[EM] Add basic distributed GPU tests. (#10861)

- Split Hist and Approx tests in unittests.
- Basic GPU tests for distributed.

python-package/xgboost/testing/dask.py

@@ -1,14 +1,16 @@
 """Tests for dask shared by different test modules."""
 
-from typing import Literal
+from typing import List, Literal, cast
 
 import numpy as np
 import pandas as pd
 from dask import array as da
 from dask import dataframe as dd
-from distributed import Client
+from distributed import Client, get_worker
 
 import xgboost as xgb
+import xgboost.testing as tm
+from xgboost.compat import concat
 from xgboost.testing.updater import get_basescore
 
 
@@ -91,3 +93,76 @@ def check_uneven_nan(
             dd.from_pandas(X, npartitions=n_workers),
             dd.from_pandas(y, npartitions=n_workers),
         )
+
+
+def check_external_memory(  # pylint: disable=too-many-locals
+    worker_id: int,
+    n_workers: int,
+    device: str,
+    comm_args: dict,
+    is_qdm: bool,
+) -> None:
+    """Basic checks for distributed external memory."""
+    n_samples_per_batch = 32
+    n_features = 4
+    n_batches = 16
+    use_cupy = device != "cpu"
+
+    n_threads = get_worker().state.nthreads
+    with xgb.collective.CommunicatorContext(dmlc_communicator="rabit", **comm_args):
+        it = tm.IteratorForTest(
+            *tm.make_batches(
+                n_samples_per_batch,
+                n_features,
+                n_batches,
+                use_cupy=use_cupy,
+                random_state=worker_id,
+            ),
+            cache="cache",
+        )
+        if is_qdm:
+            Xy: xgb.DMatrix = xgb.ExtMemQuantileDMatrix(it, nthread=n_threads)
+        else:
+            Xy = xgb.DMatrix(it, nthread=n_threads)
+        results: xgb.callback.TrainingCallback.EvalsLog = {}
+        xgb.train(
+            {"tree_method": "hist", "nthread": n_threads, "device": device},
+            Xy,
+            evals=[(Xy, "Train")],
+            num_boost_round=32,
+            evals_result=results,
+        )
+        assert tm.non_increasing(cast(List[float], results["Train"]["rmse"]))
+
+    lx, ly, lw = [], [], []
+    for i in range(n_workers):
+        x, y, w = tm.make_batches(
+            n_samples_per_batch,
+            n_features,
+            n_batches,
+            use_cupy=use_cupy,
+            random_state=i,
+        )
+        lx.extend(x)
+        ly.extend(y)
+        lw.extend(w)
+
+    X = concat(lx)
+    yconcat = concat(ly)
+    wconcat = concat(lw)
+    if is_qdm:
+        Xy = xgb.QuantileDMatrix(X, yconcat, weight=wconcat, nthread=n_threads)
+    else:
+        Xy = xgb.DMatrix(X, yconcat, weight=wconcat, nthread=n_threads)
+
+    results_local: xgb.callback.TrainingCallback.EvalsLog = {}
+    xgb.train(
+        {"tree_method": "hist", "nthread": n_threads, "device": device},
+        Xy,
+        evals=[(Xy, "Train")],
+        num_boost_round=32,
+        evals_result=results_local,
+    )
+    np.testing.assert_allclose(
+        results["Train"]["rmse"], results_local["Train"]["rmse"], rtol=1e-4
+    )

tests/cpp/tree/test_gpu_hist.cu

@@ -318,55 +318,4 @@ TEST_F(MGPUHistTest, HistColumnSplit) {
   this->DoTest([&] { VerifyHistColumnSplit(kRows, kCols, expected_tree); }, true);
   this->DoTest([&] { VerifyHistColumnSplit(kRows, kCols, expected_tree); }, false);
 }
-
-namespace {
-RegTree GetApproxTree(Context const* ctx, DMatrix* dmat) {
-  ObjInfo task{ObjInfo::kRegression};
-  std::unique_ptr<TreeUpdater> approx_maker{TreeUpdater::Create("grow_gpu_approx", ctx, &task)};
-  approx_maker->Configure(Args{});
-
-  TrainParam param;
-  param.UpdateAllowUnknown(Args{});
-
-  linalg::Matrix<GradientPair> gpair({dmat->Info().num_row_}, ctx->Device());
-  gpair.Data()->Copy(GenerateRandomGradients(dmat->Info().num_row_));
-
-  std::vector<HostDeviceVector<bst_node_t>> position(1);
-  RegTree tree;
-  approx_maker->Update(&param, &gpair, dmat, common::Span<HostDeviceVector<bst_node_t>>{position},
-                       {&tree});
-  return tree;
-}
-
-void VerifyApproxColumnSplit(bst_idx_t rows, bst_feature_t cols, RegTree const& expected_tree) {
-  auto ctx = MakeCUDACtx(DistGpuIdx());
-
-  auto Xy = RandomDataGenerator{rows, cols, 0}.GenerateDMatrix(true);
-  auto const world_size = collective::GetWorldSize();
-  auto const rank = collective::GetRank();
-  std::unique_ptr<DMatrix> sliced{Xy->SliceCol(world_size, rank)};
-
-  RegTree tree = GetApproxTree(&ctx, sliced.get());
-
-  Json json{Object{}};
-  tree.SaveModel(&json);
-  Json expected_json{Object{}};
-  expected_tree.SaveModel(&expected_json);
-  ASSERT_EQ(json, expected_json);
-}
-}  // anonymous namespace
-
-class MGPUApproxTest : public collective::BaseMGPUTest {};
-
-TEST_F(MGPUApproxTest, GPUApproxColumnSplit) {
-  auto constexpr kRows = 32;
-  auto constexpr kCols = 16;
-
-  Context ctx(MakeCUDACtx(0));
-  auto dmat = RandomDataGenerator{kRows, kCols, 0}.GenerateDMatrix(true);
-  RegTree expected_tree = GetApproxTree(&ctx, dmat.get());
-
-  this->DoTest([&] { VerifyApproxColumnSplit(kRows, kCols, expected_tree); }, true);
-  this->DoTest([&] { VerifyApproxColumnSplit(kRows, kCols, expected_tree); }, false);
-}
 }  // namespace xgboost::tree

