[[SCALAR_PH]], label %[[VECTOR_MEMCHECK:.*]] |
| ; CHECK: [[VECTOR_MEMCHECK]]: |
| ; CHECK-NEXT: [[TMP8:%.*]] = add i64 [[Y1]], -8 |
| ; CHECK-NEXT: [[TMP9:%.*]] = sub i64 [[TMP8]], [[X2]] |
| ; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP9]], 32 |
| ; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]] |
| ; CHECK: [[VECTOR_PH]]: |
| ; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i32 [[TMP0]], 4 |
| ; CHECK-NEXT: [[N_VEC:%.*]] = sub i32 [[TMP0]], [[N_MOD_VF]] |
| ; CHECK-NEXT: br label %[[VECTOR_BODY:.*]] |
| ; CHECK: [[VECTOR_BODY]]: |
| ; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[VECTOR_RECUR:%.*]] = phi <4 x i64> [ <i64 poison, i64 poison, i64 poison, i64 0>, %[[VECTOR_PH]] ], [ [[TMP10:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[TMP11:%.*]] = add <4 x i32> [[VEC_IND]], splat (i32 1) |
| ; CHECK-NEXT: [[TMP10]] = zext <4 x i32> [[TMP11]] to <4 x i64> |
| ; CHECK-NEXT: [[TMP13:%.*]] = extractelement <4 x i64> [[TMP10]], i64 0 |
| ; CHECK-NEXT: [[TMP12:%.*]] = shufflevector <4 x i64> [[VECTOR_RECUR]], <4 x i64> [[TMP10]], <4 x i32> <i32 3, i32 4, i32 5, i32 6> |
| ; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds double, ptr [[X]], i64 [[TMP13]] |
| ; CHECK-NEXT: [[TMP15:%.*]] = extractelement <4 x i64> [[TMP12]], i64 0 |
| ; CHECK-NEXT: [[TMP16:%.*]] = getelementptr inbounds double, ptr [[Y]], i64 [[TMP15]] |
| ; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x double>, ptr [[TMP14]], align 8 |
| ; CHECK-NEXT: store <4 x double> [[WIDE_LOAD]], ptr [[TMP16]], align 8 |
| ; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4 |
| ; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 4) |
| ; CHECK-NEXT: [[TMP17:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[TMP17]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP4:![0-9]+]] |
| ; CHECK: [[MIDDLE_BLOCK]]: |
| ; CHECK-NEXT: [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i64> [[TMP10]], i64 3 |
| ; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[TMP0]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[CMP_N]], label %[[FOR_END:.*]], label %[[SCALAR_PH]] |
| ; CHECK: [[SCALAR_PH]]: |
| ; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ], [ 0, %[[VECTOR_MEMCHECK]] ] |
| ; CHECK-NEXT: [[SCALAR_RECUR_INIT:%.*]] = phi i64 [ [[VECTOR_RECUR_EXTRACT]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ], [ 0, %[[VECTOR_MEMCHECK]] ] |
| ; CHECK-NEXT: br label %[[FOR_BODY:.*]] |
| ; CHECK: [[FOR_BODY]]: |
| ; CHECK-NEXT: [[PHI32:%.*]] = phi i32 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[I32NEXT:%.*]], %[[FOR_BODY]] ] |
| ; CHECK-NEXT: [[SCALAR_RECUR:%.*]] = phi i64 [ [[SCALAR_RECUR_INIT]], %[[SCALAR_PH]] ], [ [[I64NEXT:%.*]], %[[FOR_BODY]] ] |
| ; CHECK-NEXT: [[I32NEXT]] = add i32 [[PHI32]], 1 |
| ; CHECK-NEXT: [[I64NEXT]] = zext i32 [[I32NEXT]] to i64 |
| ; CHECK-NEXT: [[XIP:%.*]] = getelementptr inbounds double, ptr [[X]], i64 [[I64NEXT]] |
| ; CHECK-NEXT: [[YIP:%.*]] = getelementptr inbounds double, ptr [[Y]], i64 [[SCALAR_RECUR]] |
| ; CHECK-NEXT: [[XI:%.*]] = load double, ptr [[XIP]], align 8 |
| ; CHECK-NEXT: store double [[XI]], ptr [[YIP]], align 8 |
| ; CHECK-NEXT: [[CMP:%.*]] = icmp slt i64 [[I64NEXT]], [[N]] |
| ; CHECK-NEXT: br i1 [[CMP]], label %[[FOR_BODY]], label %[[FOR_END]], !llvm.loop [[LOOP5:![0-9]+]] |
| ; CHECK: [[FOR_END]]: |
| ; CHECK-NEXT: ret void |
| ; |
| entry: |
| br label %for.body |
| |
| for.body: |
| %phi32 = phi i32 [ 0, %entry ], [ %i32next, %for.body ] |
| %phi64 = phi i64 [ 0, %entry ], [ %i64next, %for.body ] |
| %i32next = add i32 %phi32, 1 |
| %i64next = zext i32 %i32next to i64 |
| %xip = getelementptr inbounds double, ptr %x, i64 %i64next |
| %yip = getelementptr inbounds double, ptr %y, i64 %phi64 |
| %xi = load double, ptr %xip, align 8 |
| store double %xi, ptr %yip, align 8 |
| %cmp = icmp slt i64 %i64next, %n |
