Skip to content

Commit

Permalink
Use adapter to initialize column matrix. (#7912)
Browse files Browse the repository at this point in the history
  • Loading branch information
@trivialfis
trivialfis committed May 18, 2022
1 parent 5ef33ad commit 19775ff
Show file tree
Hide file tree
Showing 4 changed files with 82 additions and 73 deletions.
58 changes: 33 additions & 25 deletions src/common/column_matrix.h
View file
Expand Up @@ -16,6 +16,7 @@
#include <utility> // std::move
#include <vector>

#include "../data/adapter.h"
#include "../data/gradient_index.h"
#include "hist_util.h"

Expand Down Expand Up @@ -128,17 +129,17 @@ class DenseColumnIter : public Column<BinIdxT> {
/**
* \brief Column major matrix for gradient index. This matrix contains both dense column
* and sparse column, the type of the column is controlled by sparse threshold. When the
* number of missing values in a column is below the threshold it classified as dense
* number of missing values in a column is below the threshold it's classified as dense
* column.
*/
class ColumnMatrix {
public:
// get number of features
bst_feature_t GetNumFeature() const { return static_cast<bst_feature_t>(type_.size()); }

// construct column matrix from GHistIndexMatrix
void Init(SparsePage const& page, const GHistIndexMatrix& gmat, double sparse_threshold,
int32_t n_threads) {
template <typename Batch>
void Init(Batch const& batch, float missing, GHistIndexMatrix const& gmat,
double sparse_threshold, int32_t n_threads) {
auto const nfeature = static_cast<bst_feature_t>(gmat.cut.Ptrs().size() - 1);
const size_t nrow = gmat.row_ptr.size() - 1;
// identify type of each column
Expand Down Expand Up @@ -190,21 +191,29 @@ class ColumnMatrix {
any_missing_ = !gmat.IsDense();

missing_flags_.clear();

// pre-fill index_ for dense columns
BinTypeSize gmat_bin_size = gmat.index.GetBinTypeSize();
if (!any_missing_) {
missing_flags_.resize(feature_offsets_[nfeature], false);
// row index is compressed, we need to dispatch it.
DispatchBinType(gmat_bin_size, [&, nrow, nfeature, n_threads](auto t) {
using RowBinIdxT = decltype(t);
SetIndexNoMissing(page, gmat.index.data<RowBinIdxT>(), nrow, nfeature, n_threads);
SetIndexNoMissing(gmat.index.data<RowBinIdxT>(), nrow, nfeature, n_threads);
});
} else {
missing_flags_.resize(feature_offsets_[nfeature], true);
SetIndexMixedColumns(page, gmat.index.data<uint32_t>(), gmat, nfeature);
SetIndexMixedColumns(batch, gmat.index.data<uint32_t>(), gmat, nfeature, missing);
}
}

// construct column matrix from GHistIndexMatrix
void Init(SparsePage const& page, const GHistIndexMatrix& gmat, double sparse_threshold,
int32_t n_threads) {
auto batch = data::SparsePageAdapterBatch{page.GetView()};
this->Init(batch, std::numeric_limits<float>::quiet_NaN(), gmat, sparse_threshold, n_threads);
}

/* Set the number of bytes based on numeric limit of maximum number of bins provided by user */
void SetTypeSize(size_t max_bin_per_feat) {
if ((max_bin_per_feat - 1) <= static_cast<int>(std::numeric_limits<uint8_t>::max())) {
Expand Down Expand Up @@ -241,8 +250,8 @@ class ColumnMatrix {
// all columns are dense column and has no missing value
// FIXME(jiamingy): We don't need a column matrix if there's no missing value.
template <typename RowBinIdxT>
void SetIndexNoMissing(SparsePage const& page, RowBinIdxT const* row_index,
const size_t n_samples, const size_t n_features, int32_t n_threads) {
void SetIndexNoMissing(RowBinIdxT const* row_index, const size_t n_samples,
const size_t n_features, int32_t n_threads) {
DispatchBinType(bins_type_size_, [&](auto t) {
using ColumnBinT = decltype(t);
auto column_index = Span<ColumnBinT>{reinterpret_cast<ColumnBinT*>(index_.data()),
Expand All @@ -263,10 +272,12 @@ class ColumnMatrix {
/**
* \brief Set column index for both dense and sparse columns
*/
void SetIndexMixedColumns(SparsePage const& page, uint32_t const* row_index,
const GHistIndexMatrix& gmat, size_t n_features) {
template <typename Batch>
void SetIndexMixedColumns(Batch const& batch, uint32_t const* row_index,
const GHistIndexMatrix& gmat, size_t n_features, float missing) {
std::vector<size_t> num_nonzeros;
num_nonzeros.resize(n_features, 0);
auto is_valid = data::IsValidFunctor {missing};

