Fix compiler warnings. (#7974)

- Remove unused parameters. There are still many warnings that are not yet addressed. Currently, the warnings in dmlc-core dominate the error log. - Remove `distributed` parameter from metric. - Fixes some warnings about signed comparison.
2022-06-06 22:56:25 +08:00
parent d48123d23b
commit 1a33b50a0d
46 changed files with 149 additions and 189 deletions
--- a/src/common/hist_util.h
+++ b/src/common/hist_util.h
@@ -171,14 +171,14 @@ inline HistogramCuts SketchOnDMatrix(DMatrix* m, int32_t max_bins, int32_t n_thr

  if (!use_sorted) {
    HostSketchContainer container(max_bins, m->Info(), reduced, HostSketchContainer::UseGroup(info),
-                                  hessian, n_threads);
+                                  n_threads);
    for (auto const& page : m->GetBatches<SparsePage>()) {
      container.PushRowPage(page, info, hessian);
    }
    container.MakeCuts(&out);
  } else {
-    SortedSketchContainer container{
-        max_bins, m->Info(), reduced, HostSketchContainer::UseGroup(info), hessian, n_threads};
+    SortedSketchContainer container{max_bins, m->Info(), reduced,
+                                    HostSketchContainer::UseGroup(info), n_threads};
    for (auto const& page : m->GetBatches<SortedCSCPage>()) {
      container.PushColPage(page, info, hessian);
    }
--- a/src/common/partition_builder.h
+++ b/src/common/partition_builder.h
@@ -168,8 +168,8 @@ class PartitionBuilder {
    const size_t n_left  = child_nodes_sizes.first;
    const size_t n_right = child_nodes_sizes.second;

-    SetNLeftElems(node_in_set, range.begin(), range.end(), n_left);
-    SetNRightElems(node_in_set, range.begin(), range.end(), n_right);
+    SetNLeftElems(node_in_set, range.begin(), n_left);
+    SetNRightElems(node_in_set, range.begin(), n_right);
  }

  /**
@@ -188,8 +188,7 @@ class PartitionBuilder {
   */
  template <typename Pred>
  void PartitionRange(const size_t node_in_set, const size_t nid, common::Range1d range,
-                      bst_feature_t fidx, common::RowSetCollection* p_row_set_collection,
-                      Pred pred) {
+                      common::RowSetCollection* p_row_set_collection, Pred pred) {
    auto& row_set_collection = *p_row_set_collection;
    const size_t* p_ridx = row_set_collection[nid].begin;
    common::Span<const size_t> ridx(p_ridx + range.begin(), p_ridx + range.end());
@@ -200,8 +199,8 @@ class PartitionBuilder {
    const size_t n_left = child_nodes_sizes.first;
    const size_t n_right = child_nodes_sizes.second;

-    this->SetNLeftElems(node_in_set, range.begin(), range.end(), n_left);
-    this->SetNRightElems(node_in_set, range.begin(), range.end(), n_right);
+    this->SetNLeftElems(node_in_set, range.begin(), n_left);
+    this->SetNRightElems(node_in_set, range.begin(), n_right);
  }

  // allocate thread local memory, should be called for each specific task
@@ -223,12 +222,12 @@ class PartitionBuilder {
    return { mem_blocks_.at(task_idx)->Right(), end - begin };
  }

-  void SetNLeftElems(int nid, size_t begin, size_t end, size_t n_left) {
+  void SetNLeftElems(int nid, size_t begin, size_t n_left) {
    size_t task_idx = GetTaskIdx(nid, begin);
    mem_blocks_.at(task_idx)->n_left = n_left;
  }

-  void SetNRightElems(int nid, size_t begin, size_t end, size_t n_right) {
+  void SetNRightElems(int nid, size_t begin, size_t n_right) {
    size_t task_idx = GetTaskIdx(nid, begin);
    mem_blocks_.at(task_idx)->n_right = n_right;
  }
--- a/src/common/quantile.cc
+++ b/src/common/quantile.cc
@@ -543,7 +543,7 @@ template class SketchContainerImpl<WXQuantileSketch<float, float>>;

