| ; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --check-globals none --version 6 |
| ; RUN: opt < %s -passes=loop-vectorize -enable-histogram-loop-vectorization -force-vector-width=2 -S | FileCheck %s |
| |
| define void @simple_histogram(ptr noalias %buckets, ptr readonly %indices) { |
| ; CHECK-LABEL: define void @simple_histogram( |
| ; CHECK-SAME: ptr noalias [[BUCKETS:%.*]], ptr readonly [[INDICES:%.*]]) { |
| ; CHECK-NEXT: [[ENTRY1:.*:]] |
| ; CHECK-NEXT: br label %[[ENTRY:.*]] |
| ; CHECK: [[ENTRY]]: |
| ; CHECK-NEXT: br label %[[FOR_BODY:.*]] |
| ; CHECK: [[FOR_BODY]]: |
| ; CHECK-NEXT: [[IV:%.*]] = phi i64 [ 0, %[[ENTRY]] ], [ [[IV_NEXT:%.*]], %[[FOR_BODY]] ] |
| ; CHECK-NEXT: [[TMP0:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[IV]] |
| ; CHECK-NEXT: [[WIDE_LOAD:%.*]] = load <2 x i32>, ptr [[TMP0]], align 4 |
| ; CHECK-NEXT: [[TMP1:%.*]] = zext <2 x i32> [[WIDE_LOAD]] to <2 x i64> |
| ; CHECK-NEXT: [[TMP7:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], <2 x i64> [[TMP1]] |
| ; CHECK-NEXT: call void @llvm.experimental.vector.histogram.add.v2p0.i32(<2 x ptr> [[TMP7]], i32 1, <2 x i1> splat (i1 true)) |
| ; CHECK-NEXT: [[IV_NEXT]] = add nuw i64 [[IV]], 2 |
| ; CHECK-NEXT: [[TMP8:%.*]] = icmp eq i64 [[IV_NEXT]], 1000 |
| ; CHECK-NEXT: br i1 [[TMP8]], label %[[SCALAR_PH:.*]], label %[[FOR_BODY]], !llvm.loop [[LOOP0:![0-9]+]] |
| ; CHECK: [[SCALAR_PH]]: |
| ; CHECK-NEXT: br label %[[FOR_BODY1:.*]] |
| ; CHECK: [[FOR_BODY1]]: |
| ; CHECK-NEXT: ret void |
| ; |
| entry: |
| br label %loop |
| |
| loop: |
| %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ] |
| %gep.indices = getelementptr inbounds i32, ptr %indices, i64 %iv |
| %l.idx = load i32, ptr %gep.indices, align 4 |
| %idxprom1 = zext i32 %l.idx to i64 |
| %gep.bucket = getelementptr inbounds i32, ptr %buckets, i64 %idxprom1 |
| %l.bucket = load i32, ptr %gep.bucket, align 4 |
| %inc = add nsw i32 %l.bucket, 1 |
| store i32 %inc, ptr %gep.bucket, align 4 |
| %iv.next = add nuw nsw i64 %iv, 1 |
| %exitcond = icmp eq i64 %iv.next, 1000 |
| br i1 %exitcond, label %exit, label %loop |
| |
| exit: |
| ret void |
| } |
| |
| define void @histogram_extra_use_of_bucket_value(ptr noalias %buckets, ptr readonly %indices, ptr noalias %out) { |
| ; CHECK-LABEL: define void @histogram_extra_use_of_bucket_value( |
| ; CHECK-SAME: ptr noalias [[BUCKETS:%.*]], ptr readonly [[INDICES:%.*]], ptr noalias [[OUT:%.*]]) { |
| ; CHECK-NEXT: [[VECTOR_PH:.*]]: |
| ; CHECK-NEXT: br label %[[VECTOR_BODY:.*]] |
| ; CHECK: [[VECTOR_BODY]]: |
| ; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[INDEX]] |
| ; CHECK-NEXT: [[L_IDX:%.*]] = load i32, ptr [[TMP1]], align 4 |
| ; CHECK-NEXT: [[TMP5:%.*]] = zext i32 [[L_IDX]] to i64 |
| ; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], i64 [[TMP5]] |
| ; CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP6]], align 4 |
| ; CHECK-NEXT: [[INC:%.*]] = add nsw i32 [[TMP10]], 1 |
| ; CHECK-NEXT: store i32 [[INC]], ptr [[TMP6]], align 4 |
| ; CHECK-NEXT: [[TMP13:%.*]] = getelementptr inbounds i32, ptr [[OUT]], i64 [[INDEX]] |
| ; CHECK-NEXT: store i32 [[TMP10]], ptr [[TMP13]], align 4 |