DispatchBinType(bins_type_size_, [&](auto t) {
using ColumnBinT = decltype(t);
Expand All @@ -276,7 +287,8 @@ class ColumnMatrix {
if (type_[fid] == kDenseColumn) {
ColumnBinT* begin = &local_index[feature_offsets_[fid]];
begin[rid] = bin_id - index_base_[fid];
// not thread-safe with bool vector.
// not thread-safe with bool vector. FIXME(jiamingy): We can directly assign
// kMissingId to the index to avoid missing flags.
missing_flags_[feature_offsets_[fid] + rid] = false;
} else {
ColumnBinT* begin = &local_index[feature_offsets_[fid]];
Expand All @@ -286,22 +298,18 @@ class ColumnMatrix {
}
};

const xgboost::Entry* data_ptr = page.data.HostVector().data();
const std::vector<bst_row_t>& offset_vec = page.offset.HostVector();
const size_t batch_size = gmat.Size();
CHECK_LT(batch_size, offset_vec.size());
size_t k{0};
for (size_t rid = 0; rid < batch_size; ++rid) {
const size_t ibegin = gmat.row_ptr[rid];
const size_t iend = gmat.row_ptr[rid + 1];
const size_t size = offset_vec[rid + 1] - offset_vec[rid];
SparsePage::Inst inst = {data_ptr + offset_vec[rid], size};

CHECK_EQ(ibegin + inst.size(), iend);
size_t j = 0;
for (size_t i = ibegin; i < iend; ++i, ++j) {
const uint32_t bin_id = row_index[i];
auto fid = inst[j].index;
get_bin_idx(bin_id, rid, fid);
auto line = batch.GetLine(rid);
for (size_t i = 0; i < line.Size(); ++i) {
auto coo = line.GetElement(i);
if (is_valid(coo)) {
auto fid = coo.column_idx;
const uint32_t bin_id = row_index[k];
get_bin_idx(bin_id, rid, fid);
++k;
}
}
}
});
Expand Down
2 changes: 2 additions & 0 deletions src/common/hist_util.h
View file
Expand Up @@ -211,6 +211,8 @@ auto DispatchBinType(BinTypeSize type, Fn&& fn) {
return fn(uint32_t{});
}
}
LOG(FATAL) << "Unreachable";
return fn(uint32_t{});
}

/**
Expand Down
18 changes: 18 additions & 0 deletions src/data/adapter.h
View file
Expand Up @@ -1131,6 +1131,24 @@ class RecordBatchesIterAdapter: public dmlc::DataIter<ArrowColumnarBatchVec> {
struct ArrowSchemaImporter schema_;
ArrowColumnarBatchVec batches_;
};

class SparsePageAdapterBatch {
HostSparsePageView page_;

public:
struct Line {
SparsePage::Inst inst;
bst_row_t ridx;
COOTuple GetElement(size_t idx) const {
return COOTuple{ridx, inst.data()[idx].index, inst.data()[idx].fvalue};
}
size_t Size() const { return inst.size(); }
};

explicit SparsePageAdapterBatch(HostSparsePageView page) : page_{std::move(page)} {}
Line GetLine(size_t ridx) const { return Line{page_[ridx], ridx}; }
size_t Size() const { return page_.Size(); }
};
}; // namespace data
} // namespace xgboost
#endif // XGBOOST_DATA_ADAPTER_H_
77 changes: 29 additions & 48 deletions tests/cpp/common/test_column_matrix.cc
View file
Expand Up @@ -31,34 +31,33 @@ TEST(DenseColumn, Test) {
ASSERT_FALSE(column_matrix.AnyMissing());
for (auto i = 0ull; i < dmat->Info().num_row_; i++) {
for (auto j = 0ull; j < dmat->Info().num_col_; j++) {
switch (column_matrix.GetTypeSize()) {
case kUint8BinsTypeSize: {
auto col = column_matrix.DenseColumn<uint8_t, false>(j);
ASSERT_EQ(gmat.index[i * dmat->Info().num_col_ + j], col.GetGlobalBinIdx(i));
} break;
case kUint16BinsTypeSize: {
auto col = column_matrix.DenseColumn<uint16_t, false>(j);
ASSERT_EQ(gmat.index[i * dmat->Info().num_col_ + j], col.GetGlobalBinIdx(i));
} break;
case kUint32BinsTypeSize: {
auto col = column_matrix.DenseColumn<uint32_t, false>(j);
ASSERT_EQ(gmat.index[i * dmat->Info().num_col_ + j], col.GetGlobalBinIdx(i));
} break;
}
DispatchBinType(column_matrix.GetTypeSize(), [&](auto dtype) {
using T = decltype(dtype);
auto col = column_matrix.DenseColumn<T, false>(j);
ASSERT_EQ(gmat.index[i * dmat->Info().num_col_ + j], col.GetGlobalBinIdx(i));
});
}
}
}
}

