Deprecate num_processes,gpus, tpu_cores, and ipus from the Trainer constructor

CHANGELOG.md

@@ -459,6 +459,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Deprecated `ModelIO.on_hpc_{save/load}` in favor of `CheckpointHooks.on_{save/load}_checkpoint` ([#10911](https://github.com/PyTorchLightning/pytorch-lightning/pull/10911))
 
 
+- Deprecated `num_processes`, `gpus`, `tpu_cores,` and `ipus` from the `Trainer` constructor in favor of using the `accelerator` and `devices` arguments ([#11040](https://github.com/PyTorchLightning/pytorch-lightning/pull/11040))
+
+
 - Deprecated `Trainer.run_stage` in favor of `Trainer.{fit,validate,test,predict}` ([#11000](https://github.com/PyTorchLightning/pytorch-lightning/pull/11000))
 
 

pytorch_lightning/trainer/connectors/accelerator_connector.py

@@ -411,7 +411,7 @@ def _check_device_config_and_set_final_flags(
         self._devices_flag = devices
 
         # TODO: Delete this method when num_processes, gpus, ipus and tpu_cores gets removed
-        self._map_deprecated_devices_specfic_info_to_accelerator_and_device_flag(
+        self._map_deprecated_devices_specific_info_to_accelerator_and_device_flag(
             devices, num_processes, gpus, ipus, tpu_cores
         )
 
@@ -421,15 +421,36 @@ def _check_device_config_and_set_final_flags(
                 " `accelerator=('auto'|'tpu'|'gpu'|'ipu'|'cpu'|'hpu)` for the devices mapping"
             )
 
-    def _map_deprecated_devices_specfic_info_to_accelerator_and_device_flag(
+    def _map_deprecated_devices_specific_info_to_accelerator_and_device_flag(
         self,
         devices: Optional[Union[List[int], str, int]],
         num_processes: Optional[int],
         gpus: Optional[Union[List[int], str, int]],
         ipus: Optional[int],
         tpu_cores: Optional[Union[List[int], str, int]],
     ) -> None:
-        """Sets the `devices_flag` and `accelerator_flag` based on num_processes, gpus, ipus, tpu_cores."""
+        """Emit deprecation warnings for num_processes, gpus, ipus, tpu_cores and set the `devices_flag` and
+        `accelerator_flag`."""
+        if num_processes is not None:
+            rank_zero_deprecation(
+                f"Setting `Trainer(num_processes={num_processes})` is deprecated in v1.6 and will be removed"
+                f" in v2.0. Please use `Trainer(accelerator='cpu', devices={num_processes})` instead."
+            )
+        if gpus is not None:
+            rank_zero_deprecation(
+                f"Setting `Trainer(gpus={gpus})` is deprecated in v1.6 and will be removed"
+                f" in v2.0. Please use `Trainer(accelerator='gpu', devices={gpus})` instead."
+            )
+        if tpu_cores is not None:
+            rank_zero_deprecation(
+                f"Setting `Trainer(tpu_cores={tpu_cores})` is deprecated in v1.6 and will be removed"
+                f" in v2.0. Please use `Trainer(accelerator='tpu', devices={tpu_cores})` instead."
+            )
+        if ipus is not None:
+            rank_zero_deprecation(
+                f"Setting `Trainer(ipus={ipus})` is deprecated in v1.6 and will be removed"
+                f" in v2.0. Please use `Trainer(accelerator='ipu', devices={ipus})` instead."
+            )
         self._gpus: Optional[Union[List[int], str, int]] = gpus
         self._tpu_cores: Optional[Union[List[int], str, int]] = tpu_cores
         deprecated_devices_specific_flag = num_processes or gpus or ipus or tpu_cores

pytorch_lightning/trainer/trainer.py

@@ -139,12 +139,12 @@ def __init__(
         gradient_clip_algorithm: Optional[str] = None,
         process_position: int = 0,
         num_nodes: int = 1,
-        num_processes: Optional[int] = None,
+        num_processes: Optional[int] = None,  # TODO: Remove in 2.0
         devices: Optional[Union[List[int], str, int]] = None,
-        gpus: Optional[Union[List[int], str, int]] = None,
+        gpus: Optional[Union[List[int], str, int]] = None,  # TODO: Remove in 2.0
         auto_select_gpus: bool = False,
-        tpu_cores: Optional[Union[List[int], str, int]] = None,
-        ipus: Optional[int] = None,
+        tpu_cores: Optional[Union[List[int], str, int]] = None,  # TODO: Remove in 2.0
+        ipus: Optional[int] = None,  # TODO: Remove in 2.0
         log_gpu_memory: Optional[str] = None,  # TODO: Remove in 1.7
         progress_bar_refresh_rate: Optional[int] = None,  # TODO: remove in v1.7
         enable_progress_bar: bool = True,
@@ -280,6 +280,10 @@ def __init__(
             gpus: Number of GPUs to train on (int) or which GPUs to train on (list or str) applied per node
                 Default: ``None``.
 
