From 0a6237d297c892e03243b00be1ee607c55313bc7 Mon Sep 17 00:00:00 2001
From: DuYicong515 <duyicong515@gmail.com>
Date: Wed, 23 Feb 2022 12:06:47 -0800
Subject: [PATCH 01/13] remove Accelerator.parallel_device_ids and deprecate
 Trainer.data_parallel_device_ids

---
 .../trainer/connectors/accelerator_connector.py      |  4 ----
 .../connectors/logger_connector/logger_connector.py  |  8 +++++++-
 pytorch_lightning/trainer/trainer.py                 | 12 ++++++++----
 tests/models/test_gpu.py                             |  7 ++++++-
 tests/trainer/test_trainer_cli.py                    |  3 ++-
 tests/utilities/test_cli.py                          |  3 ++-
 6 files changed, 25 insertions(+), 12 deletions(-)

diff --git a/pytorch_lightning/trainer/connectors/accelerator_connector.py b/pytorch_lightning/trainer/connectors/accelerator_connector.py
index aa9b25c64e28c..a28f7cfd8939c 100644
--- a/pytorch_lightning/trainer/connectors/accelerator_connector.py
+++ b/pytorch_lightning/trainer/connectors/accelerator_connector.py
@@ -819,10 +819,6 @@ def num_ipus(self) -> int:
     def gpus(self) -> Optional[Union[List[int], str, int]]:
         return self._gpus
 
-    @property
-    def parallel_device_ids(self) -> List[int]:
-        return [i for i in range(len(self.parallel_devices))] if isinstance(self.accelerator, GPUAccelerator) else []
-
     @property
     def is_distributed(self) -> bool:
         # Used for custom plugins.
diff --git a/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py b/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
index 0de5a5f882925..8800df116b963 100644
--- a/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
+++ b/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
@@ -221,7 +221,13 @@ def _log_gpus_metrics(self) -> None:
                 self.trainer.lightning_module.log(key, mem, prog_bar=False, logger=True)
             else:
                 gpu_id = int(key.split("/")[0].split(":")[1])
-                if gpu_id in self.trainer._accelerator_connector.parallel_device_ids:
+                parallel_device_ids = []
+                if isinstance(self.trainer.accelerator, GPUAccelerator):
+                    if isinstance(self.trainer.strategy, ParallelStrategy):
+                        parallel_device_ids = [i for i in range(len(self.trainer.strategy.parallel_devices))]
+                    elif isinstance(self.strategy, SingleDeviceStrategy):
+                        parallel_device_ids = [0]
+                if gpu_id in parallel_device_ids:
                     self.trainer.lightning_module.log(
                         key, mem, prog_bar=False, logger=True, on_step=True, on_epoch=False
                     )
diff --git a/pytorch_lightning/trainer/trainer.py b/pytorch_lightning/trainer/trainer.py
index f718e0f64ee43..da9af88824c48 100644
--- a/pytorch_lightning/trainer/trainer.py
+++ b/pytorch_lightning/trainer/trainer.py
@@ -58,7 +58,7 @@
     SimpleProfiler,
     XLAProfiler,
 )
-from pytorch_lightning.strategies import ParallelStrategy, Strategy
+from pytorch_lightning.strategies import ParallelStrategy, SingleDeviceStrategy, Strategy
 from pytorch_lightning.strategies.ddp_spawn import DDPSpawnStrategy
 from pytorch_lightning.trainer.callback_hook import TrainerCallbackHookMixin
 from pytorch_lightning.trainer.configuration_validator import verify_loop_configurations
@@ -2087,9 +2087,13 @@ def devices(self) -> int:
 
     @property
     def data_parallel_device_ids(self) -> Optional[List[int]]:
-        return (
-            self._accelerator_connector.parallel_device_ids if self._accelerator_connector.parallel_device_ids else None
-        )
+        rank_zero_deprecation("`Trainer.data_parallel_device_ids` was deprecated in v1.6.")
+        if isinstance(self.accelerator, GPUAccelerator):
+            if isinstance(self.strategy, ParallelStrategy):
+                return [i for i in range(len(self.strategy.parallel_devices))]
+            elif isinstance(self.strategy, SingleDeviceStrategy):
+                return [0]
+        return None
 
     @property
     def lightning_module(self) -> "pl.LightningModule":
diff --git a/tests/models/test_gpu.py b/tests/models/test_gpu.py
index 719cfb43024b1..511322d9c0698 100644
--- a/tests/models/test_gpu.py
+++ b/tests/models/test_gpu.py
@@ -27,6 +27,7 @@
 from pytorch_lightning.utilities import device_parser
 from pytorch_lightning.utilities.exceptions import MisconfigurationException
 from pytorch_lightning.utilities.imports import _compare_version, _TORCHTEXT_LEGACY
+from pytorch_lightning.utilities.rank_zero import rank_zero_only
 from tests.helpers import BoringModel
 from tests.helpers.datamodules import ClassifDataModule
 from tests.helpers.imports import Batch, Dataset, Example, Field, LabelField
@@ -190,8 +191,12 @@ def test_torchelastic_gpu_parsing(mocked_device_count, mocked_is_available, gpus
     sanitizing the gpus as only one of the GPUs is visible."""
     trainer = Trainer(gpus=gpus)
     assert isinstance(trainer._accelerator_connector.cluster_environment, TorchElasticEnvironment)
-    assert trainer.data_parallel_device_ids == device_parser.parse_gpu_ids(gpus)
     assert trainer.gpus == gpus
+    if rank_zero_only.rank == 0:
+        with pytest.deprecated_call(match="`Trainer.data_parallel_device_ids` was deprecated in v1.6."):
+            assert trainer.data_parallel_device_ids == device_parser.parse_gpu_ids(gpus)
+    else:
+        assert trainer.data_parallel_device_ids == device_parser.parse_gpu_ids(gpus)
 
