Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Avoid the use of size_t in the partitioner. (#10541)

Browse Source 
- Avoid the use of size_t in the partitioner.
- Use `Span` instead of `Elem` where `node_id` is not needed.
- Remove the `const_cast`.
- Make sure the constness is not removed in the `Elem` by making it reference only.

size_t is implementation-defined, which causes issue when we want to pass pointer or span.
This commit is contained in:
Jiaming Yuan
2024-07-11 00:43:08 +08:00
committed by GitHub
parent baba3e9eb0
commit 34b154c284
12 changed files with 203 additions and 181 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
tests/cpp/tree/hist/test_evaluate_splits.cc
View File

@@ -45,7 +45,7 @@ void TestEvaluateSplits(bool force_read_by_column) {
// dense, no missing values
GHistIndexMatrix gmat(&ctx, dmat.get(), kMaxBins, 0.5, false);
common::RowSetCollection row_set_collection;
std::vector<size_t> &row_indices = *row_set_collection.Data();
std::vector<bst_idx_t> &row_indices = *row_set_collection.Data();
row_indices.resize(kRows);
std::iota(row_indices.begin(), row_indices.end(), 0);
row_set_collection.Init();
@@ -53,7 +53,9 @@ void TestEvaluateSplits(bool force_read_by_column) {
HistMakerTrainParam hist_param;
hist.Reset(gmat.cut.Ptrs().back(), hist_param.max_cached_hist_node);
hist.AllocateHistograms({0});
common::BuildHist<false>(row_gpairs, row_set_collection[0], gmat, hist[0], force_read_by_column);
auto const &elem = row_set_collection[0];
common::BuildHist<false>(row_gpairs, common::Span{elem.begin(), elem.end()}, gmat, hist[0],
force_read_by_column);
// Compute total gradient for all data points
GradientPairPrecise total_gpair;

7
tests/cpp/tree/hist/test_histogram.cc
View File

@@ -14,7 +14,6 @@
#include <algorithm> // for max
#include <cstddef> // for size_t
#include <cstdint> // for int32_t, uint32_t
#include <functional> // for function
#include <iterator> // for back_inserter
#include <limits> // for numeric_limits
#include <memory> // for shared_ptr, allocator, unique_ptr
@@ -108,7 +107,7 @@ void TestSyncHist(bool is_distributed) {
common::RowSetCollection row_set_collection;
{
row_set_collection.Clear();
std::vector<size_t> &row_indices = *row_set_collection.Data();
std::vector<bst_idx_t> &row_indices = *row_set_collection.Data();
row_indices.resize(kNRows);
std::iota(row_indices.begin(), row_indices.end(), 0);
row_set_collection.Init();
@@ -251,7 +250,7 @@ void TestBuildHistogram(bool is_distributed, bool force_read_by_column, bool is_
common::RowSetCollection row_set_collection;
row_set_collection.Clear();
std::vector<size_t> &row_indices = *row_set_collection.Data();
std::vector<bst_idx_t> &row_indices = *row_set_collection.Data();
row_indices.resize(kNRows);
std::iota(row_indices.begin(), row_indices.end(), 0);
row_set_collection.Init();
@@ -345,7 +344,7 @@ void TestHistogramCategorical(size_t n_categories, bool force_read_by_column) {
common::RowSetCollection row_set_collection;
row_set_collection.Clear();
std::vector<size_t> &row_indices = *row_set_collection.Data();
std::vector<bst_idx_t> &row_indices = *row_set_collection.Data();
row_indices.resize(kRows);
std::iota(row_indices.begin(), row_indices.end(), 0);
row_set_collection.Init();