+                .. deprecated:: v1.6
+                    ``gpus`` has been deprecated in v1.6 and will be removed in v2.0.
+                    Please use ``accelerator='gpu'`` and ``devices=x`` instead.
+
             gradient_clip_val: The value at which to clip gradients. Passing ``gradient_clip_val=None`` disables
                 gradient clipping. If using Automatic Mixed Precision (AMP), the gradients will be unscaled before.
                 Default: ``None``.
@@ -380,6 +384,10 @@ def __init__(
             num_processes: Number of processes for distributed training with ``accelerator="cpu"``.
                 Default: ``1``.
 
+                .. deprecated:: v1.6
+                    ``num_processes`` has been deprecated in v1.6 and will be removed in v2.0.
+                    Please use ``accelerator='cpu'`` and ``devices=x`` instead.
+
             num_sanity_val_steps: Sanity check runs n validation batches before starting the training routine.
                 Set it to `-1` to run all batches in all validation dataloaders.
                 Default: ``2``.
@@ -420,9 +428,17 @@ def __init__(
             tpu_cores: How many TPU cores to train on (1 or 8) / Single TPU to train on (1)
                 Default: ``None``.
 
+                .. deprecated:: v1.6
+                    ``tpu_cores`` has been deprecated in v1.6 and will be removed in v2.0.
+                    Please use ``accelerator='tpu'`` and ``devices=x`` instead.
+
             ipus: How many IPUs to train on.
                 Default: ``None``.
 
+                .. deprecated:: v1.6
+                    ``ipus`` has been deprecated in v1.6 and will be removed in v2.0.
+                    Please use ``accelerator='ipu'`` and ``devices=x`` instead.
+
             track_grad_norm: -1 no tracking. Otherwise tracks that p-norm. May be set to 'inf' infinity-norm. If using
                 Automatic Mixed Precision (AMP), the gradients will be unscaled before logging them.
                 Default: ``-1``.

tests/accelerators/test_accelerator_connector.py

@@ -49,7 +49,6 @@
 from tests.helpers.runif import RunIf
 
 
-# TODO: please modify/sunset any test that has accelerator=ddp/ddp2/ddp_cpu/ddp_spawn @daniellepintz
 def test_accelerator_choice_cpu(tmpdir):
     trainer = Trainer(default_root_dir=tmpdir, fast_dev_run=True)
     assert isinstance(trainer.accelerator, CPUAccelerator)

tests/accelerators/test_ipu.py

@@ -101,10 +101,7 @@ def test_epoch_end(self, outputs) -> None:
 @mock.patch("pytorch_lightning.accelerators.ipu.IPUAccelerator.is_available", return_value=True)
 def test_fail_if_no_ipus(mock_ipu_acc_avail, tmpdir):
     with pytest.raises(MisconfigurationException, match="IPU Accelerator requires IPU devices to run"):
-        Trainer(default_root_dir=tmpdir, ipus=1)
-
-    with pytest.raises(MisconfigurationException, match="IPU Accelerator requires IPU devices to run"):
-        Trainer(default_root_dir=tmpdir, ipus=1, accelerator="ipu")
+        Trainer(default_root_dir=tmpdir, accelerator="ipu", devices=1)
 
 
 @RunIf(ipu=True)
@@ -398,7 +395,8 @@ def test_manual_poptorch_opts(tmpdir):
 
     trainer = Trainer(
         default_root_dir=tmpdir,
-        ipus=2,
+        accelerator="ipu",
+        devices=2,
         fast_dev_run=True,
         strategy=IPUStrategy(inference_opts=inference_opts, training_opts=training_opts),
     )
@@ -552,13 +550,13 @@ def test_precision_plugin(tmpdir):
 
