bc267dd729
* Use ptr from mmap for `GHistIndexMatrix` and `ColumnMatrix`. - Define a resource for holding various types of memory pointers. - Define ref vector for holding resources. - Swap the underlying resources for GHist and ColumnM. - Add documentation for current status. - s390x support is removed. It should work if you can compile XGBoost, all the old workaround code does is to get GCC to compile.
76 lines
2.1 KiB
C++
76 lines
2.1 KiB
C++
/**
|
|
* Copyright 2015-2023, XGBoost Contributors
|
|
* \file sparse_page_raw_format.cc
|
|
* Raw binary format of sparse page.
|
|
*/
|
|
#include <dmlc/registry.h>
|
|
|
|
#include "../common/io.h" // for AlignedResourceReadStream, AlignedFileWriteStream
|
|
#include "../common/ref_resource_view.h" // for WriteVec
|
|
#include "./sparse_page_writer.h"
|
|
#include "xgboost/data.h"
|
|
#include "xgboost/logging.h"
|
|
|
|
namespace xgboost::data {
|
|
DMLC_REGISTRY_FILE_TAG(sparse_page_raw_format);
|
|
|
|
template <typename T>
|
|
class SparsePageRawFormat : public SparsePageFormat<T> {
|
|
public:
|
|
bool Read(T* page, common::AlignedResourceReadStream* fi) override {
|
|
auto& offset_vec = page->offset.HostVector();
|
|
if (!common::ReadVec(fi, &offset_vec)) {
|
|
return false;
|
|
}
|
|
auto& data_vec = page->data.HostVector();
|
|
CHECK_NE(page->offset.Size(), 0U) << "Invalid SparsePage file";
|
|
data_vec.resize(offset_vec.back());
|
|
if (page->data.Size() != 0) {
|
|
if (!common::ReadVec(fi, &data_vec)) {
|
|
return false;
|
|
}
|
|
}
|
|
if (!fi->Read(&page->base_rowid, sizeof(page->base_rowid))) {
|
|
return false;
|
|
}
|
|
return true;
|
|
}
|
|
|
|
std::size_t Write(const T& page, common::AlignedFileWriteStream* fo) override {
|
|
const auto& offset_vec = page.offset.HostVector();
|
|
const auto& data_vec = page.data.HostVector();
|
|
CHECK(page.offset.Size() != 0 && offset_vec[0] == 0);
|
|
CHECK_EQ(offset_vec.back(), page.data.Size());
|
|
|
|
std::size_t bytes{0};
|
|
bytes += common::WriteVec(fo, offset_vec);
|
|
if (page.data.Size() != 0) {
|
|
bytes += common::WriteVec(fo, data_vec);
|
|
}
|
|
bytes += fo->Write(&page.base_rowid, sizeof(page.base_rowid));
|
|
return bytes;
|
|
}
|
|
|
|
private:
|
|
};
|
|
|
|
XGBOOST_REGISTER_SPARSE_PAGE_FORMAT(raw)
|
|
.describe("Raw binary data format.")
|
|
.set_body([]() {
|
|
return new SparsePageRawFormat<SparsePage>();
|
|
});
|
|
|
|
XGBOOST_REGISTER_CSC_PAGE_FORMAT(raw)
|
|
.describe("Raw binary data format.")
|
|
.set_body([]() {
|
|
return new SparsePageRawFormat<CSCPage>();
|
|
});
|
|
|
|
XGBOOST_REGISTER_SORTED_CSC_PAGE_FORMAT(raw)
|
|
.describe("Raw binary data format.")
|
|
.set_body([]() {
|
|
return new SparsePageRawFormat<SortedCSCPage>();
|
|
});
|
|
|
|
} // namespace xgboost::data
|