 
 @RunIf(min_gpus=1)
diff --git a/tests/trainer/test_trainer_cli.py b/tests/trainer/test_trainer_cli.py
index 11b904870294b..fdaea98e7727d 100644
--- a/tests/trainer/test_trainer_cli.py
+++ b/tests/trainer/test_trainer_cli.py
@@ -178,7 +178,8 @@ def test_argparse_args_parsing_devices(cli_args, expected_parsed, expected_devic
 
     assert args.devices == expected_parsed
     trainer = Trainer.from_argparse_args(args)
-    assert trainer.data_parallel_device_ids == expected_device_ids
+    with pytest.deprecated_call(match="`Trainer.data_parallel_device_ids` was deprecated in v1.6."):
+        assert trainer.data_parallel_device_ids == expected_device_ids
 
 
 @pytest.mark.parametrize(
diff --git a/tests/utilities/test_cli.py b/tests/utilities/test_cli.py
index 5da16737fc2d7..485836e9f7c32 100644
--- a/tests/utilities/test_cli.py
+++ b/tests/utilities/test_cli.py
@@ -192,7 +192,8 @@ def test_parse_args_parsing_gpus(monkeypatch, cli_args, expected_gpu):
         args = parser.parse_args()
 
     trainer = Trainer.from_argparse_args(args)
-    assert trainer.data_parallel_device_ids == expected_gpu
+    with pytest.deprecated_call(match="`Trainer.data_parallel_device_ids` was deprecated in v1.6."):
+        assert trainer.data_parallel_device_ids == expected_gpu
 
 
 @pytest.mark.skipif(

From 71c5e0150fd4cfdde23e30d65ce248d6d80daa34 Mon Sep 17 00:00:00 2001
From: DuYicong515 <duyicong515@gmail.com>
Date: Wed, 23 Feb 2022 14:41:03 -0800
Subject: [PATCH 02/13] rebase

---
 CHANGELOG.md | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 541b3dcc1d261..192bd5c517fd5 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -533,6 +533,9 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Deprecated `Trainer.num_gpus` in favor of `Trainer.num_devices` when GPU is used ([#12384](https://github.com/PyTorchLightning/pytorch-lightning/pull/12384))
 
 
+- Deprecated `Trainer.data_parallel_device_ids` in favor of `Trainer.device_ids` ([#12072](https://github.com/PyTorchLightning/pytorch-lightning/pull/12072))
+
+
 ### Removed
 
 - Removed deprecated parameter `method` in `pytorch_lightning.utilities.model_helpers.is_overridden` ([#10507](https://github.com/PyTorchLightning/pytorch-lightning/pull/10507))
@@ -720,7 +723,11 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Removed public attribute `sync_batchnorm` from strategies ([#11754](https://github.com/PyTorchLightning/pytorch-lightning/pull/11754))
 
 
+<<<<<<< HEAD
 - Removed `AcceleratorConnector.root_gpu` property ([#12262](https://github.com/PyTorchLightning/pytorch-lightning/pull/12262))
+=======
+- Removed deprecated `AcceleratorConnector.parallel_device_ids` property ([#12072](https://github.com/PyTorchLightning/pytorch-lightning/pull/12072))
+>>>>>>> c3e41378b (rebase)
 
 
 - Removed `AcceleratorConnector.num_gpus` property ([#12384](https://github.com/PyTorchLightning/pytorch-lightning/pull/12384))

From 0391920b273f10ca05776c4b359e0bcc9b86aa5f Mon Sep 17 00:00:00 2001
From: DuYicong515 <duyicong515@gmail.com>
Date: Wed, 23 Feb 2022 12:52:24 -0800
Subject: [PATCH 03/13] fix pre-commit

---
 .../trainer/connectors/logger_connector/logger_connector.py      | 1 +
 1 file changed, 1 insertion(+)

diff --git a/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py b/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
index 8800df116b963..b8d80be94f2df 100644
--- a/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
+++ b/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
@@ -19,6 +19,7 @@
 from pytorch_lightning.accelerators import GPUAccelerator
 from pytorch_lightning.loggers import LightningLoggerBase, TensorBoardLogger
 from pytorch_lightning.plugins.environments.slurm_environment import SLURMEnvironment
+from pytorch_lightning.strategies import ParallelStrategy, SingleDeviceStrategy
 from pytorch_lightning.trainer.connectors.logger_connector.result import _METRICS, _OUT_DICT, _PBAR_DICT
 from pytorch_lightning.trainer.states import RunningStage
 from pytorch_lightning.utilities import memory

From c1c527b3467f5d1a6b15668697681076f1aadc90 Mon Sep 17 00:00:00 2001
From: DuYicong515 <duyicong515@gmail.com>
Date: Wed, 23 Feb 2022 13:02:59 -0800
Subject: [PATCH 04/13] use trainer.data_parallel_device_ids in
 logger_connector