 HostSketchContainer::HostSketchContainer(int32_t max_bins, MetaInfo const &info,
                                         std::vector<size_t> columns_size, bool use_group,
-                                         Span<float const> hessian, int32_t n_threads)
+                                         int32_t n_threads)
    : SketchContainerImpl{columns_size, max_bins, info.feature_types.ConstHostSpan(), use_group,
                          n_threads} {
  monitor_.Init(__func__);
--- a/src/common/quantile.h
+++ b/src/common/quantile.h
@@ -774,7 +774,7 @@ class HostSketchContainer : public SketchContainerImpl<WQuantileSketch<float, fl

 public:
  HostSketchContainer(int32_t max_bins, MetaInfo const &info, std::vector<size_t> columns_size,
-                      bool use_group, Span<float const> hessian, int32_t n_threads);
+                      bool use_group, int32_t n_threads);
 };

 /**
@@ -868,7 +868,7 @@ class SortedSketchContainer : public SketchContainerImpl<WXQuantileSketch<float,
 public:
  explicit SortedSketchContainer(int32_t max_bins, MetaInfo const &info,
                                 std::vector<size_t> columns_size, bool use_group,
-                                 Span<float const> hessian, int32_t n_threads)
+                                 int32_t n_threads)
      : SketchContainerImpl{columns_size, max_bins, info.feature_types.ConstHostSpan(), use_group,
                            n_threads} {
    monitor_.Init(__func__);
--- a/src/data/iterative_device_dmatrix.cu
+++ b/src/data/iterative_device_dmatrix.cu
@@ -163,8 +163,7 @@ void IterativeDeviceDMatrix::Initialize(DataIterHandle iter_handle, float missin

 BatchSet<EllpackPage> IterativeDeviceDMatrix::GetEllpackBatches(const BatchParam& param) {
  CHECK(page_);
-  auto begin_iter =
-      BatchIterator<EllpackPage>(new SimpleBatchIteratorImpl<EllpackPage>(page_));
+  auto begin_iter = BatchIterator<EllpackPage>(new SimpleBatchIteratorImpl<EllpackPage>(page_));
  return BatchSet<EllpackPage>(begin_iter);
 }
 }  // namespace data
--- a/src/data/iterative_device_dmatrix.h
+++ b/src/data/iterative_device_dmatrix.h
@@ -45,8 +45,8 @@ class IterativeDeviceDMatrix : public DMatrix {

  bool EllpackExists() const override { return true; }
  bool SparsePageExists() const override { return false; }
-  DMatrix *Slice(common::Span<int32_t const> ridxs) override {
-    LOG(FATAL) << "Slicing DMatrix is not supported for Device DMatrix.";
+  DMatrix *Slice(common::Span<int32_t const>) override {
+    LOG(FATAL) << "Slicing DMatrix is not supported for Quantile DMatrix.";
    return nullptr;
  }
  BatchSet<SparsePage> GetRowBatches() override {
--- a/src/data/proxy_dmatrix.h
+++ b/src/data/proxy_dmatrix.h
@@ -84,7 +84,7 @@ class DMatrixProxy : public DMatrix {
  bool SingleColBlock() const override { return true; }
  bool EllpackExists() const override { return true; }
  bool SparsePageExists() const override { return false; }
-  DMatrix *Slice(common::Span<int32_t const> ridxs) override {
+  DMatrix* Slice(common::Span<int32_t const> /*ridxs*/) override {
    LOG(FATAL) << "Slicing DMatrix is not supported for Proxy DMatrix.";
    return nullptr;
  }
@@ -100,7 +100,7 @@ class DMatrixProxy : public DMatrix {
    LOG(FATAL) << "Not implemented.";
    return BatchSet<SortedCSCPage>(BatchIterator<SortedCSCPage>(nullptr));
  }
-  BatchSet<EllpackPage> GetEllpackBatches(const BatchParam& param) override {
+  BatchSet<EllpackPage> GetEllpackBatches(const BatchParam&) override {
    LOG(FATAL) << "Not implemented.";
    return BatchSet<EllpackPage>(BatchIterator<EllpackPage>(nullptr));
  }
--- a/src/gbm/gbtree.cc
+++ b/src/gbm/gbtree.cc
@@ -218,7 +218,7 @@ void CopyGradient(HostDeviceVector<GradientPair> const* in_gpair, int32_t n_thre
 }

