Based on discussion here: https://github.com/pytorch/pytorch/pull/138731
Introducing ability for subclass implement type convertion to expected_type.
```
def __coerce_same_metadata_as_tangent__(
self, expected_metadata: Any, expected_type: Optional[Type] = None
):
```
Here if `expected_type=None` means `SubclassClass` is expected.
E.g. for `DTensor` we may find tangent `AsyncCollectiveTensor` where we expected `Tensor` - in this case
`expected_type=Tensor` will be called during runtime
Adding implementation to AsyncCollectiveTensor, that just triggers `wait()`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139095
Approved by: https://github.com/bdhirsh
This PR aims to support the following use case:
```python
def all_reduce_eager(x):
y = x * x
req = dist.all_reduce(y, op=dist.ReduceOp.SUM, async_op=True)
assert isinstance(req, torch.distributed.Work)
return y
@torch.compile(fullgraph=True)
def all_reduce_wait_compiled(y):
torch.ops.c10d_functional.wait_tensor(y)
return y * y
x = torch.ones(1280, 1280, device="cuda") + self.rank
with allow_inflight_collective_as_graph_input_ctx():
y = all_reduce_eager(x)
z = all_reduce_wait_compiled(y)
```
where the collective is issued in eager (with `async_op=True`) but waited in compiled region.
This is important for internal use cases such as TorchRec, where we issue collectives in eager for SparseArch all_to_all but want to wait for them in compiled region at beginning of OverArch, so that the all_to_all can be overlapped with the DenseArch compute that runs in parallel.
----
**Update**: Did two items to prevent regression to existing use cases:
1. Added memory-stressed test case to test_c10d_nccl.py `test_unwaited` to cover existing user's "not calling work.wait() for non-functional collective" use case
2. Gated all new `register_work()` / `unregister_work()` calls with `c10d::allow_inflight_collective_as_graph_input()` check, which is a new context manager that requires explicit user enablement (i.e. not on by default, so should not affect existing users).
The risk of this new version of PR causing regression should be very low.
------
Test commands:
- `pytest -rA test/distributed/test_inductor_collectives.py::TestCollectivesMultiProc::test_eager_async_allreduce_inductor_wait`
- `pytest -rA test/test_fx.py::TestDCE::test_keep_collectives`
- `pytest -rA test/test_fx.py::TestDCE::test_keep_collectives_no_overload`
- `pytest -rA test/distributed/test_c10d_functional_native.py::TestWithNCCL::test_wait_tensor`
- `pytest -rA test/distributed/test_c10d_functional_native.py::TestWithNCCL::test_unwaited`
- `pytest -rA test/distributed/test_c10d_nccl.py::CommTest::test_wait_tensor`
- `pytest -rA test/distributed/test_c10d_nccl.py::CommTest::test_unwaited`
- `pytest -rA test/distributed/_tensor/test_tensor_ops.py::DistTensorOpsTest::test_equal`
- `pytest -rA test/distributed/_tensor/test_random_ops.py::DistTensorRandomOpTest::test_manual_seed`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_ddp_baseline_aot_eager_multiprocess`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_fsdp_aot_eager`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_fsdp_setattr`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_fsdp_unspecialized_forced_getattr_inline`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_fsdp_unspecialized_forced_getattr_no_inline`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_asymmetric_compilation`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_compiler_collectives_automatic_dynamic_scalar`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_compiler_collectives_automatic_dynamic_speculation_divergence`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_compiler_collectives_automatic_dynamic_tensor`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_compiler_collectives_dim_mismatch`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_compiler_collectives_graph_break_empty_graph_still_collective`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_compiler_collectives_missing_source`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_compiler_collectives_scalar_missing_source`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_compiler_collectives_type_mismatch`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_ddp_activation_checkpointing`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_ddp_baseline_aot_eager_multiprocess`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_fsdp_activation_checkpointing`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_fsdp_aot_eager`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_fsdp_inductor`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_fsdp_setattr`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_fsdp_unspecialized_forced_getattr_inline`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_fsdp_unspecialized_forced_getattr_no_inline`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_hf_bert_ddp_aot_eager`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_hf_bert_ddp_aot_eager_static_graph`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_hf_bert_ddp_inductor`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_hf_bert_ddp_inductor_static_graph`