---
 .../connectors/logger_connector/logger_connector.py      | 9 +--------
 1 file changed, 1 insertion(+), 8 deletions(-)

diff --git a/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py b/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
index b8d80be94f2df..59b5f36564c03 100644
--- a/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
+++ b/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
@@ -19,7 +19,6 @@
 from pytorch_lightning.accelerators import GPUAccelerator
 from pytorch_lightning.loggers import LightningLoggerBase, TensorBoardLogger
 from pytorch_lightning.plugins.environments.slurm_environment import SLURMEnvironment
-from pytorch_lightning.strategies import ParallelStrategy, SingleDeviceStrategy
 from pytorch_lightning.trainer.connectors.logger_connector.result import _METRICS, _OUT_DICT, _PBAR_DICT
 from pytorch_lightning.trainer.states import RunningStage
 from pytorch_lightning.utilities import memory
@@ -222,13 +221,7 @@ def _log_gpus_metrics(self) -> None:
                 self.trainer.lightning_module.log(key, mem, prog_bar=False, logger=True)
             else:
                 gpu_id = int(key.split("/")[0].split(":")[1])
-                parallel_device_ids = []
-                if isinstance(self.trainer.accelerator, GPUAccelerator):
-                    if isinstance(self.trainer.strategy, ParallelStrategy):
-                        parallel_device_ids = [i for i in range(len(self.trainer.strategy.parallel_devices))]
-                    elif isinstance(self.strategy, SingleDeviceStrategy):
-                        parallel_device_ids = [0]
-                if gpu_id in parallel_device_ids:
+                if gpu_id in self.trainer.data_parallel_device_ids:
                     self.trainer.lightning_module.log(
                         key, mem, prog_bar=False, logger=True, on_step=True, on_epoch=False
                     )

From f47990a0cac67210c39ce5005c0f66d68d103356 Mon Sep 17 00:00:00 2001
From: DuYicong515 <duyicong515@gmail.com>
Date: Wed, 23 Feb 2022 13:20:52 -0800
Subject: [PATCH 05/13] fix tests

---
 .../connectors/logger_connector/logger_connector.py        | 2 +-
 tests/callbacks/test_gpu_stats_monitor.py                  | 7 +++++--
 2 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py b/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
index 59b5f36564c03..69b5b158a4b4f 100644
--- a/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
+++ b/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
@@ -221,7 +221,7 @@ def _log_gpus_metrics(self) -> None:
                 self.trainer.lightning_module.log(key, mem, prog_bar=False, logger=True)
             else:
                 gpu_id = int(key.split("/")[0].split(":")[1])
-                if gpu_id in self.trainer.data_parallel_device_ids:
+                if self.trainer.data_parallel_device_ids and gpu_id in self.trainer.data_parallel_device_ids:
                     self.trainer.lightning_module.log(
                         key, mem, prog_bar=False, logger=True, on_step=True, on_epoch=False
                     )
diff --git a/tests/callbacks/test_gpu_stats_monitor.py b/tests/callbacks/test_gpu_stats_monitor.py
index 1e3a9953bf137..5af24000e06ed 100644
--- a/tests/callbacks/test_gpu_stats_monitor.py
+++ b/tests/callbacks/test_gpu_stats_monitor.py
@@ -47,7 +47,8 @@ def test_gpu_stats_monitor(tmpdir):
         logger=logger,
     )
 
-    trainer.fit(model)
+    with pytest.deprecated_call(match="`Trainer.data_parallel_device_ids` was deprecated in v1.6."):
+        trainer.fit(model)
     assert trainer.state.finished, f"Training failed with {trainer.state}"
 
     path_csv = os.path.join(logger.log_dir, ExperimentWriter.NAME_METRICS_FILE)
@@ -84,7 +85,9 @@ def test_gpu_stats_monitor_no_queries(tmpdir):
         devices=1,
         callbacks=[gpu_stats],
     )
-    with mock.patch("pytorch_lightning.loggers.tensorboard.TensorBoardLogger.log_metrics") as log_metrics_mock:
+    with mock.patch(
+        "pytorch_lightning.loggers.tensorboard.TensorBoardLogger.log_metrics"
+    ) as log_metrics_mock, pytest.deprecated_call(match="`Trainer.data_parallel_device_ids` was deprecated in v1.6."):
         trainer.fit(model)
 
     assert log_metrics_mock.mock_calls[1:] == [

From e82870ec4e806b46126c913a7abf7156b903e5d6 Mon Sep 17 00:00:00 2001
From: DuYicong515 <duyicong515@gmail.com>
Date: Thu, 24 Feb 2022 12:00:15 -0800
Subject: [PATCH 06/13] address comments

---
 CHANGELOG.md                                              | 2 ++
 pytorch_lightning/callbacks/gpu_stats_monitor.py          | 3 +--
 .../connectors/logger_connector/logger_connector.py       | 2 +-
 pytorch_lightning/trainer/trainer.py                      | 8 ++++----
 tests/models/test_gpu.py                                  | 4 ++--
 tests/trainer/test_trainer_cli.py                         | 2 +-
 6 files changed, 11 insertions(+), 10 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 192bd5c517fd5..845afac58d71b 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -368,6 +368,8 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - The strategies that support `sync_batchnorm` now only apply it when fitting ([#11919](https://github.com/PyTorchLightning/pytorch-lightning/pull/11919))
 
 
+- Changed `Trainer.data_parallel_device_ids` to return a list of GPU device indexes ([#12072](https://github.com/PyTorchLightning/pytorch-lightning/pull/12072))
+
 
 ### Deprecated
 
diff --git a/pytorch_lightning/callbacks/gpu_stats_monitor.py b/pytorch_lightning/callbacks/gpu_stats_monitor.py
index 8fb92006708f7..78fe2f70a8e1a 100644
--- a/pytorch_lightning/callbacks/gpu_stats_monitor.py
+++ b/pytorch_lightning/callbacks/gpu_stats_monitor.py
@@ -133,8 +133,7 @@ def setup(self, trainer: "pl.Trainer", pl_module: "pl.LightningModule", stage: O
             )
 
         # The logical device IDs for selected devices
-        # ignoring mypy check because `trainer.data_parallel_device_ids` is None when using CPU
-        self._device_ids = sorted(set(trainer.data_parallel_device_ids))  # type: ignore
+        self._device_ids = sorted(set(trainer.data_parallel_device_ids))
 