 @RunIf(ipu=True)
 def test_accelerator_ipu():
-    trainer = Trainer(accelerator="ipu", ipus=1)
+    trainer = Trainer(accelerator="ipu", devices=1)
     assert isinstance(trainer.accelerator, IPUAccelerator)
 
     trainer = Trainer(accelerator="ipu")
     assert isinstance(trainer.accelerator, IPUAccelerator)
 
-    trainer = Trainer(accelerator="auto", ipus=8)
+    trainer = Trainer(accelerator="auto", devices=8)
     assert isinstance(trainer.accelerator, IPUAccelerator)
 
 
@@ -592,8 +590,8 @@ def test_accelerator_ipu_with_ipus_priority():
 
 @RunIf(ipu=True)
 def test_set_devices_if_none_ipu():
-
-    trainer = Trainer(accelerator="ipu", ipus=8)
+    with pytest.deprecated_call(match=r"is deprecated in v1.6 and will be removed in v2.0."):
+        trainer = Trainer(accelerator="ipu", ipus=8)
     assert trainer.num_devices == 8
 
 

tests/accelerators/test_tpu.py

@@ -120,7 +120,8 @@ def test_accelerator_tpu_with_tpu_cores_priority():
 @RunIf(tpu=True)
 @pl_multi_process_test
 def test_set_devices_if_none_tpu():
-    trainer = Trainer(accelerator="tpu", tpu_cores=8)
+    with pytest.deprecated_call(match=r"is deprecated in v1.6 and will be removed in v2.0."):
+        trainer = Trainer(accelerator="tpu", tpu_cores=8)
     assert trainer.num_devices == 8
 
 

tests/benchmarks/test_basic_parity.py

@@ -159,7 +159,8 @@ def lightning_loop(cls_model, idx, device_type: str = "cuda", num_epochs=10):
         enable_progress_bar=False,
         enable_model_summary=False,
         enable_checkpointing=False,
-        gpus=1 if device_type == "cuda" else 0,
+        accelerator="gpu" if device_type == "cuda" else "cpu",
+        devices=1,
         logger=False,
         replace_sampler_ddp=False,
     )

tests/benchmarks/test_sharded_parity.py

@@ -137,15 +137,24 @@ def plugin_parity_test(
     ddp_model = model_cls()
     use_cuda = gpus > 0
 
-    trainer = Trainer(fast_dev_run=True, max_epochs=1, gpus=gpus, precision=precision, strategy="ddp_spawn")
+    trainer = Trainer(
+        fast_dev_run=True, max_epochs=1, accelerator="gpu", devices=gpus, precision=precision, strategy="ddp_spawn"
+    )
 
     max_memory_ddp, ddp_time = record_ddp_fit_model_stats(trainer=trainer, model=ddp_model, use_cuda=use_cuda)
 
     # Reset and train Custom DDP
     seed_everything(seed)
     custom_plugin_model = model_cls()
 
-    trainer = Trainer(fast_dev_run=True, max_epochs=1, gpus=gpus, precision=precision, strategy="ddp_sharded_spawn")
+    trainer = Trainer(
+        fast_dev_run=True,
+        max_epochs=1,
+        accelerator="gpu",
+        devices=gpus,
+        precision=precision,
+        strategy="ddp_sharded_spawn",
+    )
     assert isinstance(trainer.strategy, DDPSpawnShardedStrategy)
 
     max_memory_custom, custom_model_time = record_ddp_fit_model_stats(

tests/deprecated_api/test_remove_2-0.py

@@ -11,12 +11,39 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-"""Test deprecated functionality which will be removed in v2.0."""
+"""Test deprecated functionality which will be removed in v2.0.0."""
+from unittest import mock
+
 import pytest
 