tests/test_distributed/test_with_dask/test_external_memory.py

@@ -1,77 +1,18 @@
-from typing import List, cast
+"""Copyright 2024, XGBoost contributors"""
 
-import numpy as np
-from distributed import Client, Scheduler, Worker, get_worker
+import pytest
+from distributed import Client, Scheduler, Worker
 from distributed.utils_test import gen_cluster
 
 import xgboost as xgb
 from xgboost import testing as tm
-from xgboost.compat import concat
-
-
-def run_external_memory(worker_id: int, n_workers: int, comm_args: dict) -> None:
-    n_samples_per_batch = 32
-    n_features = 4
-    n_batches = 16
-    use_cupy = False
-
-    n_threads = get_worker().state.nthreads
-    with xgb.collective.CommunicatorContext(dmlc_communicator="rabit", **comm_args):
-        it = tm.IteratorForTest(
-            *tm.make_batches(
-                n_samples_per_batch,
-                n_features,
-                n_batches,
-                use_cupy,
-                random_state=worker_id,
-            ),
-            cache="cache",
-        )
-        Xy = xgb.DMatrix(it, nthread=n_threads)
-        results: xgb.callback.TrainingCallback.EvalsLog = {}
-        booster = xgb.train(
-            {"tree_method": "hist", "nthread": n_threads},
-            Xy,
-            evals=[(Xy, "Train")],
-            num_boost_round=32,
-            evals_result=results,
-        )
-        assert tm.non_increasing(cast(List[float], results["Train"]["rmse"]))
-
-    lx, ly, lw = [], [], []
-    for i in range(n_workers):
-        x, y, w = tm.make_batches(
-            n_samples_per_batch,
-            n_features,
-            n_batches,
-            use_cupy,
-            random_state=i,
-        )
-        lx.extend(x)
-        ly.extend(y)
-        lw.extend(w)
-
-    X = concat(lx)
-    yconcat = concat(ly)
-    wconcat = concat(lw)
-    Xy = xgb.DMatrix(X, yconcat, weight=wconcat, nthread=n_threads)
-
-    results_local: xgb.callback.TrainingCallback.EvalsLog = {}
-    booster = xgb.train(
-        {"tree_method": "hist", "nthread": n_threads},
-        Xy,
-        evals=[(Xy, "Train")],
-        num_boost_round=32,
-        evals_result=results_local,
-    )
-    np.testing.assert_allclose(
-        results["Train"]["rmse"], results_local["Train"]["rmse"], rtol=1e-4
-    )
+from xgboost.testing.dask import check_external_memory
 
 
+@pytest.mark.parametrize("is_qdm", [True, False])
 @gen_cluster(client=True)
 async def test_external_memory(
-    client: Client, s: Scheduler, a: Worker, b: Worker
+    client: Client, s: Scheduler, a: Worker, b: Worker, is_qdm: bool
 ) -> None:
     workers = tm.get_client_workers(client)
     args = await client.sync(
@@ -83,6 +24,11 @@ async def test_external_memory(
     n_workers = len(workers)
 
     futs = client.map(
-        run_external_memory, range(n_workers), n_workers=n_workers, comm_args=args
+        check_external_memory,
+        range(n_workers),
+        n_workers=n_workers,
+        device="cpu",
+        comm_args=args,
+        is_qdm=is_qdm,
     )
     await client.gather(futs)

tests/test_distributed/test_with_dask/test_with_dask.py

@@ -7,24 +7,9 @@
 import socket
 import tempfile
 from concurrent.futures import ThreadPoolExecutor
-from copy import copy
 from functools import partial
-from itertools import starmap
-from math import ceil
-from operator import attrgetter, getitem
 from pathlib import Path
-from typing import (
-    Any,
-    Dict,
-    Generator,
-    List,
-    Literal,
-    Optional,
-    Tuple,
-    Type,
-    TypeVar,
-    Union,
-)
+from typing import Any, Dict, Generator, Literal, Optional, Tuple, Type, Union
 
 import hypothesis
 import numpy as np
@@ -37,7 +22,6 @@
 import xgboost as xgb
 from xgboost import dask as dxgb
 from xgboost import testing as tm
-from xgboost.data import _is_cudf_df
 from xgboost.testing.params import hist_cache_strategy, hist_parameter_strategy
 from xgboost.testing.shared import (
     get_feature_weights,