         # The unmasked real GPU IDs
         self._gpu_ids = self._get_gpu_ids(self._device_ids)
diff --git a/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py b/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
index 69b5b158a4b4f..59b5f36564c03 100644
--- a/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
+++ b/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
@@ -221,7 +221,7 @@ def _log_gpus_metrics(self) -> None:
                 self.trainer.lightning_module.log(key, mem, prog_bar=False, logger=True)
             else:
                 gpu_id = int(key.split("/")[0].split(":")[1])
-                if self.trainer.data_parallel_device_ids and gpu_id in self.trainer.data_parallel_device_ids:
+                if gpu_id in self.trainer.data_parallel_device_ids:
                     self.trainer.lightning_module.log(
                         key, mem, prog_bar=False, logger=True, on_step=True, on_epoch=False
                     )
diff --git a/pytorch_lightning/trainer/trainer.py b/pytorch_lightning/trainer/trainer.py
index da9af88824c48..92024ccc29167 100644
--- a/pytorch_lightning/trainer/trainer.py
+++ b/pytorch_lightning/trainer/trainer.py
@@ -2086,14 +2086,14 @@ def devices(self) -> int:
         return self.num_devices
 
     @property
-    def data_parallel_device_ids(self) -> Optional[List[int]]:
-        rank_zero_deprecation("`Trainer.data_parallel_device_ids` was deprecated in v1.6.")
+    def data_parallel_device_ids(self) -> List[int]:
+        rank_zero_deprecation("`Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8.")
         if isinstance(self.accelerator, GPUAccelerator):
             if isinstance(self.strategy, ParallelStrategy):
-                return [i for i in range(len(self.strategy.parallel_devices))]
+                return [device.index for device in self.strategy.parallel_devices]
             elif isinstance(self.strategy, SingleDeviceStrategy):
                 return [0]
-        return None
+        return []
 
     @property
     def lightning_module(self) -> "pl.LightningModule":
diff --git a/tests/models/test_gpu.py b/tests/models/test_gpu.py
index 511322d9c0698..6252b601b2870 100644
--- a/tests/models/test_gpu.py
+++ b/tests/models/test_gpu.py
@@ -185,7 +185,7 @@ def test_parse_gpu_returns_none_when_no_devices_are_available(mocked_device_coun
 )
 @mock.patch("torch.cuda.device_count", return_value=1)
 @mock.patch("torch.cuda.is_available", return_value=True)
-@pytest.mark.parametrize("gpus", [[0, 1, 2], 2, "0"])
+@pytest.mark.parametrize("gpus", [[0, 1, 2], 2, "0", [0, 2]])
 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."""
@@ -194,7 +194,7 @@ def test_torchelastic_gpu_parsing(mocked_device_count, mocked_is_available, gpus
     assert trainer.gpus == gpus
     if rank_zero_only.rank == 0:
         with pytest.deprecated_call(match="`Trainer.data_parallel_device_ids` was deprecated in v1.6."):
-            assert trainer.data_parallel_device_ids == device_parser.parse_gpu_ids(gpus)
+            assert (trainer.data_parallel_device_ids or None) == device_parser.parse_gpu_ids(gpus)
     else:
         assert trainer.data_parallel_device_ids == device_parser.parse_gpu_ids(gpus)
 
diff --git a/tests/trainer/test_trainer_cli.py b/tests/trainer/test_trainer_cli.py
index fdaea98e7727d..b84fb4424dcb7 100644
--- a/tests/trainer/test_trainer_cli.py
+++ b/tests/trainer/test_trainer_cli.py
@@ -162,7 +162,7 @@ def test_argparse_args_parsing_fast_dev_run(cli_args, expected):
 
 @pytest.mark.parametrize(
     ["cli_args", "expected_parsed", "expected_device_ids"],
-    [("", None, None), ("--accelerator gpu --devices 1", "1", [0]), ("--accelerator gpu --devices 0,", "0,", [0])],
+    [("", None, []), ("--accelerator gpu --devices 1", "1", [0]), ("--accelerator gpu --devices 0,", "0,", [])],
 )
 def test_argparse_args_parsing_devices(cli_args, expected_parsed, expected_device_ids, monkeypatch):
     """Test multi type argument with bool."""

From ac12427693df82b6ce0cfc1d9c3e076acba62cbd Mon Sep 17 00:00:00 2001
From: DuYicong515 <duyicong515@gmail.com>
Date: Thu, 24 Feb 2022 17:56:59 -0800
Subject: [PATCH 07/13] update changelog

---
 CHANGELOG.md | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 845afac58d71b..3d28e20402a3a 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -725,16 +725,15 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Removed public attribute `sync_batchnorm` from strategies ([#11754](https://github.com/PyTorchLightning/pytorch-lightning/pull/11754))
 
 
-<<<<<<< HEAD
 - Removed `AcceleratorConnector.root_gpu` property ([#12262](https://github.com/PyTorchLightning/pytorch-lightning/pull/12262))
-=======
-- Removed deprecated `AcceleratorConnector.parallel_device_ids` property ([#12072](https://github.com/PyTorchLightning/pytorch-lightning/pull/12072))
->>>>>>> c3e41378b (rebase)
 
 
 - Removed `AcceleratorConnector.num_gpus` property ([#12384](https://github.com/PyTorchLightning/pytorch-lightning/pull/12384))
 
 
+- Removed `AcceleratorConnector.parallel_device_ids` property ([#12072](https://github.com/PyTorchLightning/pytorch-lightning/pull/12072))
+
+
 ### Fixed
 
 - Fixed an issue where `ModelCheckpoint` could delete older checkpoints when `dirpath` has changed during resumed training ([#12045](https://github.com/PyTorchLightning/pytorch-lightning/pull/12045))

