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

Deterministic data partitioning for external memory (#6317)

Browse Source 
* Make external memory data partitioning deterministic.

* Change the meaning of `page_size` from bytes to number of rows.

* Design a data pool.

* Note for external memory.

* Enable unity build on Windows CI.

* Force garbage collect on test.
This commit is contained in:
Jiaming Yuan
2020-11-11 06:11:06 +08:00
committed by GitHub
parent 9564886d9f
commit 43efadea2e
15 changed files with 334 additions and 88 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
tests/cpp/predictor/test_cpu_predictor.cc
View File

@@ -16,8 +16,8 @@ TEST(CpuPredictor, Basic) {
std::unique_ptr<Predictor> cpu_predictor =
std::unique_ptr<Predictor>(Predictor::Create("cpu_predictor", &lparam));
int kRows = 5;
int kCols = 5;
size_t constexpr kRows = 5;
size_t constexpr kCols = 5;
LearnerModelParam param;
param.num_feature = kCols;
@@ -85,7 +85,11 @@ TEST(CpuPredictor, Basic) {
TEST(CpuPredictor, ExternalMemory) {
dmlc::TemporaryDirectory tmpdir;
std::string filename = tmpdir.path + "/big.libsvm";
std::unique_ptr<DMatrix> dmat = CreateSparsePageDMatrix(12, 64, filename);
size_t constexpr kPageSize = 64, kEntriesPerCol = 3;
size_t constexpr kEntries = kPageSize * kEntriesPerCol * 2;
std::unique_ptr<DMatrix> dmat = CreateSparsePageDMatrix(kEntries, kPageSize, filename);
auto lparam = CreateEmptyGenericParam(GPUIDX);
std::unique_ptr<Predictor> cpu_predictor =

6
tests/cpp/predictor/test_gpu_predictor.cu
View File

@@ -105,9 +105,9 @@ TEST(GPUPredictor, ExternalMemoryTest) {
std::string file0 = tmpdir.path + "/big_0.libsvm";
std::string file1 = tmpdir.path + "/big_1.libsvm";
std::string file2 = tmpdir.path + "/big_2.libsvm";
dmats.push_back(CreateSparsePageDMatrix(9, 64UL, file0));
dmats.push_back(CreateSparsePageDMatrix(128, 128UL, file1));
dmats.push_back(CreateSparsePageDMatrix(1024, 1024UL, file2));
dmats.push_back(CreateSparsePageDMatrix(400, 64UL, file0));
dmats.push_back(CreateSparsePageDMatrix(800, 128UL, file1));
dmats.push_back(CreateSparsePageDMatrix(8000, 1024UL, file2));
for (const auto& dmat: dmats) {
dmat->Info().base_margin_.Resize(dmat->Info().num_row_ * n_classes, 0.5);