| br i1 %cmp, label %for.body, label %for.end |
| |
| for.end: |
| ret void |
| } |
| |
| define i64 @do_not_replace_first_order_recurrence_with_versioned_iv_for_wide_pointer_use(ptr %y, ptr %x, i32 %n) { |
| ; CHECK-LABEL: define i64 @do_not_replace_first_order_recurrence_with_versioned_iv_for_wide_pointer_use( |
| ; CHECK-SAME: ptr [[Y:%.*]], ptr [[X:%.*]], i32 [[N:%.*]]) { |
| ; CHECK-NEXT: [[ENTRY:.*]]: |
| ; CHECK-NEXT: [[X2:%.*]] = ptrtoaddr ptr [[X]] to i64 |
| ; CHECK-NEXT: [[Y1:%.*]] = ptrtoaddr ptr [[Y]] to i64 |
| ; CHECK-NEXT: [[SMAX:%.*]] = call i32 @llvm.smax.i32(i32 [[N]], i32 1) |
| ; CHECK-NEXT: [[TMP0:%.*]] = zext nneg i32 [[SMAX]] to i64 |
| ; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[TMP0]], 4 |
| ; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]] |
| ; CHECK: [[VECTOR_MEMCHECK]]: |
| ; CHECK-NEXT: [[TMP1:%.*]] = add i64 [[Y1]], -8 |
| ; CHECK-NEXT: [[TMP2:%.*]] = sub i64 [[TMP1]], [[X2]] |
| ; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP2]], 32 |
| ; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]] |
| ; CHECK: [[VECTOR_PH]]: |
| ; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[TMP0]], 4 |
| ; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[TMP0]], [[N_MOD_VF]] |
| ; CHECK-NEXT: [[IND_END:%.*]] = trunc i64 [[N_VEC]] to i32 |
| ; CHECK-NEXT: br label %[[VECTOR_BODY:.*]] |
| ; CHECK: [[VECTOR_BODY]]: |
| ; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[OFFSET_IDX:%.*]] = trunc i64 [[INDEX]] to i32 |
| ; CHECK-NEXT: [[TMP4:%.*]] = add <4 x i32> [[VEC_IND]], splat (i32 1) |
| ; CHECK-NEXT: [[TMP6:%.*]] = extractelement <4 x i32> [[TMP4]], i64 0 |
| ; CHECK-NEXT: [[TMP5:%.*]] = zext <4 x i32> [[TMP4]] to <4 x i64> |
| ; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i64, ptr [[X]], i32 [[TMP6]] |
| ; CHECK-NEXT: [[TMP11:%.*]] = getelementptr inbounds i64, ptr [[Y]], i32 [[OFFSET_IDX]] |
| ; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x i64>, ptr [[TMP7]], align 8 |
| ; CHECK-NEXT: [[TMP10:%.*]] = add <4 x i64> [[WIDE_LOAD]], [[TMP5]] |
| ; CHECK-NEXT: store <4 x i64> [[TMP10]], ptr [[TMP11]], align 8 |
| ; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4 |
| ; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 4) |
| ; CHECK-NEXT: [[TMP12:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[TMP12]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP6:![0-9]+]] |
| ; CHECK: [[MIDDLE_BLOCK]]: |
| ; CHECK-NEXT: [[IND_ESCAPE:%.*]] = sub i64 [[N_VEC]], 1 |
| ; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[TMP0]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[CMP_N]], label %[[FOR_END:.*]], label %[[SCALAR_PH]] |
| ; CHECK: [[SCALAR_PH]]: |
| ; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[IND_END]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_MEMCHECK]] ] |
| ; CHECK-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_MEMCHECK]] ] |
| ; CHECK-NEXT: br label %[[FOR_BODY:.*]] |
| ; CHECK: [[FOR_BODY]]: |
| ; CHECK-NEXT: [[PHI32:%.*]] = phi i32 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[I32NEXT:%.*]], %[[FOR_BODY]] ] |
| ; CHECK-NEXT: [[PHI64:%.*]] = phi i64 [ [[BC_RESUME_VAL3]], %[[SCALAR_PH]] ], [ [[I64NEXT:%.*]], %[[FOR_BODY]] ] |
| ; CHECK-NEXT: [[I32NEXT]] = add i32 [[PHI32]], 1 |
| ; CHECK-NEXT: [[I64NEXT]] = zext i32 [[I32NEXT]] to i64 |
| ; CHECK-NEXT: [[XIP:%.*]] = getelementptr inbounds i64, ptr [[X]], i32 [[I32NEXT]] |
| ; CHECK-NEXT: [[YIP:%.*]] = getelementptr inbounds i64, ptr [[Y]], i32 [[PHI32]] |
| ; CHECK-NEXT: [[XI:%.*]] = load i64, ptr [[XIP]], align 8 |
| ; CHECK-NEXT: [[ADD:%.*]] = add i64 [[XI]], [[I64NEXT]] |
| ; CHECK-NEXT: store i64 [[ADD]], ptr [[YIP]], align 8 |
| ; CHECK-NEXT: [[CMP:%.*]] = icmp slt i32 [[I32NEXT]], [[N]] |
| ; CHECK-NEXT: br i1 [[CMP]], label %[[FOR_BODY]], label %[[FOR_END]], !llvm.loop [[LOOP7:![0-9]+]] |
| ; CHECK: [[FOR_END]]: |
