[C10D] Fix spelling of MultiProcContinuousTest (#160892)

Pull Request resolved: https://github.com/pytorch/pytorch/pull/160892
Approved by: https://github.com/fduwjj

test/distributed/_test_template.py

@@ -1,10 +1,10 @@
 # Owner(s): ["oncall: distributed"]
 
-from torch.testing._internal.common_distributed import MultiProcContinousTest
+from torch.testing._internal.common_distributed import MultiProcContinuousTest
 from torch.testing._internal.common_utils import run_tests
 
 
-class TestTemplate(MultiProcContinousTest):
+class TestTemplate(MultiProcContinuousTest):
     def testABC(self):
         print(f"rank {self.rank} of {self.world_size} testing ABC")
 

test/distributed/checkpoint/test_pg_transport.py

@@ -25,7 +25,7 @@ from torch.distributed.distributed_c10d import _get_default_group
 from torch.distributed.tensor import DTensor
 from torch.testing._internal.common_cuda import TEST_MULTIGPU
 from torch.testing._internal.common_distributed import (
-    MultiProcContinousTest,
+    MultiProcContinuousTest,
     requires_nccl,
 )
 from torch.testing._internal.common_utils import (
@@ -201,7 +201,7 @@ def _test_pg_transport_with_sharded_tensor(self, device) -> None:
     torch.testing.assert_close(expected_local_tensor, received_local_tensor)
 
 
-class PgTransportCPU(MultiProcContinousTest):
+class PgTransportCPU(MultiProcContinuousTest):
     world_size = 8
     timeout: timedelta = timedelta(seconds=20)
 
@@ -227,7 +227,7 @@ class PgTransportCPU(MultiProcContinousTest):
         _test_pg_transport_with_sharded_tensor(self, self.device)
 
 
-class PgTransportCUDA(MultiProcContinousTest):
+class PgTransportCUDA(MultiProcContinuousTest):
     world_size = 2
     timeout: timedelta = timedelta(seconds=20)
 

test/distributed/pipelining/test_schedule_multiproc.py

@@ -31,7 +31,7 @@ from torch.distributed.pipelining.schedules import _PipelineScheduleRuntime
 from torch.nn.modules.loss import MSELoss
 from torch.testing._internal.common_cuda import TEST_MULTIGPU
 from torch.testing._internal.common_distributed import (
-    MultiProcContinousTest,
+    MultiProcContinuousTest,
     requires_nccl,
 )
 from torch.testing._internal.common_utils import (
@@ -199,7 +199,7 @@ def zero_gradients(stage_modules):
         stage_module.zero_grad()
 
 
-class ScheduleTest(MultiProcContinousTest):
+class ScheduleTest(MultiProcContinuousTest):
     world_size = 4
 
     @classmethod
@@ -802,7 +802,7 @@ class ScheduleTest(MultiProcContinousTest):
 instantiate_parametrized_tests(ScheduleTest)
 
 
-class CustomSchedulesTest(MultiProcContinousTest):
+class CustomSchedulesTest(MultiProcContinuousTest):
     """
     These schedules are from the ScheduleRegistry and require world_size == 2
     The schedules test weird and unconventional schedules for edge cases

test/distributed/pipelining/test_stage.py

@@ -16,7 +16,7 @@ from torch.distributed.pipelining import (
 from torch.distributed.pipelining._utils import PipeliningShapeError
 from torch.testing._internal.common_cuda import TEST_MULTIGPU
 from torch.testing._internal.common_distributed import (
-    MultiProcContinousTest,
+    MultiProcContinuousTest,
     MultiProcessTestCase,
     requires_nccl,
 )
@@ -63,7 +63,7 @@ def get_flatten_hook():
     return flatten_hook
 
 
-class StageTest(MultiProcContinousTest):
+class StageTest(MultiProcContinuousTest):
     @classmethod
     def backend_str(cls) -> str:
         # Testing with NCCL backend

test/distributed/test_c10d_ops_nccl.py

@@ -25,7 +25,7 @@ import torch.distributed as dist
 from torch.testing._internal.common_cuda import TEST_MULTIGPU
 from torch.testing._internal.common_distributed import (
     init_multigpu_helper,
-    MultiProcContinousTest,
+    MultiProcContinuousTest,
     requires_nccl,
     requires_nccl_version,
     sm_is_or_higher_than,
@@ -45,7 +45,7 @@ if TEST_WITH_DEV_DBG_ASAN:
     sys.exit(0)
 