 void GBTree::UpdateTreeLeaf(DMatrix const* p_fmat, HostDeviceVector<float> const& predictions,
-                            ObjFunction const* obj, size_t gidx,
+                            ObjFunction const* obj,
                            std::vector<std::unique_ptr<RegTree>>* p_trees) {
  CHECK(!updaters_.empty());
  if (!updaters_.back()->HasNodePosition()) {
@@ -257,7 +257,7 @@ void GBTree::DoBoost(DMatrix* p_fmat, HostDeviceVector<GradientPair>* in_gpair,
  if (ngroup == 1) {
    std::vector<std::unique_ptr<RegTree>> ret;
    BoostNewTrees(in_gpair, p_fmat, 0, &ret);
-    UpdateTreeLeaf(p_fmat, predt->predictions, obj, 0, &ret);
+    UpdateTreeLeaf(p_fmat, predt->predictions, obj, &ret);
    const size_t num_new_trees = ret.size();
    new_trees.push_back(std::move(ret));
    auto v_predt = out.Slice(linalg::All(), 0);
@@ -274,7 +274,7 @@ void GBTree::DoBoost(DMatrix* p_fmat, HostDeviceVector<GradientPair>* in_gpair,
      CopyGradient(in_gpair, ctx_->Threads(), ngroup, gid, &tmp);
      std::vector<std::unique_ptr<RegTree>> ret;
      BoostNewTrees(&tmp, p_fmat, gid, &ret);
-      UpdateTreeLeaf(p_fmat, predt->predictions, obj, gid, &ret);
+      UpdateTreeLeaf(p_fmat, predt->predictions, obj, &ret);
      const size_t num_new_trees = ret.size();
      new_trees.push_back(std::move(ret));
      auto v_predt = out.Slice(linalg::All(), gid);
@@ -289,7 +289,7 @@ void GBTree::DoBoost(DMatrix* p_fmat, HostDeviceVector<GradientPair>* in_gpair,
  }

  monitor_.Stop("BoostNewTrees");
-  this->CommitModel(std::move(new_trees), p_fmat, predt);
+  this->CommitModel(std::move(new_trees));
 }

 void GBTree::InitUpdater(Args const& cfg) {
@@ -378,9 +378,7 @@ void GBTree::BoostNewTrees(HostDeviceVector<GradientPair>* gpair, DMatrix* p_fma
  }
 }

-void GBTree::CommitModel(std::vector<std::vector<std::unique_ptr<RegTree>>>&& new_trees,
-                         DMatrix* m,
-                         PredictionCacheEntry* predts) {
+void GBTree::CommitModel(std::vector<std::vector<std::unique_ptr<RegTree>>>&& new_trees) {
  monitor_.Start("CommitModel");
  for (uint32_t gid = 0; gid < model_.learner_model_param->num_output_group; ++gid) {
    model_.CommitModel(std::move(new_trees[gid]), gid);
@@ -490,15 +488,14 @@ void GBTree::Slice(int32_t layer_begin, int32_t layer_end, int32_t step,
           "want to update a portion of trees.";
  }

-  *out_of_bound = detail::SliceTrees(
-      layer_begin, layer_end, step, this->model_, tparam_, layer_trees,
-      [&](auto const &in_it, auto const &out_it) {
-        auto new_tree =
-            std::make_unique<RegTree>(*this->model_.trees.at(in_it));
-        bst_group_t group = this->model_.tree_info[in_it];
-        out_trees.at(out_it) = std::move(new_tree);
-        out_trees_info.at(out_it) = group;
-      });
+  *out_of_bound = detail::SliceTrees(layer_begin, layer_end, step, this->model_, layer_trees,
+                                     [&](auto const& in_it, auto const& out_it) {
+                                       auto new_tree =
+                                           std::make_unique<RegTree>(*this->model_.trees.at(in_it));
+                                       bst_group_t group = this->model_.tree_info[in_it];
+                                       out_trees.at(out_it) = std::move(new_tree);
+                                       out_trees_info.at(out_it) = group;
+                                     });
 }