From 09613094148eee42c5d9d3868d3c728b25159f43 Mon Sep 17 00:00:00 2001
From: DuYicong515 <duyicong515@gmail.com>
Date: Thu, 24 Feb 2022 21:06:49 -0800
Subject: [PATCH 08/13] rebase

---
 tests/callbacks/test_gpu_stats_monitor.py | 8 ++++++--
 tests/models/test_gpu.py                  | 6 ++++--
 tests/trainer/test_trainer_cli.py         | 4 +++-
 tests/utilities/test_cli.py               | 4 +++-
 4 files changed, 16 insertions(+), 6 deletions(-)

diff --git a/tests/callbacks/test_gpu_stats_monitor.py b/tests/callbacks/test_gpu_stats_monitor.py
index 5af24000e06ed..a905b9da48ff0 100644
--- a/tests/callbacks/test_gpu_stats_monitor.py
+++ b/tests/callbacks/test_gpu_stats_monitor.py
@@ -47,7 +47,9 @@ def test_gpu_stats_monitor(tmpdir):
         logger=logger,
     )
 
-    with pytest.deprecated_call(match="`Trainer.data_parallel_device_ids` was deprecated in v1.6."):
+    with pytest.deprecated_call(
+        match="Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8."
+    ):
         trainer.fit(model)
     assert trainer.state.finished, f"Training failed with {trainer.state}"
 
@@ -87,7 +89,9 @@ def test_gpu_stats_monitor_no_queries(tmpdir):
     )
     with mock.patch(
         "pytorch_lightning.loggers.tensorboard.TensorBoardLogger.log_metrics"
-    ) as log_metrics_mock, pytest.deprecated_call(match="`Trainer.data_parallel_device_ids` was deprecated in v1.6."):
+    ) as log_metrics_mock, pytest.deprecated_call(
+        match="Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8."
+    ):
         trainer.fit(model)
 
     assert log_metrics_mock.mock_calls[1:] == [
diff --git a/tests/models/test_gpu.py b/tests/models/test_gpu.py
index 6252b601b2870..74535d7a569a2 100644
--- a/tests/models/test_gpu.py
+++ b/tests/models/test_gpu.py
@@ -193,10 +193,12 @@ def test_torchelastic_gpu_parsing(mocked_device_count, mocked_is_available, gpus
     assert isinstance(trainer._accelerator_connector.cluster_environment, TorchElasticEnvironment)
     assert trainer.gpus == gpus
     if rank_zero_only.rank == 0:
-        with pytest.deprecated_call(match="`Trainer.data_parallel_device_ids` was deprecated in v1.6."):
+        with pytest.deprecated_call(
+            match="Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8."
+        ):
             assert (trainer.data_parallel_device_ids or None) == device_parser.parse_gpu_ids(gpus)
     else:
-        assert trainer.data_parallel_device_ids == device_parser.parse_gpu_ids(gpus)
+        assert (trainer.data_parallel_device_ids or None) == device_parser.parse_gpu_ids(gpus)
 
 
 @RunIf(min_gpus=1)
diff --git a/tests/trainer/test_trainer_cli.py b/tests/trainer/test_trainer_cli.py
index b84fb4424dcb7..00f223c689b3e 100644
--- a/tests/trainer/test_trainer_cli.py
+++ b/tests/trainer/test_trainer_cli.py
@@ -178,7 +178,9 @@ def test_argparse_args_parsing_devices(cli_args, expected_parsed, expected_devic
 
     assert args.devices == expected_parsed
     trainer = Trainer.from_argparse_args(args)
-    with pytest.deprecated_call(match="`Trainer.data_parallel_device_ids` was deprecated in v1.6."):
+    with pytest.deprecated_call(
+        match="Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8."
+    ):
         assert trainer.data_parallel_device_ids == expected_device_ids
 
 
diff --git a/tests/utilities/test_cli.py b/tests/utilities/test_cli.py
index 485836e9f7c32..46bbbbb6253fb 100644
--- a/tests/utilities/test_cli.py
+++ b/tests/utilities/test_cli.py
@@ -192,7 +192,9 @@ def test_parse_args_parsing_gpus(monkeypatch, cli_args, expected_gpu):
         args = parser.parse_args()
 
     trainer = Trainer.from_argparse_args(args)
-    with pytest.deprecated_call(match="`Trainer.data_parallel_device_ids` was deprecated in v1.6."):
+    with pytest.deprecated_call(
+        match="Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8."
+    ):
         assert trainer.data_parallel_device_ids == expected_gpu
 
 

From 25cc8fc0073d340374ab7a4de5fc2832cf7df1bc Mon Sep 17 00:00:00 2001
From: DuYicong515 <duyicong515@gmail.com>
Date: Fri, 25 Feb 2022 12:08:43 -0800
Subject: [PATCH 09/13] rebase and use Trainer.device_ids

---
 .../callbacks/gpu_stats_monitor.py            |  2 +-
 .../logger_connector/logger_connector.py      |  2 +-
 pytorch_lightning/trainer/trainer.py          | 14 +++++-----
 tests/callbacks/test_gpu_stats_monitor.py     | 11 ++------
 tests/deprecated_api/test_remove_1-8.py       | 27 +++++++++++++++++++
 tests/models/test_gpu.py                      | 11 +++-----
 tests/trainer/test_trainer_cli.py             | 12 +++------
 tests/utilities/test_cli.py                   |  9 +++----
 8 files changed, 48 insertions(+), 40 deletions(-)

diff --git a/pytorch_lightning/callbacks/gpu_stats_monitor.py b/pytorch_lightning/callbacks/gpu_stats_monitor.py
index 78fe2f70a8e1a..607dc4cf0efbd 100644
--- a/pytorch_lightning/callbacks/gpu_stats_monitor.py
+++ b/pytorch_lightning/callbacks/gpu_stats_monitor.py
@@ -133,7 +133,7 @@ def setup(self, trainer: "pl.Trainer", pl_module: "pl.LightningModule", stage: O
             )
 