 
-class ProcessGroupNCCLOpTest(MultiProcContinousTest):
+class ProcessGroupNCCLOpTest(MultiProcContinuousTest):
     @classmethod
     def backend_str(cls) -> str:
         return "nccl"

test/distributed/test_composability.py

@@ -19,7 +19,7 @@ from torch.distributed.tensor import DTensor
 from torch.nn.parallel import DistributedDataParallel as DDP
 from torch.testing._internal.common_cuda import TEST_MULTIGPU
 from torch.testing._internal.common_distributed import (
-    MultiProcContinousTest,
+    MultiProcContinuousTest,
     requires_nccl,
     skip_if_lt_x_gpu,
 )
@@ -91,7 +91,7 @@ def loss_fn(y, target, scale=1e-4):
     return torch.nn.functional.cross_entropy(y, target) * scale
 
 
-class ComposabilityTest(MultiProcContinousTest):
+class ComposabilityTest(MultiProcContinuousTest):
     @classmethod
     def backend_str(cls) -> str:
         # Testing with NCCL backend

test/distributed/test_cupy_as_tensor.py

@@ -7,7 +7,7 @@ from dataclasses import dataclass
 
 import torch
 from torch.multiprocessing.reductions import reduce_tensor
-from torch.testing._internal.common_distributed import MultiProcContinousTest
+from torch.testing._internal.common_distributed import MultiProcContinuousTest
 from torch.testing._internal.common_utils import (
     requires_cuda_p2p_access,
     run_tests,
@@ -46,7 +46,7 @@ def from_buffer(
 
 
 @requires_cuda_p2p_access()
-class CupyAsTensorTest(MultiProcContinousTest):
+class CupyAsTensorTest(MultiProcContinuousTest):
     @classmethod
     def backend_str(cls):
         return "gloo"

test/distributed/test_nccl.py

@@ -14,7 +14,7 @@ from torch.testing._internal.common_device_type import (
     instantiate_device_type_tests,
 )
 from torch.testing._internal.common_distributed import (
-    MultiProcContinousTest,
+    MultiProcContinuousTest,
     skip_if_lt_x_gpu,
 )
 from torch.testing._internal.common_utils import (
@@ -246,7 +246,7 @@ class TestNCCL(TestCase):
 
 
 @requires_cuda_p2p_access()
-class NCCLSymmetricMemoryTest(MultiProcContinousTest):
+class NCCLSymmetricMemoryTest(MultiProcContinuousTest):
     @property
     def device(self) -> torch.device:
         return torch.device("cuda", self.rank)

test/distributed/test_nvshmem.py

@@ -7,7 +7,7 @@
 import torch
 import torch.distributed as dist
 import torch.distributed._symmetric_memory as symm_mem
-from torch.testing._internal.common_distributed import MultiProcContinousTest
+from torch.testing._internal.common_distributed import MultiProcContinuousTest
 from torch.testing._internal.common_utils import (
     instantiate_parametrized_tests,
     parametrize,
@@ -33,7 +33,7 @@ device_module = torch.get_device_module(device_type)
 
 @requires_nvshmem()
 @requires_cuda_p2p_access()
-class NVSHMEMSymmetricMemoryTest(MultiProcContinousTest):
+class NVSHMEMSymmetricMemoryTest(MultiProcContinuousTest):
     def _init_device(self) -> None:
         # TODO: relieve this (seems to hang if without)
         device_module.set_device(self.device)
@@ -128,7 +128,7 @@ class NVSHMEMSymmetricMemoryTest(MultiProcContinousTest):
 @instantiate_parametrized_tests
 @requires_nvshmem()
 @requires_cuda_p2p_access()
-class NVSHMEMAll2AllTest(MultiProcContinousTest):
+class NVSHMEMAll2AllTest(MultiProcContinuousTest):
     def _init_device(self) -> None:
         # TODO: relieve this (seems to hang if without)
         device_module.set_device(self.device)

test/distributed/test_nvshmem_triton.py

@@ -9,7 +9,7 @@ import torch.distributed as dist
 import torch.distributed._symmetric_memory as symm_mem
 import torch.distributed._symmetric_memory._nvshmem_triton as nvshmem
 from torch._inductor.runtime.triton_compat import triton
-from torch.testing._internal.common_distributed import MultiProcContinousTest
+from torch.testing._internal.common_distributed import MultiProcContinuousTest
 from torch.testing._internal.common_utils import (
     instantiate_parametrized_tests,
     parametrize,
@@ -246,7 +246,7 @@ def nvshmem_reduce_kernel(
 
