Update checks of HAS_GPU in tests to handle case where cudf is not installed (#118)

Update checks of HAS_GPU to handle case where cudf is not installed

Co-authored-by: Karl Higley <[email protected]>
Co-authored-by: edknv <[email protected]>
Co-authored-by: Julio Perez <[email protected]>

Author: 4 people

tests/unit/dataloader/test_tf_dataloader.py

@@ -26,7 +26,7 @@
 import pytest
 from sklearn.metrics import roc_auc_score
 
-from merlin.core.compat import HAS_GPU, cupy
+from merlin.core.compat import HAS_GPU, cudf, cupy
 from merlin.core.dispatch import make_df, random_uniform
 from merlin.io import Dataset
 from merlin.schema import Tags
@@ -357,7 +357,7 @@ def add_sample_weight(features, labels, sample_weight_col_name="sample_weight"):
 # TODO: include parts_per_chunk test
 @pytest.mark.parametrize("gpu_memory_frac", [0.01, 0.06])
 @pytest.mark.parametrize("batch_size", [1, 10, 100])
-@pytest.mark.parametrize("cpu", [False, True] if HAS_GPU else [True])
+@pytest.mark.parametrize("cpu", [False, True] if HAS_GPU and cudf else [True])
 def test_tensorflow_dataloader(
     tmpdir,
     cpu,
@@ -633,7 +633,7 @@ def test_horovod_multigpu(tmpdir):
 
 
 @pytest.mark.parametrize("batch_size", [1000])
-@pytest.mark.parametrize("cpu", [False, True] if HAS_GPU else [True])
+@pytest.mark.parametrize("cpu", [False, True] if HAS_GPU and cudf else [True])
 def test_dataloader_schema(tmpdir, dataset, batch_size, cpu):
     with tf_loader(
         dataset,

tests/unit/dataloader/test_torch_dataloader.py

@@ -23,7 +23,7 @@
 from conftest import assert_eq
 
 from merlin.core import dispatch
-from merlin.core.compat import HAS_GPU
+from merlin.core.compat import HAS_GPU, cudf
 from merlin.core.dispatch import make_df
 from merlin.io import Dataset
 from merlin.schema import Tags
@@ -150,7 +150,7 @@ def test_torch_drp_reset(tmpdir, batch_size, drop_last, num_rows):
                     # Each column has only one unique value
                     # We test that each value in chunk (output of dataloader)
                     # is equal to every value in dataframe
-                    if dispatch.HAS_GPU:
+                    if cudf and isinstance(df, cudf.DataFrame):
                         assert (
                             np.expand_dims(chunk[0][col].cpu().numpy(), 1) == df[col].values_host
                         ).all()
@@ -224,7 +224,7 @@ def test_gpu_file_iterator_ds(df, dataset, batch):
 
 @pytest.mark.parametrize("part_mem_fraction", [0.001, 0.06])
 @pytest.mark.parametrize("batch_size", [1000])
-@pytest.mark.parametrize("cpu", [False, True] if HAS_GPU else [True])
+@pytest.mark.parametrize("cpu", [False, True] if HAS_GPU and cudf else [True])
 def test_dataloader_break(dataset, batch_size, part_mem_fraction, cpu):
     dataloader = torch_loader(
         dataset,
@@ -257,7 +257,7 @@ def test_dataloader_break(dataset, batch_size, part_mem_fraction, cpu):
 
 @pytest.mark.parametrize("part_mem_fraction", [0.001, 0.06])
 @pytest.mark.parametrize("batch_size", [1000])
-@pytest.mark.parametrize("cpu", [False, True] if HAS_GPU else [True])
+@pytest.mark.parametrize("cpu", [False, True] if HAS_GPU and cudf else [True])
 def test_dataloader(df, dataset, batch_size, part_mem_fraction, cpu):
     dataloader = torch_loader(
         dataset,
@@ -336,7 +336,7 @@ def test_mh_support(multihot_dataset):
 
 
 @pytest.mark.parametrize("batch_size", [1000])
-@pytest.mark.parametrize("cpu", [False, True] if HAS_GPU else [True])
+@pytest.mark.parametrize("cpu", [False, True] if HAS_GPU and cudf else [True])
 def test_dataloader_schema(df, dataset, batch_size, cpu):
     with torch_loader(
         dataset,