 void GBTree::PredictBatch(DMatrix* p_fmat,
@@ -674,11 +671,10 @@ class Dart : public GBTree {
    auto p_dart = dynamic_cast<Dart*>(out);
    CHECK(p_dart);
    CHECK(p_dart->weight_drop_.empty());
-    detail::SliceTrees(
-        layer_begin, layer_end, step, model_, tparam_, this->LayerTrees(),
-        [&](auto const& in_it, auto const&) {
-          p_dart->weight_drop_.push_back(this->weight_drop_.at(in_it));
-        });
+    detail::SliceTrees(layer_begin, layer_end, step, model_, this->LayerTrees(),
+                       [&](auto const& in_it, auto const&) {
+                         p_dart->weight_drop_.push_back(this->weight_drop_.at(in_it));
+                       });
  }

  void SaveModel(Json *p_out) const override {
@@ -901,9 +897,7 @@ class Dart : public GBTree {

 protected:
  // commit new trees all at once
-  void
-  CommitModel(std::vector<std::vector<std::unique_ptr<RegTree>>>&& new_trees,
-              DMatrix*, PredictionCacheEntry*) override {
+  void CommitModel(std::vector<std::vector<std::unique_ptr<RegTree>>>&& new_trees) override {
    int num_new_trees = 0;
    for (uint32_t gid = 0; gid < model_.learner_model_param->num_output_group; ++gid) {
      num_new_trees += new_trees[gid].size();
--- a/src/gbm/gbtree.h
+++ b/src/gbm/gbtree.h
@@ -162,9 +162,8 @@ inline std::pair<uint32_t, uint32_t> LayerToTree(gbm::GBTreeModel const &model,

 // Call fn for each pair of input output tree.  Return true if index is out of bound.
 template <typename Func>
-inline bool SliceTrees(int32_t layer_begin, int32_t layer_end, int32_t step,
-                       GBTreeModel const &model, GBTreeTrainParam const &tparam,
-                       uint32_t layer_trees, Func fn) {
+bool SliceTrees(int32_t layer_begin, int32_t layer_end, int32_t step, GBTreeModel const& model,
+                uint32_t layer_trees, Func fn) {
  uint32_t tree_begin, tree_end;
  std::tie(tree_begin, tree_end) = detail::LayerToTree(model, layer_begin, layer_end);
  if (tree_end > model.trees.size()) {
@@ -206,8 +205,7 @@ class GBTree : public GradientBooster {
   * \brief Optionally update the leaf value.
   */
  void UpdateTreeLeaf(DMatrix const* p_fmat, HostDeviceVector<float> const& predictions,
-                      ObjFunction const* obj, size_t gidx,
-                      std::vector<std::unique_ptr<RegTree>>* p_trees);
+                      ObjFunction const* obj, std::vector<std::unique_ptr<RegTree>>* p_trees);

  /*! \brief Carry out one iteration of boosting */
  void DoBoost(DMatrix* p_fmat, HostDeviceVector<GradientPair>* in_gpair,
@@ -325,7 +323,7 @@ class GBTree : public GradientBooster {
    };

    if (importance_type == "weight") {
-      add_score([&](auto const &p_tree, bst_node_t, bst_feature_t split) {
+      add_score([&](auto const&, bst_node_t, bst_feature_t split) {
        gain_map[split] = split_counts[split];
      });
    } else if (importance_type == "gain" || importance_type == "total_gain") {
@@ -423,9 +421,7 @@ class GBTree : public GradientBooster {
                                                 DMatrix* f_dmat = nullptr) const;

  // commit new trees all at once
-  virtual void CommitModel(std::vector<std::vector<std::unique_ptr<RegTree>>>&& new_trees,
-                           DMatrix* m,
-                           PredictionCacheEntry* predts);
+  virtual void CommitModel(std::vector<std::vector<std::unique_ptr<RegTree>>>&& new_trees);

  // --- data structure ---
  GBTreeModel model_;
--- a/src/learner.cc
+++ b/src/learner.cc
@@ -1234,8 +1234,7 @@ class LearnerImpl : public LearnerIO {

      obj_->EvalTransform(&out);
      for (auto& ev : metrics_) {
-        os << '\t' << data_names[i] << '-' << ev->Name() << ':'
-           << ev->Eval(out, m->Info(), tparam_.dsplit == DataSplitMode::kRow);
+        os << '\t' << data_names[i] << '-' << ev->Name() << ':' << ev->Eval(out, m->Info());
      }
    }