         # The logical device IDs for selected devices
-        self._device_ids = sorted(set(trainer.data_parallel_device_ids))
+        self._device_ids = sorted(set(trainer.device_ids))
 
         # The unmasked real GPU IDs
         self._gpu_ids = self._get_gpu_ids(self._device_ids)
diff --git a/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py b/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
index 59b5f36564c03..9cc91c5adf171 100644
--- a/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
+++ b/pytorch_lightning/trainer/connectors/logger_connector/logger_connector.py
@@ -221,7 +221,7 @@ def _log_gpus_metrics(self) -> None:
                 self.trainer.lightning_module.log(key, mem, prog_bar=False, logger=True)
             else:
                 gpu_id = int(key.split("/")[0].split(":")[1])
-                if gpu_id in self.trainer.data_parallel_device_ids:
+                if gpu_id in self.trainer.device_ids:
                     self.trainer.lightning_module.log(
                         key, mem, prog_bar=False, logger=True, on_step=True, on_epoch=False
                     )
diff --git a/pytorch_lightning/trainer/trainer.py b/pytorch_lightning/trainer/trainer.py
index 92024ccc29167..c32a5a2b2ec68 100644
--- a/pytorch_lightning/trainer/trainer.py
+++ b/pytorch_lightning/trainer/trainer.py
@@ -58,7 +58,7 @@
     SimpleProfiler,
     XLAProfiler,
 )
-from pytorch_lightning.strategies import ParallelStrategy, SingleDeviceStrategy, Strategy
+from pytorch_lightning.strategies import ParallelStrategy, Strategy
 from pytorch_lightning.strategies.ddp_spawn import DDPSpawnStrategy
 from pytorch_lightning.trainer.callback_hook import TrainerCallbackHookMixin
 from pytorch_lightning.trainer.configuration_validator import verify_loop_configurations
@@ -2087,13 +2087,11 @@ def devices(self) -> int:
 
     @property
     def data_parallel_device_ids(self) -> List[int]:
-        rank_zero_deprecation("`Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8.")
-        if isinstance(self.accelerator, GPUAccelerator):
-            if isinstance(self.strategy, ParallelStrategy):
-                return [device.index for device in self.strategy.parallel_devices]
-            elif isinstance(self.strategy, SingleDeviceStrategy):
-                return [0]
-        return []
+        rank_zero_deprecation(
+            "`Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8."
+            " Please use `Trainer.device_ids` instead."
+        )
+        return self.device_ids if isinstance(self.accelerator, GPUAccelerator) else []
 
     @property
     def lightning_module(self) -> "pl.LightningModule":
diff --git a/tests/callbacks/test_gpu_stats_monitor.py b/tests/callbacks/test_gpu_stats_monitor.py
index a905b9da48ff0..1e3a9953bf137 100644
--- a/tests/callbacks/test_gpu_stats_monitor.py
+++ b/tests/callbacks/test_gpu_stats_monitor.py
@@ -47,10 +47,7 @@ def test_gpu_stats_monitor(tmpdir):
         logger=logger,
     )
 
-    with pytest.deprecated_call(
-        match="Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8."
-    ):
-        trainer.fit(model)
+    trainer.fit(model)
     assert trainer.state.finished, f"Training failed with {trainer.state}"
 
     path_csv = os.path.join(logger.log_dir, ExperimentWriter.NAME_METRICS_FILE)
@@ -87,11 +84,7 @@ def test_gpu_stats_monitor_no_queries(tmpdir):
         devices=1,
         callbacks=[gpu_stats],
     )
-    with mock.patch(
-        "pytorch_lightning.loggers.tensorboard.TensorBoardLogger.log_metrics"
-    ) as log_metrics_mock, pytest.deprecated_call(
-        match="Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8."
-    ):
+    with mock.patch("pytorch_lightning.loggers.tensorboard.TensorBoardLogger.log_metrics") as log_metrics_mock:
         trainer.fit(model)
 