| ; CHECK-NEXT: [[INDEX_NEXT]] = add nuw nsw i64 [[INDEX]], 1 |
| ; CHECK-NEXT: [[TMP14:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000 |
| ; CHECK-NEXT: br i1 [[TMP14]], label %[[LOOP:.*]], label %[[VECTOR_BODY]] |
| ; CHECK: [[LOOP]]: |
| ; CHECK-NEXT: ret void |
| ; |
| entry: |
| br label %loop |
| |
| loop: |
| %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ] |
| %gep.indices = getelementptr inbounds i32, ptr %indices, i64 %iv |
| %l.idx = load i32, ptr %gep.indices, align 4 |
| %idxprom1 = zext i32 %l.idx to i64 |
| %gep.bucket = getelementptr inbounds i32, ptr %buckets, i64 %idxprom1 |
| %l.bucket = load i32, ptr %gep.bucket, align 4 |
| %inc = add nsw i32 %l.bucket, 1 |
| store i32 %inc, ptr %gep.bucket, align 4 |
| %gep.out = getelementptr inbounds i32, ptr %out, i64 %iv |
| store i32 %l.bucket, ptr %gep.out, align 4 |
| %iv.next = add nuw nsw i64 %iv, 1 |
| %exitcond = icmp eq i64 %iv.next, 1000 |
| br i1 %exitcond, label %exit, label %loop |
| |
| exit: |
| ret void |
| } |
| |
| define void @histogram_extra_use_of_update(ptr noalias %buckets, ptr readonly %indices, ptr noalias %out) { |
| ; CHECK-LABEL: define void @histogram_extra_use_of_update( |
| ; CHECK-SAME: ptr noalias [[BUCKETS:%.*]], ptr readonly [[INDICES:%.*]], ptr noalias [[OUT:%.*]]) { |
| ; CHECK-NEXT: [[VECTOR_PH:.*]]: |
| ; CHECK-NEXT: br label %[[VECTOR_BODY:.*]] |
| ; CHECK: [[VECTOR_BODY]]: |
| ; CHECK-NEXT: [[INDEX:%.*]] = phi i64 [ 0, %[[VECTOR_PH]] ], [ [[INDEX_NEXT:%.*]], %[[VECTOR_BODY]] ] |
| ; CHECK-NEXT: [[TMP1:%.*]] = getelementptr inbounds i32, ptr [[INDICES]], i64 [[INDEX]] |
| ; CHECK-NEXT: [[L_IDX:%.*]] = load i32, ptr [[TMP1]], align 4 |
| ; CHECK-NEXT: [[TMP5:%.*]] = zext i32 [[L_IDX]] to i64 |
| ; CHECK-NEXT: [[TMP6:%.*]] = getelementptr inbounds i32, ptr [[BUCKETS]], i64 [[TMP5]] |
| ; CHECK-NEXT: [[TMP10:%.*]] = load i32, ptr [[TMP6]], align 4 |
| ; CHECK-NEXT: [[INC:%.*]] = add nsw i32 [[TMP10]], 1 |
| ; CHECK-NEXT: store i32 [[INC]], ptr [[TMP6]], align 4 |
| ; CHECK-NEXT: [[TMP14:%.*]] = getelementptr inbounds i32, ptr [[OUT]], i64 [[INDEX]] |
| ; CHECK-NEXT: store i32 [[INC]], ptr [[TMP14]], align 4 |
| ; CHECK-NEXT: [[INDEX_NEXT]] = add nuw nsw i64 [[INDEX]], 1 |
| ; CHECK-NEXT: [[TMP15:%.*]] = icmp eq i64 [[INDEX_NEXT]], 1000 |
| ; CHECK-NEXT: br i1 [[TMP15]], label %[[LOOP:.*]], label %[[VECTOR_BODY]] |
| ; CHECK: [[LOOP]]: |
| ; CHECK-NEXT: ret void |
| ; |
| entry: |
| br label %loop |
| |
| loop: |
| %iv = phi i64 [ 0, %entry ], [ %iv.next, %loop ] |
| %gep.indices = getelementptr inbounds i32, ptr %indices, i64 %iv |
| %l.idx = load i32, ptr %gep.indices, align 4 |
| %idxprom1 = zext i32 %l.idx to i64 |
| %gep.bucket = getelementptr inbounds i32, ptr %buckets, i64 %idxprom1 |
| %l.bucket = load i32, ptr %gep.bucket, align 4 |
| %inc = add nsw i32 %l.bucket, 1 |
| store i32 %inc, ptr %gep.bucket, align 4 |
| %gep.out = getelementptr inbounds i32, ptr %out, i64 %iv |
| store i32 %inc, ptr %gep.out, align 4 |
| %iv.next = add nuw nsw i64 %iv, 1 |
| %exitcond = icmp eq i64 %iv.next, 1000 |
| br i1 %exitcond, label %exit, label %loop |
| |
| exit: |
| ret void |
| } |