| ; CHECK-NEXT: [[PHI64_LCSSA:%.*]] = phi i64 [ [[PHI64]], %[[FOR_BODY]] ], [ [[IND_ESCAPE]], %[[MIDDLE_BLOCK]] ] |
| ; CHECK-NEXT: ret i64 [[PHI64_LCSSA]] |
| ; |
| entry: |
| br label %for.body |
| |
| for.body: |
| %phi32 = phi i32 [ 0, %entry ], [ %i32next, %for.body ] |
| %phi64 = phi i64 [ 0, %entry ], [ %i64next, %for.body ] |
| %i32next = add i32 %phi32, 1 |
| %i64next = zext i32 %i32next to i64 |
| %xip = getelementptr inbounds i64, ptr %x, i32 %i32next |
| %yip = getelementptr inbounds i64, ptr %y, i32 %phi32 |
| %xi = load i64, ptr %xip, align 8 |
| %add = add i64 %xi, %i64next |
| store i64 %add, ptr %yip, align 8 |
| %cmp = icmp slt i32 %i32next, %n |
| br i1 %cmp, label %for.body, label %for.end |
| |
| for.end: |
| ret i64 %phi64 |
| } |
| |
| define void @replace_first_order_recurrence_with_versioned_iv_for_uniform_non_pointer_use(ptr %y, ptr %x, i64 %n) { |
| ; CHECK-LABEL: define void @replace_first_order_recurrence_with_versioned_iv_for_uniform_non_pointer_use( |
| ; CHECK-SAME: ptr [[Y:%.*]], ptr [[X:%.*]], i64 [[N:%.*]]) { |
| ; CHECK-NEXT: [[ENTRY:.*]]: |
| ; CHECK-NEXT: [[X2:%.*]] = ptrtoaddr ptr [[X]] to i64 |
| ; CHECK-NEXT: [[Y1:%.*]] = ptrtoaddr ptr [[Y]] to i64 |
| ; CHECK-NEXT: [[SMAX3:%.*]] = call i64 @llvm.smax.i64(i64 [[N]], i64 1) |
| ; CHECK-NEXT: [[TMP0:%.*]] = trunc i64 [[SMAX3]] to i32 |
| ; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[TMP0]], 4 |
| ; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_SCEVCHECK:.*]] |
| ; CHECK: [[VECTOR_SCEVCHECK]]: |
| ; CHECK-NEXT: [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 [[N]], i64 1) |
| ; CHECK-NEXT: [[TMP1:%.*]] = add nsw i64 [[SMAX]], -1 |
| ; CHECK-NEXT: [[TMP2:%.*]] = trunc i64 [[TMP1]] to i32 |
| ; CHECK-NEXT: [[TMP3:%.*]] = add i32 1, [[TMP2]] |
| ; CHECK-NEXT: [[TMP4:%.*]] = icmp ult i32 [[TMP3]], 1 |
| ; CHECK-NEXT: [[TMP9:%.*]] = icmp ugt i64 [[TMP1]], 4294967295 |
| ; CHECK-NEXT: [[TMP16:%.*]] = or i1 [[TMP4]], [[TMP9]] |
| ; CHECK-NEXT: [[TMP11:%.*]] = trunc i64 [[TMP1]] to i32 |
| ; CHECK-NEXT: [[TMP12:%.*]] = add i32 1, [[TMP11]] |
| ; CHECK-NEXT: [[TMP13:%.*]] = icmp slt i32 [[TMP12]], 1 |
| ; CHECK-NEXT: [[TMP14:%.*]] = icmp ugt i64 [[TMP1]], 4294967295 |
| ; CHECK-NEXT: [[TMP15:%.*]] = or i1 [[TMP13]], [[TMP14]] |
| ; CHECK-NEXT: [[TMP17:%.*]] = or i1 [[TMP16]], [[TMP15]] |
| ; CHECK-NEXT: br i1 [[TMP17]], label %[[SCALAR_PH]], label %[[VECTOR_MEMCHECK:.*]] |
| ; CHECK: [[VECTOR_MEMCHECK]]: |
| ; CHECK-NEXT: [[TMP18:%.*]] = add i64 [[Y1]], -8 |
| ; CHECK-NEXT: [[TMP19:%.*]] = sub i64 [[TMP18]], [[X2]] |
| ; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP19]], 32 |
| ; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]] |
| ; CHECK: [[VECTOR_PH]]: |
| ; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i32 [[TMP0]], 4 |
| ; CHECK-NEXT: [[N_VEC:%.*]] = sub i32 [[TMP0]], [[N_MOD_VF]] |
| ; CHECK-NEXT: br label %[[VECTOR_BODY:.*]] |
| ; CHECK: [[VECTOR_BODY]]: |
| ; CHECK-NEXT: [[TMP20:%.*]] = phi i32 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[TMP21:%.*]] = add <4 x i32> [[VEC_IND]], splat (i32 1) |
| ; CHECK-NEXT: [[TMP24:%.*]] = extractelement <4 x i32> [[TMP21]], i64 0 |
| ; CHECK-NEXT: [[TMP25:%.*]] = getelementptr inbounds double, ptr [[X]], i32 [[TMP24]] |
| ; CHECK-NEXT: [[TMP26:%.*]] = getelementptr inbounds double, ptr [[Y]], i32 [[TMP20]] |
| ; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x double>, ptr [[TMP25]], align 8 |
| ; CHECK-NEXT: store <4 x double> [[WIDE_LOAD]], ptr [[TMP26]], align 8 |
| ; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[TMP20]], 4 |
| ; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 4) |
| ; CHECK-NEXT: [[TMP22:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[TMP22]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP8:![0-9]+]] |
| ; CHECK: [[MIDDLE_BLOCK]]: |