     assert log_metrics_mock.mock_calls[1:] == [
diff --git a/tests/deprecated_api/test_remove_1-8.py b/tests/deprecated_api/test_remove_1-8.py
index 035f76a748ee1..f556c241e6c82 100644
--- a/tests/deprecated_api/test_remove_1-8.py
+++ b/tests/deprecated_api/test_remove_1-8.py
@@ -962,3 +962,30 @@ def test_trainer_num_gpu_0(monkeypatch, gpus, expected_num_gpus, strategy):
         " Please use `Trainer.num_devices` instead."
     ):
         assert Trainer(gpus=gpus, strategy=strategy).num_gpus == expected_num_gpus
+
+
+@pytest.mark.parametrize(
+    ["trainer_kwargs", "expected_data_parallel_device_ids"],
+    [
+        ({}, []),
+        ({"devices": 1}, []),
+        ({"devices": "1"}, []),
+        ({"accelerator": "gpu", "devices": 1}, [0]),
+        ({"accelerator": "gpu", "devices": 2}, [0, 1]),
+        ({"accelerator": "gpu", "devices": [1]}, [1]),
+        ({"accelerator": "gpu", "devices": "0"}, []),
+        ({"accelerator": "gpu", "devices": "0,"}, [0]),
+    ],
+)
+def test_trainer_data_parallel_device_ids(monkeypatch, trainer_kwargs, expected_data_parallel_device_ids):
+    """Test multi type argument with bool."""
+
+    monkeypatch.setattr(torch.cuda, "is_available", lambda: True)
+    monkeypatch.setattr(torch.cuda, "device_count", lambda: 2)
+
+    trainer = Trainer(**trainer_kwargs)
+    with pytest.deprecated_call(
+        match="`Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8."
+        " Please use `Trainer.device_ids` instead."
+    ):
+        assert trainer.data_parallel_device_ids == expected_data_parallel_device_ids
diff --git a/tests/models/test_gpu.py b/tests/models/test_gpu.py
index 74535d7a569a2..306604737db53 100644
--- a/tests/models/test_gpu.py
+++ b/tests/models/test_gpu.py
@@ -27,7 +27,6 @@
 from pytorch_lightning.utilities import device_parser
 from pytorch_lightning.utilities.exceptions import MisconfigurationException
 from pytorch_lightning.utilities.imports import _compare_version, _TORCHTEXT_LEGACY
-from pytorch_lightning.utilities.rank_zero import rank_zero_only
 from tests.helpers import BoringModel
 from tests.helpers.datamodules import ClassifDataModule
 from tests.helpers.imports import Batch, Dataset, Example, Field, LabelField
@@ -192,13 +191,9 @@ def test_torchelastic_gpu_parsing(mocked_device_count, mocked_is_available, gpus
     trainer = Trainer(gpus=gpus)
     assert isinstance(trainer._accelerator_connector.cluster_environment, TorchElasticEnvironment)
     assert trainer.gpus == gpus
-    if rank_zero_only.rank == 0:
-        with pytest.deprecated_call(
-            match="Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8."
-        ):
-            assert (trainer.data_parallel_device_ids or None) == device_parser.parse_gpu_ids(gpus)
-    else:
-        assert (trainer.data_parallel_device_ids or None) == device_parser.parse_gpu_ids(gpus)
+    # when use gpu
+    if device_parser.parse_gpu_ids(gpus) is not None:
+        assert trainer.device_ids == device_parser.parse_gpu_ids(gpus)
 
 
 @RunIf(min_gpus=1)
diff --git a/tests/trainer/test_trainer_cli.py b/tests/trainer/test_trainer_cli.py
index 00f223c689b3e..37a02e895d560 100644
--- a/tests/trainer/test_trainer_cli.py
+++ b/tests/trainer/test_trainer_cli.py
@@ -161,10 +161,10 @@ def test_argparse_args_parsing_fast_dev_run(cli_args, expected):
 
 
 @pytest.mark.parametrize(
-    ["cli_args", "expected_parsed", "expected_device_ids"],
-    [("", None, []), ("--accelerator gpu --devices 1", "1", [0]), ("--accelerator gpu --devices 0,", "0,", [])],
+    ["cli_args", "expected_parsed"],
+    [("", None), ("--accelerator gpu --devices 1", "1"), ("--accelerator gpu --devices 0,", "0,")],
 )
-def test_argparse_args_parsing_devices(cli_args, expected_parsed, expected_device_ids, monkeypatch):
+def test_argparse_args_parsing_devices(cli_args, expected_parsed, monkeypatch):
     """Test multi type argument with bool."""
 
     monkeypatch.setattr(torch.cuda, "is_available", lambda: True)
@@ -177,11 +177,7 @@ def test_argparse_args_parsing_devices(cli_args, expected_parsed, expected_devic
         args = Trainer.parse_argparser(parser)
 
     assert args.devices == expected_parsed
-    trainer = Trainer.from_argparse_args(args)
-    with pytest.deprecated_call(
-        match="Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8."
-    ):
-        assert trainer.data_parallel_device_ids == expected_device_ids
+    assert Trainer.from_argparse_args(args)
 
 
 @pytest.mark.parametrize(
diff --git a/tests/utilities/test_cli.py b/tests/utilities/test_cli.py
index 46bbbbb6253fb..8afd831901ba7 100644
--- a/tests/utilities/test_cli.py
+++ b/tests/utilities/test_cli.py
@@ -180,7 +180,9 @@ def test_parse_args_parsing_complex_types(cli_args, expected, instantiate):
         assert Trainer.from_argparse_args(args)
 
 
-@pytest.mark.parametrize(["cli_args", "expected_gpu"], [("--gpus 1", [0]), ("--gpus 0,", [0]), ("--gpus 0,1", [0, 1])])
+@pytest.mark.parametrize(
+    ["cli_args", "expected_gpu"], [("--gpus 1", [0]), ("--gpus 0,", [0]), ("--gpus 1,", [1]), ("--gpus 0,1", [0, 1])]
+)
 def test_parse_args_parsing_gpus(monkeypatch, cli_args, expected_gpu):
     """Test parsing of gpus and instantiation of Trainer."""
     monkeypatch.setattr("torch.cuda.device_count", lambda: 2)
@@ -192,10 +194,7 @@ def test_parse_args_parsing_gpus(monkeypatch, cli_args, expected_gpu):
         args = parser.parse_args()
 
     trainer = Trainer.from_argparse_args(args)
-    with pytest.deprecated_call(
-        match="Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8."
-    ):
-        assert trainer.data_parallel_device_ids == expected_gpu
+    assert trainer.device_ids == expected_gpu
 
 
 @pytest.mark.skipif(

From 0a268f6ee636f51a46294b0b0d5a28ea4e20a3ea Mon Sep 17 00:00:00 2001
From: DuYicong515 <duyicong515@gmail.com>
Date: Sun, 20 Mar 2022 11:42:38 -0700
Subject: [PATCH 10/13] keep returning None when GPU not used

---
 pytorch_lightning/trainer/trainer.py    |  4 ++--
 tests/deprecated_api/test_remove_1-8.py | 14 +++++++-------
 2 files changed, 9 insertions(+), 9 deletions(-)

diff --git a/pytorch_lightning/trainer/trainer.py b/pytorch_lightning/trainer/trainer.py
index c32a5a2b2ec68..6365c82eca61a 100644
--- a/pytorch_lightning/trainer/trainer.py
+++ b/pytorch_lightning/trainer/trainer.py
@@ -2086,12 +2086,12 @@ def devices(self) -> int:
         return self.num_devices
 