 from pytorch_lightning import Trainer
 from tests.callbacks.test_callbacks import OldStatefulCallback
 from tests.helpers import BoringModel
+from tests.helpers.runif import RunIf
+
+
+def test_v2_0_0_deprecated_num_processes(tmpdir):
+    with pytest.deprecated_call(match=r"is deprecated in v1.6 and will be removed in v2.0."):
+        _ = Trainer(default_root_dir=tmpdir, num_processes=2)
+
+
+@mock.patch("torch.cuda.is_available", return_value=True)
+@mock.patch("torch.cuda.device_count", return_value=2)
+def test_v2_0_0_deprecated_gpus(mock_is_available, mock_device_count, tmpdir):
+    with pytest.deprecated_call(match=r"is deprecated in v1.6 and will be removed in v2.0."):
+        _ = Trainer(default_root_dir=tmpdir, gpus=0)
+
+
+@RunIf(tpu=True)
+def test_v2_0_0_deprecated_tpu_cores(tmpdir):
+    with pytest.deprecated_call(match=r"is deprecated in v1.6 and will be removed in v2.0."):
+        _ = Trainer(default_root_dir=tmpdir, tpu_cores=1)
+
+
+@RunIf(ipu=True)
+def test_v2_0_0_deprecated_ipus(tmpdir):
+    with pytest.deprecated_call(match=r"is deprecated in v1.6 and will be removed in v2.0."):
+        _ = Trainer(default_root_dir=tmpdir, ipus=4)
 
 
 def test_v2_0_resume_from_checkpoint_trainer_constructor(tmpdir):

tests/models/test_gpu.py

@@ -191,7 +191,8 @@ def test_parse_gpu_returns_none_when_no_devices_are_available(mocked_device_coun
 def test_torchelastic_gpu_parsing(mocked_device_count, mocked_is_available, gpus):
     """Ensure when using torchelastic and nproc_per_node is set to the default of 1 per GPU device That we omit
     sanitizing the gpus as only one of the GPUs is visible."""
-    trainer = Trainer(gpus=gpus)
+    with pytest.deprecated_call(match=r"is deprecated in v1.6 and will be removed in v2.0."):
+        trainer = Trainer(gpus=gpus)
     assert isinstance(trainer._accelerator_connector.cluster_environment, TorchElasticEnvironment)
     # when use gpu
     if device_parser.parse_gpu_ids(gpus) is not None:

tests/models/test_hooks.py

@@ -438,10 +438,13 @@ def _predict_batch(trainer, model, batches):
     [
         {},
         # these precision plugins modify the optimization flow, so testing them explicitly
-        pytest.param(dict(gpus=1, precision=16, amp_backend="native"), marks=RunIf(min_gpus=1)),
-        pytest.param(dict(gpus=1, precision=16, amp_backend="apex"), marks=RunIf(min_gpus=1, amp_apex=True)),
+        pytest.param(dict(accelerator="gpu", devices=1, precision=16, amp_backend="native"), marks=RunIf(min_gpus=1)),
         pytest.param(
-            dict(gpus=1, precision=16, strategy="deepspeed"), marks=RunIf(min_gpus=1, standalone=True, deepspeed=True)
+            dict(accelerator="gpu", devices=1, precision=16, amp_backend="apex"), marks=RunIf(min_gpus=1, amp_apex=True)
+        ),
+        pytest.param(
+            dict(accelerator="gpu", devices=1, precision=16, strategy="deepspeed"),
+            marks=RunIf(min_gpus=1, standalone=True, deepspeed=True),
         ),
     ],
 )
@@ -496,7 +499,7 @@ def training_step(self, batch, batch_idx):
     }
     if kwargs.get("amp_backend") == "native" or kwargs.get("amp_backend") == "apex":
         saved_ckpt[trainer.precision_plugin.__class__.__qualname__] = ANY
-    device = torch.device("cuda:0" if "gpus" in kwargs else "cpu")
+    device = torch.device("cuda:0" if "accelerator" in kwargs and kwargs["accelerator"] == "gpu" else "cpu")
     expected = [
         dict(name="Callback.on_init_start", args=(trainer,)),
         dict(name="Callback.on_init_end", args=(trainer,)),

tests/models/test_tpu.py

@@ -305,7 +305,8 @@ def test_tpu_cores_with_argparse(cli_args, expected):
 
     for k, v in expected.items():
         assert getattr(args, k) == v
-    assert Trainer.from_argparse_args(args)
+    with pytest.deprecated_call(match=r"is deprecated in v1.6 and will be removed in v2.0."):
+        assert Trainer.from_argparse_args(args)
 
 
 @RunIf(tpu=True)

tests/strategies/test_ddp.py

@@ -88,7 +88,14 @@ def test_torch_distributed_backend_env_variables(tmpdir):
         with pytest.deprecated_call(match="Environment variable `PL_TORCH_DISTRIBUTED_BACKEND` was deprecated in v1.6"):
             with pytest.raises(ValueError, match="Invalid backend: 'undefined'"):
                 model = BoringModel()
-                trainer = Trainer(default_root_dir=tmpdir, fast_dev_run=True, strategy="ddp", gpus=2, logger=False)
+                trainer = Trainer(
+                    default_root_dir=tmpdir,
+                    fast_dev_run=True,
+                    strategy="ddp",
+                    accelerator="gpu",
+                    devices=2,
+                    logger=False,
+                )
                 trainer.fit(model)
 