--- a/src/metric/auc.cc
+++ b/src/metric/auc.cc
@@ -254,8 +254,7 @@ std::pair<double, uint32_t> RankingAUC(std::vector<float> const &predts,

 template <typename Curve>
 class EvalAUC : public Metric {
-  double Eval(const HostDeviceVector<bst_float> &preds, const MetaInfo &info,
-              bool distributed) override {
+  double Eval(const HostDeviceVector<bst_float> &preds, const MetaInfo &info) override {
    double auc {0};
    if (tparam_->gpu_id != GenericParameter::kCpuId) {
      preds.SetDevice(tparam_->gpu_id);
--- a/src/metric/auc.cu
+++ b/src/metric/auc.cu
@@ -312,10 +312,8 @@ void SegmentedReduceAUC(common::Span<size_t const> d_unique_idx,
 * up each class in all kernels.
 */
 template <bool scale, typename Fn>
-double GPUMultiClassAUCOVR(common::Span<float const> predts,
-                           MetaInfo const &info, int32_t device,
-                           common::Span<uint32_t> d_class_ptr, size_t n_classes,
-                           std::shared_ptr<DeviceAUCCache> cache, Fn area_fn) {
+double GPUMultiClassAUCOVR(MetaInfo const &info, int32_t device, common::Span<uint32_t> d_class_ptr,
+                           size_t n_classes, std::shared_ptr<DeviceAUCCache> cache, Fn area_fn) {
  dh::safe_cuda(cudaSetDevice(device));
  /**
   * Sorted idx
@@ -478,8 +476,7 @@ double GPUMultiClassROCAUC(common::Span<float const> predts,
                              double tp, size_t /*class_id*/) {
    return TrapezoidArea(fp_prev, fp, tp_prev, tp);
  };
-  return GPUMultiClassAUCOVR<true>(predts, info, device, dh::ToSpan(class_ptr),
-                                   n_classes, cache, fn);
+  return GPUMultiClassAUCOVR<true>(info, device, dh::ToSpan(class_ptr), n_classes, cache, fn);
 }

 namespace {
@@ -704,8 +701,7 @@ double GPUMultiClassPRAUC(common::Span<float const> predts,
    return detail::CalcDeltaPRAUC(fp_prev, fp, tp_prev, tp,
                                  d_totals[class_id].first);
  };
-  return GPUMultiClassAUCOVR<false>(predts, info, device, d_class_ptr,
-                                    n_classes, cache, fn);
+  return GPUMultiClassAUCOVR<false>(info, device, d_class_ptr, n_classes, cache, fn);
 }

 template <typename Fn>
--- a/src/metric/elementwise_metric.cu
+++ b/src/metric/elementwise_metric.cu
@@ -178,8 +178,7 @@ class PseudoErrorLoss : public Metric {
    out["pseudo_huber_param"] = ToJson(param_);
  }

-  double Eval(const HostDeviceVector<bst_float>& preds, const MetaInfo& info,
-              bool distributed) override {
+  double Eval(const HostDeviceVector<bst_float>& preds, const MetaInfo& info) override {
    CHECK_EQ(info.labels.Shape(0), info.num_row_);
    auto labels = info.labels.View(tparam_->gpu_id);
    preds.SetDevice(tparam_->gpu_id);
@@ -197,7 +196,7 @@ class PseudoErrorLoss : public Metric {
          return std::make_tuple(v, wt);
        });
    double dat[2]{result.Residue(), result.Weights()};
-    if (distributed) {
+    if (rabit::IsDistributed()) {
      rabit::Allreduce<rabit::op::Sum>(dat, 2);
    }
    return EvalRowMAPE::GetFinal(dat[0], dat[1]);
@@ -342,8 +341,7 @@ struct EvalEWiseBase : public Metric {
  EvalEWiseBase() = default;
  explicit EvalEWiseBase(char const* policy_param) : policy_{policy_param} {}