     @property
-    def data_parallel_device_ids(self) -> List[int]:
+    def data_parallel_device_ids(self) -> Optional[List[int]]:
         rank_zero_deprecation(
             "`Trainer.data_parallel_device_ids` was deprecated in v1.6 and will be removed in v1.8."
             " Please use `Trainer.device_ids` instead."
         )
-        return self.device_ids if isinstance(self.accelerator, GPUAccelerator) else []
+        return self.device_ids if isinstance(self.accelerator, GPUAccelerator) else None
 
     @property
     def lightning_module(self) -> "pl.LightningModule":
diff --git a/tests/deprecated_api/test_remove_1-8.py b/tests/deprecated_api/test_remove_1-8.py
index f556c241e6c82..c5ad12b11a205 100644
--- a/tests/deprecated_api/test_remove_1-8.py
+++ b/tests/deprecated_api/test_remove_1-8.py
@@ -967,21 +967,21 @@ def test_trainer_num_gpu_0(monkeypatch, gpus, expected_num_gpus, strategy):
 @pytest.mark.parametrize(
     ["trainer_kwargs", "expected_data_parallel_device_ids"],
     [
-        ({}, []),
-        ({"devices": 1}, []),
-        ({"devices": "1"}, []),
+        ({}, None),
+        ({"devices": 1}, None),
+        ({"devices": "1"}, None),
         ({"accelerator": "gpu", "devices": 1}, [0]),
         ({"accelerator": "gpu", "devices": 2}, [0, 1]),
         ({"accelerator": "gpu", "devices": [1]}, [1]),
-        ({"accelerator": "gpu", "devices": "0"}, []),
+        ({"accelerator": "gpu", "devices": "0"}, None),
         ({"accelerator": "gpu", "devices": "0,"}, [0]),
     ],
 )
 def test_trainer_data_parallel_device_ids(monkeypatch, trainer_kwargs, expected_data_parallel_device_ids):
     """Test multi type argument with bool."""
-
-    monkeypatch.setattr(torch.cuda, "is_available", lambda: True)
-    monkeypatch.setattr(torch.cuda, "device_count", lambda: 2)
+    if trainer_kwargs.get("accelerator") == "gpu":
+        monkeypatch.setattr(torch.cuda, "is_available", lambda: True)
+        monkeypatch.setattr(torch.cuda, "device_count", lambda: 2)
 
     trainer = Trainer(**trainer_kwargs)
     with pytest.deprecated_call(

From ab6469186a2f776926a93135479d82bf209d01e0 Mon Sep 17 00:00:00 2001
From: "pre-commit-ci[bot]"
 <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Mon, 21 Mar 2022 21:32:31 +0000
Subject: [PATCH 11/13] [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci
---
 tests/deprecated_api/test_remove_1-8.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tests/deprecated_api/test_remove_1-8.py b/tests/deprecated_api/test_remove_1-8.py
index 7f9c84de36548..553eb6f8f860e 100644
--- a/tests/deprecated_api/test_remove_1-8.py
+++ b/tests/deprecated_api/test_remove_1-8.py
@@ -985,7 +985,7 @@ def stop(self, action_name: str) -> None:
     # No deprecation message
     CustomProfiler2()
 
-    
+
 @pytest.mark.parametrize(
     ["trainer_kwargs", "expected_data_parallel_device_ids"],
     [

From 07b7dc5143b1db268c30908e1829eadab60f68f0 Mon Sep 17 00:00:00 2001
From: DuYicong515 <duyicong515@gmail.com>
Date: Mon, 21 Mar 2022 14:35:49 -0700
Subject: [PATCH 12/13] update changelog

---
 CHANGELOG.md | 3 ---
 1 file changed, 3 deletions(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 944250f9d44f8..86fb00b23784d 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -368,9 +368,6 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - The strategies that support `sync_batchnorm` now only apply it when fitting ([#11919](https://github.com/PyTorchLightning/pytorch-lightning/pull/11919))
 
 
-- Changed `Trainer.data_parallel_device_ids` to return a list of GPU device indexes ([#12072](https://github.com/PyTorchLightning/pytorch-lightning/pull/12072))
-
-
 ### Deprecated
 
 - Deprecated `training_type_plugin` property in favor of `strategy` in `Trainer` and updated the references ([#11141](https://github.com/PyTorchLightning/pytorch-lightning/pull/11141))

From 26bcf3bb3ab0c72785ec8f89c570f09c7c3c45e2 Mon Sep 17 00:00:00 2001
From: "pre-commit-ci[bot]"
 <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Date: Tue, 22 Mar 2022 16:31:41 +0000
Subject: [PATCH 13/13] [pre-commit.ci] auto fixes from pre-commit.com hooks

for more information, see https://pre-commit.ci
---
 CHANGELOG.md | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/CHANGELOG.md b/CHANGELOG.md
index ab21e6125e5ed..cda7110dc7429 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -538,7 +538,7 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Deprecated `Trainer.num_processes` in favor of `Trainer.num_devices` ([#12388](https://github.com/PyTorchLightning/pytorch-lightning/pull/12388))
 
 
-- Deprecated `Trainer.data_parallel_device_ids` in favor of `Trainer.device_ids` ([#12072](https://github.com/PyTorchLightning/pytorch-lightning/pull/12072))  
+- Deprecated `Trainer.data_parallel_device_ids` in favor of `Trainer.device_ids` ([#12072](https://github.com/PyTorchLightning/pytorch-lightning/pull/12072))
 
 
 ### Removed