 @instantiate_parametrized_tests
 @requires_nvshmem()
-class NVSHMEMTritonTest(MultiProcContinousTest):
+class NVSHMEMTritonTest(MultiProcContinuousTest):
     def _init_device(self) -> None:
         # TODO: relieve this (seems to hang if without)
         device_module.set_device(self.device)

test/distributed/test_p2p_ipc.py

@@ -6,7 +6,7 @@
 
 import torch
 from torch.multiprocessing.reductions import reduce_tensor
-from torch.testing._internal.common_distributed import MultiProcContinousTest
+from torch.testing._internal.common_distributed import MultiProcContinuousTest
 from torch.testing._internal.common_utils import (
     requires_cuda_p2p_access,
     run_tests,
@@ -20,7 +20,7 @@ device_module = torch.get_device_module(device_type)
 
 
 @requires_cuda_p2p_access()
-class P2PIpcTest(MultiProcContinousTest):
+class P2PIpcTest(MultiProcContinuousTest):
     @classmethod
     def backend_str(cls):
         return "gloo"

test/distributed/test_symmetric_memory.py

@@ -24,7 +24,7 @@ from torch.distributed._symmetric_memory import (
 from torch.testing._internal.common_cuda import _get_torch_cuda_version, SM90OrLater
 from torch.testing._internal.common_device_type import e4m3_type
 from torch.testing._internal.common_distributed import (
-    MultiProcContinousTest,
+    MultiProcContinuousTest,
     MultiProcessTestCase,
     requires_multicast_support,
     skip_if_lt_x_gpu,
@@ -52,7 +52,7 @@ device_module = torch.get_device_module(device_type)
 
 @instantiate_parametrized_tests
 @requires_cuda_p2p_access()
-class SymmetricMemoryTest(MultiProcContinousTest):
+class SymmetricMemoryTest(MultiProcContinuousTest):
     @property
     def device(self) -> torch.device:
         return torch.device(device_type, self.rank)
@@ -636,7 +636,7 @@ class SymmetricMemoryTest(MultiProcContinousTest):
 
 # This Test class is used to test the error handling of SymmetricMemory APIs.
 # Since a process restart is often needed after each test, we use the
-# MultiProcessTestCase instead of MultiProcContinousTest.
+# MultiProcessTestCase instead of MultiProcContinuousTest.
 @requires_cuda_p2p_access()
 class SymmMemNegativeTest(MultiProcessTestCase):
     def setUp(self) -> None:
@@ -746,7 +746,7 @@ class SymmMemNegativeTest(MultiProcessTestCase):
 
 @instantiate_parametrized_tests
 @requires_cuda_p2p_access()
-class SymmMemCollectiveTest(MultiProcContinousTest):
+class SymmMemCollectiveTest(MultiProcContinuousTest):
     @property
     def device(self) -> torch.device:
         return torch.device(device_type, self.rank)
@@ -993,7 +993,7 @@ class SymmMemCollectiveTest(MultiProcContinousTest):
 
 @instantiate_parametrized_tests
 @requires_cuda_p2p_access()
-class LoweringTest(MultiProcContinousTest):
+class LoweringTest(MultiProcContinuousTest):
     def _init_process(self) -> None:
         torch.cuda.set_device(self.device)
         enable_symm_mem_for_group(dist.group.WORLD.group_name)

torch/testing/_internal/common_distributed.py

@@ -1568,7 +1568,7 @@ class DynamoDistributedMultiProcTestCase(DistributedTestBase):
         self.run_test(test_name, parent_pipe)
 
 
-class MultiProcContinousTest(TestCase):
+class MultiProcContinuousTest(TestCase):
     # Class variables:
     MAIN_PROCESS_RANK = -1
     # number of test processes

torch/testing/_internal/distributed/_tensor/common_dtensor.py

@@ -31,7 +31,7 @@ from torch.distributed.tensor.parallel import (
     SequenceParallel,
 )
 from torch.testing._internal.common_distributed import (
-    MultiProcContinousTest,
+    MultiProcContinuousTest,
     MultiProcessTestCase,
     MultiThreadedTestCase,
     run_subtests,
@@ -337,7 +337,7 @@ def skip_unless_torch_gpu(method: T) -> T:
     return cast(T, skip_if_lt_x_gpu(NUM_DEVICES)(method))
 
 
-class DTensorContinuousTestBase(MultiProcContinousTest):
+class DTensorContinuousTestBase(MultiProcContinuousTest):
     @classmethod
     def device_type(cls) -> str:
         # if enough GPU/XPU/HPU we can use those devices, otherwise we fallback to CPU