-  double Eval(HostDeviceVector<bst_float> const& preds, const MetaInfo& info,
-              bool distributed) override {
+  double Eval(HostDeviceVector<bst_float> const& preds, const MetaInfo& info) override {
    CHECK_EQ(preds.Size(), info.labels.Size())
        << "label and prediction size not match, "
        << "hint: use merror or mlogloss for multi-class classification";
@@ -367,10 +365,7 @@ struct EvalEWiseBase : public Metric {
        });

    double dat[2]{result.Residue(), result.Weights()};
-
-    if (distributed) {
-      rabit::Allreduce<rabit::op::Sum>(dat, 2);
-    }
+    rabit::Allreduce<rabit::op::Sum>(dat, 2);
    return Policy::GetFinal(dat[0], dat[1]);
  }

--- a/src/metric/multiclass_metric.cu
+++ b/src/metric/multiclass_metric.cu
@@ -167,8 +167,7 @@ class MultiClassMetricsReduction {
 */
 template<typename Derived>
 struct EvalMClassBase : public Metric {
-  double Eval(const HostDeviceVector<float> &preds, const MetaInfo &info,
-              bool distributed) override {
+  double Eval(const HostDeviceVector<float> &preds, const MetaInfo &info) override {
    if (info.labels.Size() == 0) {
      CHECK_EQ(preds.Size(), 0);
    } else {
@@ -186,9 +185,7 @@ struct EvalMClassBase : public Metric {
      dat[0] = result.Residue();
      dat[1] = result.Weights();
    }
-    if (distributed) {
-      rabit::Allreduce<rabit::op::Sum>(dat, 2);
-    }
+    rabit::Allreduce<rabit::op::Sum>(dat, 2);
    return Derived::GetFinal(dat[0], dat[1]);
  }
  /*!
--- a/src/metric/rank_metric.cc
+++ b/src/metric/rank_metric.cc
@@ -102,9 +102,8 @@ struct EvalAMS : public Metric {
    name_ = os.str();
  }

-  double Eval(const HostDeviceVector<bst_float> &preds, const MetaInfo &info,
-              bool distributed) override {
-    CHECK(!distributed) << "metric AMS do not support distributed evaluation";
+  double Eval(const HostDeviceVector<bst_float>& preds, const MetaInfo& info) override {
+    CHECK(!rabit::IsDistributed()) << "metric AMS do not support distributed evaluation";
    using namespace std;  // NOLINT(*)

    const auto ndata = static_cast<bst_omp_uint>(info.labels.Size());
@@ -161,8 +160,7 @@ struct EvalRank : public Metric, public EvalRankConfig {
  std::unique_ptr<xgboost::Metric> rank_gpu_;

 public:
-  double Eval(const HostDeviceVector<bst_float> &preds, const MetaInfo &info,
-              bool distributed) override {
+  double Eval(const HostDeviceVector<bst_float>& preds, const MetaInfo& info) override {
    CHECK_EQ(preds.Size(), info.labels.Size())
        << "label size predict size not match";

@@ -185,7 +183,7 @@ struct EvalRank : public Metric, public EvalRankConfig {
        rank_gpu_.reset(GPUMetric::CreateGPUMetric(this->Name(), tparam_));
      }
      if (rank_gpu_) {
-        sum_metric = rank_gpu_->Eval(preds, info, distributed);
+        sum_metric = rank_gpu_->Eval(preds, info);
      }
    }

@@ -218,7 +216,7 @@ struct EvalRank : public Metric, public EvalRankConfig {
      exc.Rethrow();
    }

-    if (distributed) {
+    if (rabit::IsDistributed()) {
      double dat[2]{sum_metric, static_cast<double>(ngroups)};
      // approximately estimate the metric using mean
      rabit::Allreduce<rabit::op::Sum>(dat, 2);
@@ -342,9 +340,8 @@ struct EvalMAP : public EvalRank {
 struct EvalCox : public Metric {
 public:
  EvalCox() = default;
-  double Eval(const HostDeviceVector<bst_float> &preds, const MetaInfo &info,
-              bool distributed) override {
-    CHECK(!distributed) << "Cox metric does not support distributed evaluation";
+  double Eval(const HostDeviceVector<bst_float>& preds, const MetaInfo& info) override {
+    CHECK(!rabit::IsDistributed()) << "Cox metric does not support distributed evaluation";
    using namespace std;  // NOLINT(*)