| ; CHECK-NEXT: [[TMP23:%.*]] = zext <4 x i32> [[TMP21]] to <4 x i64> |
| ; CHECK-NEXT: [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i64> [[TMP23]], i64 3 |
| ; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[TMP0]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[CMP_N]], label %[[FOR_END:.*]], label %[[SCALAR_PH]] |
| ; CHECK: [[SCALAR_PH]]: |
| ; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ], [ 0, %[[VECTOR_MEMCHECK]] ] |
| ; CHECK-NEXT: [[SCALAR_RECUR_INIT:%.*]] = phi i64 [ [[VECTOR_RECUR_EXTRACT]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ], [ 0, %[[VECTOR_MEMCHECK]] ] |
| ; CHECK-NEXT: br label %[[FOR_BODY:.*]] |
| ; CHECK: [[FOR_BODY]]: |
| ; CHECK-NEXT: [[PHI32:%.*]] = phi i32 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[I32NEXT:%.*]], %[[FOR_BODY]] ] |
| ; CHECK-NEXT: [[PHI64:%.*]] = phi i64 [ [[SCALAR_RECUR_INIT]], %[[SCALAR_PH]] ], [ [[I64NEXT:%.*]], %[[FOR_BODY]] ] |
| ; CHECK-NEXT: [[I32NEXT]] = add i32 [[PHI32]], 1 |
| ; CHECK-NEXT: [[I64NEXT]] = zext i32 [[I32NEXT]] to i64 |
| ; CHECK-NEXT: [[XIP:%.*]] = getelementptr inbounds double, ptr [[X]], i32 [[I32NEXT]] |
| ; CHECK-NEXT: [[YIP:%.*]] = getelementptr inbounds double, ptr [[Y]], i32 [[PHI32]] |
| ; CHECK-NEXT: [[XI:%.*]] = load double, ptr [[XIP]], align 8 |
| ; CHECK-NEXT: store double [[XI]], ptr [[YIP]], align 8 |
| ; CHECK-NEXT: [[CMP:%.*]] = icmp slt i64 [[I64NEXT]], [[N]] |
| ; CHECK-NEXT: br i1 [[CMP]], label %[[FOR_BODY]], label %[[FOR_END]], !llvm.loop [[LOOP9:![0-9]+]] |
| ; CHECK: [[FOR_END]]: |
| ; CHECK-NEXT: ret void |
| ; |
| entry: |
| br label %for.body |
| |
| for.body: |
| %phi32 = phi i32 [ 0, %entry ], [ %i32next, %for.body ] |
| %phi64 = phi i64 [ 0, %entry ], [ %i64next, %for.body ] |
| %i32next = add i32 %phi32, 1 |
| %i64next = zext i32 %i32next to i64 |
| %xip = getelementptr inbounds double, ptr %x, i32 %i32next |
| %yip = getelementptr inbounds double, ptr %y, i32 %phi32 |
| %xi = load double, ptr %xip, align 8 |
| store double %xi, ptr %yip, align 8 |
| %cmp = icmp slt i64 %i64next, %n |
| br i1 %cmp, label %for.body, label %for.end |
| |
| for.end: |
| ret void |
| } |
| |
| define void @replace_for_removes_predicate(ptr %y, ptr %x, i64 %n) { |
| ; CHECK-LABEL: define void @replace_for_removes_predicate( |
| ; CHECK-SAME: ptr [[Y:%.*]], ptr [[X:%.*]], i64 [[N:%.*]]) { |
| ; CHECK-NEXT: [[ENTRY:.*]]: |
| ; CHECK-NEXT: [[X2:%.*]] = ptrtoaddr ptr [[X]] to i64 |
| ; CHECK-NEXT: [[Y1:%.*]] = ptrtoaddr ptr [[Y]] to i64 |
| ; CHECK-NEXT: [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 [[N]], i64 1) |
| ; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[SMAX]], 4 |
| ; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_MEMCHECK:.*]] |
| ; CHECK: [[VECTOR_MEMCHECK]]: |
| ; CHECK-NEXT: [[TMP0:%.*]] = sub i64 [[Y1]], [[X2]] |
| ; CHECK-NEXT: [[DIFF_CHECK:%.*]] = icmp ult i64 [[TMP0]], 32 |
| ; CHECK-NEXT: br i1 [[DIFF_CHECK]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]] |
| ; CHECK: [[VECTOR_PH]]: |
| ; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[SMAX]], 4 |
| ; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[SMAX]], [[N_MOD_VF]] |
| ; CHECK-NEXT: [[TMP1:%.*]] = trunc i64 [[N_VEC]] to i32 |
| ; CHECK-NEXT: br label %[[VECTOR_BODY:.*]] |
| ; CHECK: [[VECTOR_BODY]]: |
| ; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[VECTOR_RECUR:%.*]] = phi <4 x i64> [ <i64 poison, i64 poison, i64 poison, i64 0>, %[[VECTOR_PH]] ], [ [[TMP3:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[TMP2:%.*]] = add <4 x i32> [[VEC_IND]], splat (i32 1) |
| ; CHECK-NEXT: [[TMP3]] = zext <4 x i32> [[TMP2]] to <4 x i64> |
| ; CHECK-NEXT: [[TMP4:%.*]] = shufflevector <4 x i64> [[VECTOR_RECUR]], <4 x i64> [[TMP3]], <4 x i32> <i32 3, i32 4, i32 5, i32 6> |
| ; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds double, ptr [[X]], i64 [[INDEX]] |
| ; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds double, ptr [[Y]], i64 [[INDEX]] |
| ; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <4 x double>, ptr [[TMP5]], align 8 |
| ; CHECK-NEXT: [[TMP7:%.*]] = sitofp <4 x i64> [[TMP4]] to <4 x double> |
| ; CHECK-NEXT: [[TMP8:%.*]] = fadd <4 x double> [[WIDE_LOAD]], [[TMP7]] |
| ; CHECK-NEXT: store <4 x double> [[TMP8]], ptr [[TMP6]], align 8 |
| ; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4 |
| ; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 4) |
| ; CHECK-NEXT: [[TMP9:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[TMP9]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP10:![0-9]+]] |
| ; CHECK: [[MIDDLE_BLOCK]]: |
| ; CHECK-NEXT: [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i64> [[TMP3]], i64 3 |
| ; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[SMAX]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[CMP_N]], label %[[FOR_END:.*]], label %[[SCALAR_PH]] |
| ; CHECK: [[SCALAR_PH]]: |
| ; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_MEMCHECK]] ] |
| ; CHECK-NEXT: [[BC_RESUME_VAL3:%.*]] = phi i32 [ [[TMP1]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_MEMCHECK]] ] |
| ; CHECK-NEXT: [[SCALAR_RECUR_INIT:%.*]] = phi i64 [ [[VECTOR_RECUR_EXTRACT]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_MEMCHECK]] ] |
| ; CHECK-NEXT: br label %[[FOR_BODY:.*]] |
| ; CHECK: [[FOR_BODY]]: |
| ; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[FOR_BODY]] ] |
| ; CHECK-NEXT: [[PHI32:%.*]] = phi i32 [ [[BC_RESUME_VAL3]], %[[SCALAR_PH]] ], [ [[I32NEXT:%.*]], %[[FOR_BODY]] ] |
| ; CHECK-NEXT: [[PHI64:%.*]] = phi i64 [ [[SCALAR_RECUR_INIT]], %[[SCALAR_PH]] ], [ [[I64NEXT:%.*]], %[[FOR_BODY]] ] |
| ; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1 |
| ; CHECK-NEXT: [[I32NEXT]] = add i32 [[PHI32]], 1 |
| ; CHECK-NEXT: [[I64NEXT]] = zext i32 [[I32NEXT]] to i64 |
| ; CHECK-NEXT: [[XIP:%.*]] = getelementptr inbounds double, ptr [[X]], i64 [[IV]] |
| ; CHECK-NEXT: [[YIP:%.*]] = getelementptr inbounds double, ptr [[Y]], i64 [[IV]] |
| ; CHECK-NEXT: [[XI:%.*]] = load double, ptr [[XIP]], align 8 |
| ; CHECK-NEXT: [[PHI64_F:%.*]] = sitofp i64 [[PHI64]] to double |
| ; CHECK-NEXT: [[RESULT:%.*]] = fadd double [[XI]], [[PHI64_F]] |
| ; CHECK-NEXT: store double [[RESULT]], ptr [[YIP]], align 8 |
| ; CHECK-NEXT: [[CMP:%.*]] = icmp slt i64 [[IV_NEXT]], [[N]] |
| ; CHECK-NEXT: br i1 [[CMP]], label %[[FOR_BODY]], label %[[FOR_END]], !llvm.loop [[LOOP11:![0-9]+]] |
| ; CHECK: [[FOR_END]]: |
| ; CHECK-NEXT: ret void |
| ; |
| entry: |
| br label %for.body |
| |
| for.body: |
| %iv = phi i64 [ 0, %entry ], [ %iv.next, %for.body ] |
| %phi32 = phi i32 [ 0, %entry ], [ %i32next, %for.body ] |
| %phi64 = phi i64 [ 0, %entry ], [ %i64next, %for.body ] |
| %iv.next = add nuw nsw i64 %iv, 1 |
| %i32next = add i32 %phi32, 1 |
| %i64next = zext i32 %i32next to i64 |
| %xip = getelementptr inbounds double, ptr %x, i64 %iv |
| %yip = getelementptr inbounds double, ptr %y, i64 %iv |
| %xi = load double, ptr %xip, align 8 |
| %phi64_f = sitofp i64 %phi64 to double |
| %result = fadd double %xi, %phi64_f |
| store double %result, ptr %yip, align 8 |
| %cmp = icmp slt i64 %iv.next, %n |
| br i1 %cmp, label %for.body, label %for.end |
| |
| for.end: |
| ret void |
| } |
| |
| define void @for_iv_kept_predicate_needed(ptr %dst, i64 %n) { |
| ; CHECK-LABEL: define void @for_iv_kept_predicate_needed( |
| ; CHECK-SAME: ptr [[DST:%.*]], i64 [[N:%.*]]) { |
| ; CHECK-NEXT: [[ENTRY:.*]]: |
| ; CHECK-NEXT: [[SMAX1:%.*]] = call i64 @llvm.smax.i64(i64 [[N]], i64 1) |
| ; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[SMAX1]], 4 |
| ; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_SCEVCHECK:.*]] |
| ; CHECK: [[VECTOR_SCEVCHECK]]: |
| ; CHECK-NEXT: [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 [[N]], i64 1) |
| ; CHECK-NEXT: [[TMP0:%.*]] = add nsw i64 [[SMAX]], -1 |