 

tests/strategies/test_ddp_strategy_with_comm_hook.py

@@ -179,7 +179,8 @@ def test_post_local_sgd_model_averaging(average_parameters_mock, tmpdir):
 
     trainer = Trainer(
         fast_dev_run=True,
-        gpus=2,
+        accelerator="gpu",
+        devices=2,
         strategy=ddp_strategy,
         default_root_dir=tmpdir,
         sync_batchnorm=True,

tests/strategies/test_deepspeed_strategy.py

@@ -265,7 +265,8 @@ def setup(self, trainer, pl_module, stage: Optional[str] = None) -> None:
         default_root_dir=tmpdir,
         fast_dev_run=True,
         callbacks=ck,
-        gpus=1,
+        accelerator="gpu",
+        devices=1,
         strategy=DeepSpeedStrategy(logging_batch_size_per_gpu=value, zero_optimization=False),
     )
     with pytest.raises(SystemExit):
@@ -438,7 +439,8 @@ def setup(self, trainer, pl_module, stage=None) -> None:
         max_epochs=1,
         strategy=DeepSpeedStrategy(config=deepspeed_zero_config),
         precision=16,
-        gpus=1,
+        accelerator="gpu",
+        devices=1,
         callbacks=[TestCallback()],
     )
     with pytest.raises(SystemExit):
@@ -688,7 +690,9 @@ def test_deepspeed_multigpu_stage_3_checkpointing(tmpdir, automatic_optimization
         model = ModelParallelClassificationModel()
     else:
         model = ManualModelParallelClassificationModel()
-    trainer = Trainer(default_root_dir=tmpdir, gpus=2, strategy=DeepSpeedStrategy(stage=3), precision=16)
+    trainer = Trainer(
+        default_root_dir=tmpdir, accelerator="gpu", devices=2, strategy=DeepSpeedStrategy(stage=3), precision=16
+    )
 
     results = trainer.test(model, datamodule=dm, ckpt_path=ck.best_model_path)
     assert results[0]["test_acc"] > 0.7

tests/trainer/properties/test_auto_gpu_select.py

@@ -22,7 +22,6 @@
 from tests.helpers.runif import RunIf
 
 
-# TODO: add pytest.deprecated_call @daniellepintz
 @RunIf(min_gpus=2)
 @pytest.mark.parametrize(
     ["auto_select_gpus", "gpus", "expected_error"],
@@ -36,10 +35,11 @@ def test_trainer_with_gpus_options_combination_at_available_gpus_env(auto_select
                 "auto_select_gpus=True, gpus=0 is not a valid configuration."
                 " Please select a valid number of GPU resources when using auto_select_gpus."
             ),
-        ):
+        ), pytest.deprecated_call(match=r"is deprecated in v1.6 and will be removed in v2.0."):
             Trainer(auto_select_gpus=auto_select_gpus, gpus=gpus)
     else:
-        Trainer(auto_select_gpus=auto_select_gpus, gpus=gpus)
+        with pytest.deprecated_call(match=r"is deprecated in v1.6 and will be removed in v2.0."):
+            Trainer(auto_select_gpus=auto_select_gpus, gpus=gpus)
 
 
 @RunIf(min_gpus=2)

tests/trainer/test_supporters.py

@@ -348,8 +348,8 @@ def __init__(self, data_source, name) -> None:
     )
 
     with mock.patch.dict(os.environ, {"PL_FAULT_TOLERANT_TRAINING": str(int(use_fault_tolerant))}):
-
-        trainer = Trainer(replace_sampler_ddp=replace_sampler_ddp, strategy="ddp", accelerator="gpu", devices=2)
+        with pytest.deprecated_call(match=r"is deprecated in v1.6 and will be removed in v2.0."):
+            trainer = Trainer(replace_sampler_ddp=replace_sampler_ddp, strategy="ddp", accelerator="gpu", devices=2)
         dataloader = trainer._data_connector._prepare_dataloader(dataloader, shuffle=True)
         _count = 0
         _has_fastforward_sampler = False