    const auto ndata = static_cast<bst_omp_uint>(info.labels.Size());
--- a/src/metric/rank_metric.cu
+++ b/src/metric/rank_metric.cu
@@ -29,8 +29,7 @@ DMLC_REGISTRY_FILE_TAG(rank_metric_gpu);
 template <typename EvalMetricT>
 struct EvalRankGpu : public GPUMetric, public EvalRankConfig {
 public:
-  double Eval(const HostDeviceVector<bst_float> &preds, const MetaInfo &info,
-              bool distributed) override {
+  double Eval(const HostDeviceVector<bst_float> &preds, const MetaInfo &info) override {
    // Sanity check is done by the caller
    std::vector<unsigned> tgptr(2, 0);
    tgptr[1] = static_cast<unsigned>(preds.Size());
--- a/src/metric/survival_metric.cu
+++ b/src/metric/survival_metric.cu
@@ -206,20 +206,15 @@ template <typename Policy> struct EvalEWiseSurvivalBase : public Metric {
    CHECK(tparam_);
  }

-  double Eval(const HostDeviceVector<float> &preds, const MetaInfo &info,
-              bool distributed) override {
+  double Eval(const HostDeviceVector<float>& preds, const MetaInfo& info) override {
    CHECK_EQ(preds.Size(), info.labels_lower_bound_.Size());
    CHECK_EQ(preds.Size(), info.labels_upper_bound_.Size());
    CHECK(tparam_);
-    auto result =
-        reducer_.Reduce(*tparam_, info.weights_, info.labels_lower_bound_,
-                        info.labels_upper_bound_, preds);
+    auto result = reducer_.Reduce(*tparam_, info.weights_, info.labels_lower_bound_,
+                                  info.labels_upper_bound_, preds);

-    double dat[2] {result.Residue(), result.Weights()};
-
-    if (distributed) {
-      rabit::Allreduce<rabit::op::Sum>(dat, 2);
-    }
+    double dat[2]{result.Residue(), result.Weights()};
+    rabit::Allreduce<rabit::op::Sum>(dat, 2);
    return Policy::GetFinal(dat[0], dat[1]);
  }

@@ -240,10 +235,9 @@ struct AFTNLogLikDispatcher : public Metric {
    return "aft-nloglik";
  }

-  double Eval(const HostDeviceVector<bst_float> &preds, const MetaInfo &info,
-              bool distributed) override {
+  double Eval(const HostDeviceVector<bst_float>& preds, const MetaInfo& info) override {
    CHECK(metric_) << "AFT metric must be configured first, with distribution type and scale";
-    return metric_->Eval(preds, info, distributed);
+    return metric_->Eval(preds, info);
  }