| ; CHECK-NEXT: [[TMP1:%.*]] = icmp ugt i64 [[TMP0]], 65535 |
| ; CHECK-NEXT: br i1 [[TMP1]], label %[[SCALAR_PH]], label %[[VECTOR_PH:.*]] |
| ; CHECK: [[VECTOR_PH]]: |
| ; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[SMAX1]], 4 |
| ; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[SMAX1]], [[N_MOD_VF]] |
| ; CHECK-NEXT: [[TMP2:%.*]] = trunc i64 [[N_VEC]] to i16 |
| ; CHECK-NEXT: br label %[[VECTOR_BODY:.*]] |
| ; CHECK: [[VECTOR_BODY]]: |
| ; CHECK-NEXT: [[INDEX1:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i64> [ <i64 0, i64 1, i64 2, i64 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[VEC_IND2:%.*]] = phi <4 x i16> [ <i16 0, i16 1, i16 2, i16 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT3:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[VECTOR_RECUR:%.*]] = phi <4 x i64> [ <i64 poison, i64 poison, i64 poison, i64 0>, %[[VECTOR_PH]] ], [ [[TMP6:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[TMP7:%.*]] = add <4 x i16> [[VEC_IND2]], splat (i16 1) |
| ; CHECK-NEXT: [[TMP6]] = zext <4 x i16> [[TMP7]] to <4 x i64> |
| ; CHECK-NEXT: [[TMP5:%.*]] = shufflevector <4 x i64> [[VECTOR_RECUR]], <4 x i64> [[TMP6]], <4 x i32> <i32 3, i32 4, i32 5, i32 6> |
| ; CHECK-NEXT: [[INDEX:%.*]] = extractelement <4 x i64> [[TMP5]], i64 0 |
| ; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i64, ptr [[DST]], i64 [[INDEX]] |
| ; CHECK-NEXT: store <4 x i64> [[VEC_IND]], ptr [[TMP3]], align 8 |
| ; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX1]], 4 |
| ; CHECK-NEXT: [[VEC_IND_NEXT]] = add nuw nsw <4 x i64> [[VEC_IND]], splat (i64 4) |
| ; CHECK-NEXT: [[VEC_IND_NEXT3]] = add <4 x i16> [[VEC_IND2]], splat (i16 4) |
| ; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[TMP4]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP12:![0-9]+]] |
| ; CHECK: [[MIDDLE_BLOCK]]: |
| ; CHECK-NEXT: [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i64> [[TMP6]], i64 3 |
| ; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[SMAX1]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]] |
| ; CHECK: [[SCALAR_PH]]: |
| ; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ] |
| ; CHECK-NEXT: [[BC_RESUME_VAL4:%.*]] = phi i16 [ [[TMP2]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ] |
| ; CHECK-NEXT: [[SCALAR_RECUR_INIT:%.*]] = phi i64 [ [[VECTOR_RECUR_EXTRACT]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_SCEVCHECK]] ] |
| ; CHECK-NEXT: br label %[[LOOP:.*]] |
| ; CHECK: [[LOOP]]: |
| ; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ] |
| ; CHECK-NEXT: [[NARROW:%.*]] = phi i16 [ [[BC_RESUME_VAL4]], %[[SCALAR_PH]] ], [ [[INC:%.*]], %[[LOOP]] ] |
| ; CHECK-NEXT: [[PREV:%.*]] = phi i64 [ [[SCALAR_RECUR_INIT]], %[[SCALAR_PH]] ], [ [[EXT:%.*]], %[[LOOP]] ] |
| ; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1 |
| ; CHECK-NEXT: [[INC]] = add i16 [[NARROW]], 1 |
| ; CHECK-NEXT: [[EXT]] = zext i16 [[INC]] to i64 |
| ; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i64, ptr [[DST]], i64 [[PREV]] |
| ; CHECK-NEXT: store i64 [[IV]], ptr [[GEP]], align 8 |
| ; CHECK-NEXT: [[CMP:%.*]] = icmp slt i64 [[IV_NEXT]], [[N]] |
| ; CHECK-NEXT: br i1 [[CMP]], label %[[LOOP]], label %[[EXIT]], !llvm.loop [[LOOP13:![0-9]+]] |
| ; CHECK: [[EXIT]]: |
| ; CHECK-NEXT: ret void |
| ; |
| entry: |
| br label %loop |
| |
| loop: |
| %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ] |
| %narrow = phi i16 [ 0, %entry ], [ %inc, %loop ] |
| %prev = phi i64 [ 0, %entry ], [ %ext, %loop ] |
| %iv.next = add nuw nsw i64 %iv, 1 |
| %inc = add i16 %narrow, 1 |
| %ext = zext i16 %inc to i64 |
| %gep = getelementptr inbounds i64, ptr %dst, i64 %prev |
| store i64 %iv, ptr %gep, align 8 |
| %cmp = icmp slt i64 %iv.next, %n |
| br i1 %cmp, label %loop, label %exit |
| |
| exit: |
| ret void |
| } |
| |
| define void @for_iv_dropped_predicate_removed(ptr %dst, i64 %n) { |
