[CUDA] consolidate CUDA versions (#5677)

* [ci] speed up if-else, swig, and lint conda setup

* add 'source activate'

* python constraint

* start removing cuda v1

* comment out CI

* remove more references

* revert some unnecessaary changes

* revert a few more mistakes

* revert another change that ignored params

* sigh

* remove CUDATreeLearner

* fix tests, docs

* fix quoting in setup.py

* restore all CI

* Apply suggestions from code review
Co-authored-by: shiyu1994 <shiyu_k1994@qq.com>

* Apply suggestions from code review

* completely remove cuda_exp, update docs

---------
Co-authored-by: shiyu1994 <shiyu_k1994@qq.com>

tests/python_package_test/test_dask.py

@@ -57,7 +57,7 @@ task_to_local_factory = {
 
 pytestmark = [
     pytest.mark.skipif(getenv('TASK', '') == 'mpi', reason='Fails to run with MPI interface'),
-    pytest.mark.skipif(getenv('TASK', '') == 'cuda_exp', reason='Fails to run with CUDA Experimental interface')
+    pytest.mark.skipif(getenv('TASK', '') == 'cuda', reason='Fails to run with CUDA interface')
 ]
 
 

tests/python_package_test/test_engine.py

@@ -588,7 +588,7 @@ def test_multi_class_error():
     assert results['training']['multi_error@2'][-1] == pytest.approx(0)
 
 
-@pytest.mark.skipif(getenv('TASK', '') == 'cuda_exp', reason='Skip due to differences in implementation details of CUDA Experimental version')
+@pytest.mark.skipif(getenv('TASK', '') == 'cuda', reason='Skip due to differences in implementation details of CUDA version')
 def test_auc_mu():
     # should give same result as binary auc for 2 classes
     X, y = load_digits(n_class=10, return_X_y=True)
@@ -1661,7 +1661,7 @@ def generate_trainset_for_monotone_constraints_tests(x3_to_category=True):
     return trainset
 
 
-@pytest.mark.skipif(getenv('TASK', '') == 'cuda_exp', reason='Monotone constraints are not yet supported by CUDA Experimental version')
+@pytest.mark.skipif(getenv('TASK', '') == 'cuda', reason='Monotone constraints are not yet supported by CUDA version')
 @pytest.mark.parametrize("test_with_categorical_variable", [True, False])
 def test_monotone_constraints(test_with_categorical_variable):
     def is_increasing(y):
@@ -1751,7 +1751,7 @@ def test_monotone_constraints(test_with_categorical_variable):
                 assert are_interactions_enforced(constrained_model, feature_sets)
 
 
-@pytest.mark.skipif(getenv('TASK', '') == 'cuda_exp', reason='Monotone constraints are not yet supported by CUDA Experimental version')
+@pytest.mark.skipif(getenv('TASK', '') == 'cuda', reason='Monotone constraints are not yet supported by CUDA version')
 def test_monotone_penalty():
     def are_first_splits_non_monotone(tree, n, monotone_constraints):
         if n <= 0:
@@ -1791,7 +1791,7 @@ def test_monotone_penalty():
 
 
 # test if a penalty as high as the depth indeed prohibits all monotone splits
-@pytest.mark.skipif(getenv('TASK', '') == 'cuda_exp', reason='Monotone constraints are not yet supported by CUDA Experimental version')
+@pytest.mark.skipif(getenv('TASK', '') == 'cuda', reason='Monotone constraints are not yet supported by CUDA version')
 def test_monotone_penalty_max():
     max_depth = 5
     monotone_constraints = [1, -1, 0]
@@ -2652,7 +2652,7 @@ def test_model_size():
         pytest.skipTest('not enough RAM')
 
 
-@pytest.mark.skipif(getenv('TASK', '') == 'cuda_exp', reason='Skip due to differences in implementation details of CUDA Experimental version')
+@pytest.mark.skipif(getenv('TASK', '') == 'cuda', reason='Skip due to differences in implementation details of CUDA version')
 def test_get_split_value_histogram():
     X, y = make_synthetic_regression()
     X = np.repeat(X, 3, axis=0)
@@ -2735,7 +2735,7 @@ def test_get_split_value_histogram():
         gbm.get_split_value_histogram(2)
 