  void Configure(const Args& args) override {
--- a/src/tree/gpu_hist/row_partitioner.cuh
+++ b/src/tree/gpu_hist/row_partitioner.cuh
@@ -116,7 +116,7 @@ class RowPartitioner {
    Segment segment = ridx_segments_.at(nidx);  // rows belongs to node nidx
    auto d_ridx = ridx_.CurrentSpan();
    auto d_position = position_.CurrentSpan();
-    if (left_counts_.size() <= nidx) {
+    if (left_counts_.size() <= static_cast<size_t>(nidx)) {
      left_counts_.resize((nidx * 2) + 1);
      thrust::fill(left_counts_.begin(), left_counts_.end(), 0);
    }
--- a/src/tree/hist/evaluate_splits.h
+++ b/src/tree/hist/evaluate_splits.h
@@ -203,8 +203,8 @@ class HistEvaluator {
  // Returns the sum of gradients corresponding to the data points that contains
  // a non-missing value for the particular feature fid.
  template <int d_step>
-  GradStats EnumerateSplit(common::HistogramCuts const &cut, common::Span<size_t const> sorted_idx,
-                           const common::GHistRow &hist, bst_feature_t fidx, bst_node_t nidx,
+  GradStats EnumerateSplit(common::HistogramCuts const &cut, const common::GHistRow &hist,
+                           bst_feature_t fidx, bst_node_t nidx,
                           TreeEvaluator::SplitEvaluator<TrainParam> const &evaluator,
                           SplitEntry *p_best) const {
    static_assert(d_step == +1 || d_step == -1, "Invalid step.");
@@ -333,9 +333,9 @@ class HistEvaluator {
            EnumeratePart<-1>(cut, sorted_idx, histogram, fidx, nidx, evaluator, best);
          }
        } else {
-          auto grad_stats = EnumerateSplit<+1>(cut, {}, histogram, fidx, nidx, evaluator, best);
+          auto grad_stats = EnumerateSplit<+1>(cut, histogram, fidx, nidx, evaluator, best);
          if (SplitContainsMissingValues(grad_stats, snode_[nidx])) {
-            EnumerateSplit<-1>(cut, {}, histogram, fidx, nidx, evaluator, best);
+            EnumerateSplit<-1>(cut, histogram, fidx, nidx, evaluator, best);
          }
        }
      }
@@ -440,7 +440,7 @@ template <typename Partitioner, typename ExpandEntry>
 void UpdatePredictionCacheImpl(GenericParameter const *ctx, RegTree const *p_last_tree,
                               std::vector<Partitioner> const &partitioner,
                               HistEvaluator<ExpandEntry> const &hist_evaluator,
-                               TrainParam const &param, linalg::VectorView<float> out_preds) {
+                               linalg::VectorView<float> out_preds) {
  CHECK_GT(out_preds.Size(), 0U);

  CHECK(p_last_tree);
--- a/src/tree/updater_approx.cc
+++ b/src/tree/updater_approx.cc
@@ -116,7 +116,7 @@ class GloablApproxBuilder {
    // Caching prediction seems redundant for approx tree method, as sketching takes up
    // majority of training time.
    CHECK_EQ(out_preds.Size(), data->Info().num_row_);
-    UpdatePredictionCacheImpl(ctx_, p_last_tree_, partitioner_, evaluator_, param_, out_preds);
+    UpdatePredictionCacheImpl(ctx_, p_last_tree_, partitioner_, evaluator_, out_preds);
    monitor_->Stop(__func__);
  }

--- a/src/tree/updater_approx.h
+++ b/src/tree/updater_approx.h
@@ -83,7 +83,7 @@ class ApproxRowPartitioner {
      const size_t task_id = partition_builder_.GetTaskIdx(node_in_set, r.begin());
      partition_builder_.AllocateForTask(task_id);
      partition_builder_.PartitionRange(
-          node_in_set, nid, r, fidx, &row_set_collection_, [&](size_t row_id) {
+          node_in_set, nid, r, &row_set_collection_, [&](size_t row_id) {
            auto cut_value = SearchCutValue(row_id, fidx, index, cut_ptrs, cut_values);
            if (std::isnan(cut_value)) {
              return candidate.split.DefaultLeft();
--- a/src/tree/updater_gpu_hist.cu
+++ b/src/tree/updater_gpu_hist.cu
@@ -563,7 +563,7 @@ struct GPUHistMakerDevice {
    // when processing a large batch
    this->AllReduceHist(hist_nidx.at(0), reducer, hist_nidx.size());

-    for (int i = 0; i < subtraction_nidx.size(); i++) {
+    for (size_t i = 0; i < subtraction_nidx.size(); i++) {
      auto build_hist_nidx = hist_nidx.at(i);
      auto subtraction_trick_nidx = subtraction_nidx.at(i);
      auto parent_nidx = candidates.at(i).nid;
--- a/src/tree/updater_quantile_hist.cc
+++ b/src/tree/updater_quantile_hist.cc
@@ -257,7 +257,7 @@ bool QuantileHistMaker::Builder::UpdatePredictionCache(DMatrix const *data,
  }
  monitor_->Start(__func__);
  CHECK_EQ(out_preds.Size(), data->Info().num_row_);
-  UpdatePredictionCacheImpl(ctx_, p_last_tree_, partitioner_, *evaluator_, param_, out_preds);
+  UpdatePredictionCacheImpl(ctx_, p_last_tree_, partitioner_, *evaluator_, out_preds);
  monitor_->Stop(__func__);
  return true;
 }