template <typename BinIdxType>
inline void CheckSparseColumn(const SparseColumnIter<BinIdxType>& col_input,
const GHistIndexMatrix& gmat) {
const SparseColumnIter<BinIdxType>& col =
static_cast<const SparseColumnIter<BinIdxType>&>(col_input);
void CheckSparseColumn(SparseColumnIter<BinIdxType>* p_col, const GHistIndexMatrix& gmat) {
auto& col = *p_col;

size_t n_samples = gmat.row_ptr.size() - 1;
ASSERT_EQ(col.Size(), gmat.index.Size());
for (auto i = 0ull; i < col.Size(); i++) {
ASSERT_EQ(gmat.index[gmat.row_ptr[col.GetRowIdx(i)]], col.GetGlobalBinIdx(i));
}

for (auto i = 0ull; i < n_samples; i++) {
if (col[i] == Column<BinIdxType>::kMissingId) {
auto beg = gmat.row_ptr[i];
auto end = gmat.row_ptr[i + 1];
ASSERT_EQ(end - beg, 0);
}
}
}

TEST(SparseColumn, Test) {
Expand All @@ -72,26 +71,17 @@ TEST(SparseColumn, Test) {
for (auto const& page : dmat->GetBatches<SparsePage>()) {
column_matrix.Init(page, gmat, 1.0, common::OmpGetNumThreads(0));
}
switch (column_matrix.GetTypeSize()) {
case kUint8BinsTypeSize: {
auto col = column_matrix.SparseColumn<uint8_t>(0, 0);
CheckSparseColumn(col, gmat);
} break;
case kUint16BinsTypeSize: {
auto col = column_matrix.SparseColumn<uint16_t>(0, 0);
CheckSparseColumn(col, gmat);
} break;
case kUint32BinsTypeSize: {
auto col = column_matrix.SparseColumn<uint32_t>(0, 0);
CheckSparseColumn(col, gmat);
} break;
}
common::DispatchBinType(column_matrix.GetTypeSize(), [&](auto dtype) {
using T = decltype(dtype);
auto col = column_matrix.SparseColumn<T>(0, 0);
CheckSparseColumn(&col, gmat);
});
}
}

template <typename BinIdxType>
inline void CheckColumWithMissingValue(const DenseColumnIter<BinIdxType, true>& col,
const GHistIndexMatrix& gmat) {
void CheckColumWithMissingValue(const DenseColumnIter<BinIdxType, true>& col,
const GHistIndexMatrix& gmat) {
for (auto i = 0ull; i < col.Size(); i++) {
if (col.IsMissing(i)) continue;
EXPECT_EQ(gmat.index[gmat.row_ptr[i]], col.GetGlobalBinIdx(i));
Expand All @@ -110,20 +100,11 @@ TEST(DenseColumnWithMissing, Test) {
column_matrix.Init(page, gmat, 0.2, common::OmpGetNumThreads(0));
}
ASSERT_TRUE(column_matrix.AnyMissing());
switch (column_matrix.GetTypeSize()) {
case kUint8BinsTypeSize: {
auto col = column_matrix.DenseColumn<uint8_t, true>(0);
CheckColumWithMissingValue(col, gmat);
} break;
case kUint16BinsTypeSize: {
auto col = column_matrix.DenseColumn<uint16_t, true>(0);
CheckColumWithMissingValue(col, gmat);
} break;
case kUint32BinsTypeSize: {
auto col = column_matrix.DenseColumn<uint32_t, true>(0);
CheckColumWithMissingValue(col, gmat);
} break;
}
DispatchBinType(column_matrix.GetTypeSize(), [&](auto dtype) {
using T = decltype(dtype);
auto col = column_matrix.DenseColumn<T, true>(0);
CheckColumWithMissingValue(col, gmat);
});
}
}

Expand Down

0 comments on commit 19775ff

Please sign in to comment.