| ; CHECK-LABEL: define void @for_iv_dropped_predicate_removed( |
| ; CHECK-SAME: ptr [[DST:%.*]], i64 [[N:%.*]]) { |
| ; CHECK-NEXT: [[ENTRY:.*]]: |
| ; CHECK-NEXT: [[SMAX:%.*]] = call i64 @llvm.smax.i64(i64 [[N]], i64 1) |
| ; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i64 [[SMAX]], 4 |
| ; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]] |
| ; CHECK: [[VECTOR_PH]]: |
| ; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i64 [[SMAX]], 4 |
| ; CHECK-NEXT: [[N_VEC:%.*]] = sub i64 [[SMAX]], [[N_MOD_VF]] |
| ; CHECK-NEXT: [[TMP0:%.*]] = trunc i64 [[N_VEC]] to i16 |
| ; CHECK-NEXT: br label %[[VECTOR_BODY:.*]] |
| ; CHECK: [[VECTOR_BODY]]: |
| ; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i16> [ <i16 0, i16 1, i16 2, i16 3>, %[[VECTOR_PH]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[VECTOR_RECUR:%.*]] = phi <4 x i64> [ <i64 poison, i64 poison, i64 poison, i64 0>, %[[VECTOR_PH]] ], [ [[TMP2:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[TMP1:%.*]] = add <4 x i16> [[VEC_IND]], splat (i16 1) |
| ; CHECK-NEXT: [[TMP2]] = zext <4 x i16> [[TMP1]] to <4 x i64> |
| ; CHECK-NEXT: [[TMP3:%.*]] = shufflevector <4 x i64> [[VECTOR_RECUR]], <4 x i64> [[TMP2]], <4 x i32> <i32 3, i32 4, i32 5, i32 6> |
| ; CHECK-NEXT: [[TMP4:%.*]] = sitofp <4 x i64> [[TMP3]] to <4 x double> |
| ; CHECK-NEXT: [[TMP5:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[INDEX]] |
| ; CHECK-NEXT: store <4 x double> [[TMP4]], ptr [[TMP5]], align 8 |
| ; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i64 [[INDEX]], 4 |
| ; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i16> [[VEC_IND]], splat (i16 4) |
| ; CHECK-NEXT: [[TMP6:%.*]] = icmp eq i64 [[INDEX_NEXT]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[TMP6]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP14:![0-9]+]] |
| ; CHECK: [[MIDDLE_BLOCK]]: |
| ; CHECK-NEXT: [[VECTOR_RECUR_EXTRACT:%.*]] = extractelement <4 x i64> [[TMP2]], i64 3 |
| ; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i64 [[SMAX]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]] |
| ; CHECK: [[SCALAR_PH]]: |
| ; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i64 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ] |
| ; CHECK-NEXT: [[BC_RESUME_VAL1:%.*]] = phi i16 [ [[TMP0]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ] |
| ; CHECK-NEXT: [[SCALAR_RECUR_INIT:%.*]] = phi i64 [ [[VECTOR_RECUR_EXTRACT]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ] |
| ; CHECK-NEXT: br label %[[LOOP:.*]] |
| ; CHECK: [[LOOP]]: |
| ; CHECK-NEXT: [[IV:%.*]] = phi i64 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ] |
| ; CHECK-NEXT: [[NARROW:%.*]] = phi i16 [ [[BC_RESUME_VAL1]], %[[SCALAR_PH]] ], [ [[INC:%.*]], %[[LOOP]] ] |
| ; CHECK-NEXT: [[PREV:%.*]] = phi i64 [ [[SCALAR_RECUR_INIT]], %[[SCALAR_PH]] ], [ [[EXT:%.*]], %[[LOOP]] ] |
| ; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i64 [[IV]], 1 |
| ; CHECK-NEXT: [[INC]] = add i16 [[NARROW]], 1 |
| ; CHECK-NEXT: [[EXT]] = zext i16 [[INC]] to i64 |
| ; CHECK-NEXT: [[F:%.*]] = sitofp i64 [[PREV]] to double |
| ; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds double, ptr [[DST]], i64 [[IV]] |
| ; CHECK-NEXT: store double [[F]], ptr [[GEP]], align 8 |
| ; CHECK-NEXT: [[CMP:%.*]] = icmp slt i64 [[IV_NEXT]], [[N]] |
| ; CHECK-NEXT: br i1 [[CMP]], label %[[LOOP]], label %[[EXIT]], !llvm.loop [[LOOP15:![0-9]+]] |
| ; CHECK: [[EXIT]]: |
| ; CHECK-NEXT: ret void |
| ; |
| entry: |
| br label %loop |
| |
| loop: |
| %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ] |
| %narrow = phi i16 [ 0, %entry ], [ %inc, %loop ] |
| %prev = phi i64 [ 0, %entry ], [ %ext, %loop ] |
| %iv.next = add nuw nsw i64 %iv, 1 |
| %inc = add i16 %narrow, 1 |
| %ext = zext i16 %inc to i64 |
| %f = sitofp i64 %prev to double |
| %gep = getelementptr inbounds double, ptr %dst, i64 %iv |