 
-@pytest.mark.skipif(getenv('TASK', '') == 'cuda_exp', reason='Skip due to differences in implementation details of CUDA Experimental version')
+@pytest.mark.skipif(getenv('TASK', '') == 'cuda', reason='Skip due to differences in implementation details of CUDA version')
 def test_early_stopping_for_only_first_metric():
 
     def metrics_combination_train_regression(valid_sets, metric_list, assumed_iteration,
@@ -3573,7 +3573,7 @@ def test_dump_model_hook():
     assert "LV" in dumped_model_str
 
 
-@pytest.mark.skipif(getenv('TASK', '') == 'cuda_exp', reason='Forced splits are not yet supported by CUDA Experimental version')
+@pytest.mark.skipif(getenv('TASK', '') == 'cuda', reason='Forced splits are not yet supported by CUDA version')
 def test_force_split_with_feature_fraction(tmp_path):
     X, y = make_synthetic_regression()
     X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.1, random_state=42)

tests/python_package_test/test_sklearn.py

@@ -121,7 +121,7 @@ def test_regression():
     assert gbm.evals_result_['valid_0']['l2'][gbm.best_iteration_ - 1] == pytest.approx(ret)
 
 
-@pytest.mark.skipif(getenv('TASK', '') == 'cuda_exp', reason='Skip due to differences in implementation details of CUDA Experimental version')
+@pytest.mark.skipif(getenv('TASK', '') == 'cuda', reason='Skip due to differences in implementation details of CUDA version')
 def test_multiclass():
     X, y = load_digits(n_class=10, return_X_y=True)
     X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.1, random_state=42)
@@ -134,7 +134,7 @@ def test_multiclass():
     assert gbm.evals_result_['valid_0']['multi_logloss'][gbm.best_iteration_ - 1] == pytest.approx(ret)
 
 
-@pytest.mark.skipif(getenv('TASK', '') == 'cuda_exp', reason='Skip due to differences in implementation details of CUDA Experimental version')
+@pytest.mark.skipif(getenv('TASK', '') == 'cuda', reason='Skip due to differences in implementation details of CUDA version')
 def test_lambdarank():
     rank_example_dir = Path(__file__).absolute().parents[2] / 'examples' / 'lambdarank'
     X_train, y_train = load_svmlight_file(str(rank_example_dir / 'rank.train'))
@@ -1091,7 +1091,7 @@ def test_nan_handle():
     np.testing.assert_allclose(gbm.evals_result_['training']['l2'], np.nan)
 
 
-@pytest.mark.skipif(getenv('TASK', '') == 'cuda_exp', reason='Skip due to differences in implementation details of CUDA Experimental version')
+@pytest.mark.skipif(getenv('TASK', '') == 'cuda', reason='Skip due to differences in implementation details of CUDA version')
 def test_first_metric_only():
 
     def fit_and_check(eval_set_names, metric_names, assumed_iteration, first_metric_only):

tests/python_package_test/test_utilities.py

@@ -91,15 +91,15 @@ WARNING | More than one metric available, picking one to plot.
         "INFO | [LightGBM] [Warning] CUDA currently requires double precision calculations.",
         "INFO | [LightGBM] [Info] LightGBM using CUDA trainer with DP float!!"
     ]
-    cuda_exp_lines = [
-        "INFO | [LightGBM] [Warning] Metric auc is not implemented in cuda_exp version. Fall back to evaluation on CPU.",
-        "INFO | [LightGBM] [Warning] Metric binary_error is not implemented in cuda_exp version. Fall back to evaluation on CPU.",
+    cuda_lines = [
+        "INFO | [LightGBM] [Warning] Metric auc is not implemented in cuda version. Fall back to evaluation on CPU.",
+        "INFO | [LightGBM] [Warning] Metric binary_error is not implemented in cuda version. Fall back to evaluation on CPU.",
     ]
     with open(log_filename, "rt", encoding="utf-8") as f:
         actual_log = f.read().strip()
         actual_log_wo_gpu_stuff = []
         for line in actual_log.split("\n"):
-            if not any(line.startswith(gpu_or_cuda_exp_line) for gpu_or_cuda_exp_line in gpu_lines + cuda_exp_lines):
+            if not any(line.startswith(gpu_or_cuda_line) for gpu_or_cuda_line in gpu_lines + cuda_lines):
                 actual_log_wo_gpu_stuff.append(line)
 
     assert "\n".join(actual_log_wo_gpu_stuff) == expected_log