Revert "Fix: skip importing DistributedOptimizer for Windows (#10071)"

This reverts commit c3614f1.

Author: awaelchli

docs/source/advanced/mixed_precision.rst

@@ -50,7 +50,7 @@ BFloat16 Mixed precision is similar to FP16 mixed precision, however we maintain
 Since BFloat16 is more stable than FP16 during training, we do not need to worry about any gradient scaling or nan gradient values that comes with using FP16 mixed precision.
 
 .. testcode::
-    :skipif: not _TORCH_GREATER_EQUAL_DEV_1_10 or not torch.cuda.is_available()
+    :skipif: not _TORCH_GREATER_EQUAL_DEV_1_10
 
     Trainer(gpus=1, precision="bf16")
 

pytorch_lightning/core/lightning.py

@@ -38,7 +38,6 @@
 from pytorch_lightning.core.saving import ModelIO
 from pytorch_lightning.trainer.connectors.logger_connector.fx_validator import _FxValidator
 from pytorch_lightning.utilities import (
-    _IS_WINDOWS,
     _TORCH_GREATER_EQUAL_DEV_1_10,
     GradClipAlgorithmType,
     rank_zero_deprecation,
@@ -2042,7 +2041,7 @@ def _register_sharded_tensor_state_dict_hooks_if_available(self) -> None:
 
         These hooks ensure that ShardedTensors are included when saving, and are loaded the LightningModule correctly.
         """
-        if not _TORCH_GREATER_EQUAL_DEV_1_10 or _IS_WINDOWS:
+        if not _TORCH_GREATER_EQUAL_DEV_1_10:
             return
 
         from torch.distributed._sharded_tensor import pre_load_state_dict_hook, state_dict_hook

pytorch_lightning/plugins/training_type/ddp.py

@@ -42,7 +42,6 @@
 from pytorch_lightning.utilities import (
     _FAIRSCALE_AVAILABLE,
     _HYDRA_AVAILABLE,
-    _IS_WINDOWS,
     _TORCH_GREATER_EQUAL_1_7,
     _TORCH_GREATER_EQUAL_1_8,
     _TORCH_GREATER_EQUAL_1_9,
@@ -58,9 +57,7 @@
 from pytorch_lightning.utilities.types import STEP_OUTPUT
 
 if _TORCH_GREATER_EQUAL_1_10:
-    if not _IS_WINDOWS:
-        from torch.distributed.optim import DistributedOptimizer
-    from torch.distributed.optim import PostLocalSGDOptimizer, ZeroRedundancyOptimizer
+    from torch.distributed.optim import DistributedOptimizer, PostLocalSGDOptimizer, ZeroRedundancyOptimizer
 
 if _FAIRSCALE_AVAILABLE:
     from fairscale.optim import OSS
@@ -336,9 +333,8 @@ def _reinit_optimizers_with_post_localSGD(self, warmup_steps: int):
             if isinstance(optimizer, LightningOptimizer):
                 optimizer = optimizer._optimizer
 
-            is_distributed_optimizer = isinstance(optimizer, DistributedOptimizer) if not _IS_WINDOWS else False
             if (
-                is_distributed_optimizer
+                isinstance(optimizer, DistributedOptimizer)
                 or isinstance(optimizer, ZeroRedundancyOptimizer)
                 or (_FAIRSCALE_AVAILABLE and isinstance(optimizer, OSS))
             ):

pytorch_lightning/utilities/__init__.py

@@ -38,7 +38,6 @@
     _HYDRA_EXPERIMENTAL_AVAILABLE,
     _IPU_AVAILABLE,
     _IS_INTERACTIVE,
-    _IS_WINDOWS,
     _JSONARGPARSE_AVAILABLE,
     _module_available,
     _OMEGACONF_AVAILABLE,

tests/core/test_lightning_module.py

@@ -21,7 +21,7 @@
 
 from pytorch_lightning import Trainer
 from pytorch_lightning.loggers import TensorBoardLogger
-from pytorch_lightning.utilities import _IS_WINDOWS, _TORCH_GREATER_EQUAL_DEV_1_10
+from pytorch_lightning.utilities import _TORCH_GREATER_EQUAL_DEV_1_10
 from pytorch_lightning.utilities.exceptions import MisconfigurationException
 from tests.helpers import BoringModel
 from tests.helpers.runif import RunIf
@@ -315,7 +315,6 @@ def __init__(self, spec):
 @pytest.mark.skipif(
     not _TORCH_GREATER_EQUAL_DEV_1_10, reason="Test requires the torch version to support `ShardedTensor`"
 )
-@pytest.mark.skipif(_IS_WINDOWS, reason="Not supported on Windows")
 def test_sharded_tensor_state_dict(tmpdir, single_process_pg):
     spec = dist._sharding_spec.ChunkShardingSpec(
         dim=0,