| store double %f, ptr %gep, align 8 |
| %cmp = icmp slt i64 %iv.next, %n |
| br i1 %cmp, label %loop, label %exit |
| |
| exit: |
| ret void |
| } |
| |
| define void @induction_with_predicate_replaceable_by_for(ptr %dst, i32 %len) { |
| ; CHECK-LABEL: define void @induction_with_predicate_replaceable_by_for( |
| ; CHECK-SAME: ptr [[DST:%.*]], i32 [[LEN:%.*]]) { |
| ; CHECK-NEXT: [[ENTRY:.*]]: |
| ; CHECK-NEXT: [[UMAX1:%.*]] = call i32 @llvm.umax.i32(i32 [[LEN]], i32 1) |
| ; CHECK-NEXT: [[MIN_ITERS_CHECK:%.*]] = icmp ult i32 [[UMAX1]], 4 |
| ; CHECK-NEXT: br i1 [[MIN_ITERS_CHECK]], label %[[SCALAR_PH:.*]], label %[[VECTOR_PH:.*]] |
| ; CHECK: [[VECTOR_PH]]: |
| ; CHECK-NEXT: [[UMAX:%.*]] = call i32 @llvm.umax.i32(i32 [[LEN]], i32 1) |
| ; CHECK-NEXT: [[TMP0:%.*]] = add i32 [[UMAX]], -1 |
| ; CHECK-NEXT: [[TMP1:%.*]] = icmp ugt i32 [[TMP0]], 255 |
| ; CHECK-NEXT: br i1 [[TMP1]], label %[[SCALAR_PH]], label %[[VECTOR_PH1:.*]] |
| ; CHECK: [[VECTOR_PH1]]: |
| ; CHECK-NEXT: [[N_MOD_VF:%.*]] = urem i32 [[UMAX1]], 4 |
| ; CHECK-NEXT: [[N_VEC:%.*]] = sub i32 [[UMAX1]], [[N_MOD_VF]] |
| ; CHECK-NEXT: [[TMP2:%.*]] = trunc i32 [[N_VEC]] to i8 |
| ; CHECK-NEXT: br label %[[VECTOR_BODY:.*]] |
| ; CHECK: [[VECTOR_BODY]]: |
| ; CHECK-NEXT: [[INDEX:%.*]] = phi i32 [ 0, %[[VECTOR_PH1]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[VEC_IND:%.*]] = phi <4 x i32> [ <i32 0, i32 1, i32 2, i32 3>, %[[VECTOR_PH1]] ], [ [[VEC_IND_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[TMP3:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 [[INDEX]] |
| ; CHECK-NEXT: store <4 x i32> [[VEC_IND]], ptr [[TMP3]], align 4 |
| ; CHECK-NEXT: [[INDEX_NEXT]] = add nuw i32 [[INDEX]], 4 |
| ; CHECK-NEXT: [[VEC_IND_NEXT]] = add <4 x i32> [[VEC_IND]], splat (i32 4) |
| ; CHECK-NEXT: [[TMP4:%.*]] = icmp eq i32 [[INDEX_NEXT]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[TMP4]], label %[[MIDDLE_BLOCK:.*]], label %[[VECTOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]] |
| ; CHECK: [[MIDDLE_BLOCK]]: |
| ; CHECK-NEXT: [[CMP_N:%.*]] = icmp eq i32 [[UMAX1]], [[N_VEC]] |
| ; CHECK-NEXT: br i1 [[CMP_N]], label %[[EXIT:.*]], label %[[SCALAR_PH]] |
| ; CHECK: [[SCALAR_PH]]: |
| ; CHECK-NEXT: [[BC_RESUME_VAL:%.*]] = phi i32 [ [[N_VEC]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_PH]] ] |
| ; CHECK-NEXT: [[BC_RESUME_VAL2:%.*]] = phi i8 [ [[TMP2]], %[[MIDDLE_BLOCK]] ], [ 0, %[[ENTRY]] ], [ 0, %[[VECTOR_PH]] ] |
| ; CHECK-NEXT: br label %[[LOOP:.*]] |
| ; CHECK: [[LOOP]]: |
| ; CHECK-NEXT: [[IV:%.*]] = phi i32 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[IV_NEXT:%.*]], %[[LOOP]] ] |
| ; CHECK-NEXT: [[NARROW:%.*]] = phi i8 [ [[BC_RESUME_VAL2]], %[[SCALAR_PH]] ], [ [[INC:%.*]], %[[LOOP]] ] |
| ; CHECK-NEXT: [[SPHI:%.*]] = phi i32 [ [[BC_RESUME_VAL]], %[[SCALAR_PH]] ], [ [[EXT:%.*]], %[[LOOP]] ] |
| ; CHECK-NEXT: [[GEP:%.*]] = getelementptr inbounds i32, ptr [[DST]], i32 [[IV]] |
| ; CHECK-NEXT: store i32 [[SPHI]], ptr [[GEP]], align 4 |
| ; CHECK-NEXT: [[INC]] = add i8 [[NARROW]], 1 |
| ; CHECK-NEXT: [[EXT]] = zext i8 [[INC]] to i32 |
| ; CHECK-NEXT: [[IV_NEXT]] = add nuw nsw i32 [[IV]], 1 |
| ; CHECK-NEXT: [[C:%.*]] = icmp ult i32 [[IV_NEXT]], [[LEN]] |
| ; CHECK-NEXT: br i1 [[C]], label %[[LOOP]], label %[[EXIT]], !llvm.loop [[LOOP3:![0-9]+]] |
| ; CHECK: [[EXIT]]: |
| ; CHECK-NEXT: ret void |
| ; |
| entry: |
| br label %loop |
| |
| loop: |
| %iv = phi i32 [ 0, %entry ], [ %iv.next, %loop ] |
| %narrow = phi i8 [ 0, %entry ], [ %inc, %loop ] |
| %sphi = phi i32 [ 0, %entry ], [ %ext, %loop ] |
| %gep = getelementptr inbounds i32, ptr %dst, i32 %iv |
| store i32 %sphi, ptr %gep, align 4 |
| %inc = add i8 %narrow, 1 |
| %ext = zext i8 %inc to i32 |
| %iv.next = add nuw nsw i32 %iv, 1 |
| %c = icmp ult i32 %iv.next, %len |
| br i1 %c, label %loop, label %exit |
| |
| exit: |
| ret void |
| } |