- `pytest -rA test/distributed/test_dynamo_distributed.py::TestMultiProc::test_hf_bert_fsdp_activation_checkpointing`
- `pytest -rA test/distributed/_tensor/test_experimental_ops.py::DistOtherOpsTest::test_bernoulli`
- `pytest -rA test/distributed/_tensor/test_dtensor_compile.py::TestDTensorCompileE2E::test_tp_compile_fullgraph_is_seq_parallel_True`
- `pytest -rA test/distributed/test_inductor_collectives.py::TestCollectivesMultiProc::test_allreduce_inductor_cudagraph_trees`
- `python benchmarks/dynamo/torchbench.py --ci --accuracy --timing --explain --inductor --device cuda --inference --bfloat16 --total-partitions 2 --partition-id 1 --output inference_torchbench.csv --only moco`
------
Differential Revision: [D65023311](https://our.internmc.facebook.com/intern/diff/D65023311)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/137763
Approved by: https://github.com/yifuwang
reland of https://github.com/pytorch/pytorch/pull/133113
I have to create a new PR because the previous reverted PR could not either be rebased, or imported successfully :(
----
Moving DTensor to be in the public namespace, to formally add the documentation page that includes all the public APIs. This includes:
* many path renames and path import fixes
* a dedicated doc page without too much content yet (adding in the next PRs)
* To preserve the BC for users still using the torch.distributed._tensor, I added a shim script to redirect old path calls to the new module
The BC preserving is evidented by the fact that all DTensor tests are still working without changing the public imports. So it's safe to land the changes
Pull Request resolved: https://github.com/pytorch/pytorch/pull/134203
Approved by: https://github.com/tianyu-l
Moving DTensor to be in the public namespace, to formally add the
documentation page that includes all the public APIs. This includes:
* many path renames and path import fixes
* a dedicated doc page without too much content yet (adding in the next
PRs)
* To preserve the BC for users still using the `torch.distributed._tensor`,
I added a shim script to redirect old path calls to the new module
The BC preserving is evidented by the fact that all DTensor tests are still
working without changing the public imports. So it's safe to land the
changes
Pull Request resolved: https://github.com/pytorch/pytorch/pull/133113
Approved by: https://github.com/XilunWu
ghstack dependencies: #133305, #133306
Use `typing_extensions.deprecated` for deprecation annotation if possible. Otherwise, add `category=FutureWarning` to `warnings.warn("message")` if the category is missing.
Note that only warnings that their messages contain `[Dd]eprecat(ed|ion)` are updated in this PR.
Resolves#126888
- #126888
This PR is split from PR #126898.
- #126898
------
Pull Request resolved: https://github.com/pytorch/pytorch/pull/127689
Approved by: https://github.com/Skylion007
Use `typing_extensions.deprecated` for deprecation annotation if possible. Otherwise, add `category=FutureWarning` to `warnings.warn("message")` if the category is missing.
Note that only warnings that their messages contain `[Dd]eprecat(ed|ion)` are updated in this PR.
UPDATE: Use `FutureWarning` instead of `DeprecationWarning`.
Resolves#126888
- #126888
Pull Request resolved: https://github.com/pytorch/pytorch/pull/126898
Approved by: https://github.com/albanD
This PR adds `torch.ops._c10d_functional.all_gather_into_tensor_out`.
It's important for tracing FSDP2, because FSDP2 pre-allocates the output buffer of AllGather, and makes input buffer an alias of the output buffer, and expects both of them to be used to achieve lower memory usage. If we don't preserve this behavior and instead functionalize the AllGather op, AllGather op will then create a brand-new output buffer (instead of reusing), thus significantly increasing the memory usage.
The expectation is that we will "re-inplace" the AllGather op by switching to the out variant in Inductor post-grad stage via an FX pass, so this API is not expected to be directly used by users.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/126334
Approved by: https://github.com/yifuwang, https://github.com/wanchaol
@wanchaol was seeing the loss eventually become NaN when compiling individual transformer blocks in torchtitan - with this patch I no longer see the NaN loss.
The problem is the following:
(1) It is possible to have graph inputs to a compiled region that are AsyncCollectiveTensors. In particular: when we compile individual transformer blocks in the llama model, the first layer (embedding layer) is run in eager mode, and it outputs an AsyncCollectiveTensor that is fed to the first transformer block
(2) ideally, we would like that AsyncCollectiveTensor graph input to desugar into a `wait_tensor()` op that shows up at the beginning of the graph.
(3) the way this is supposed to happen is: AOTAutograd traces through the __torch_dispatch__ of AsyncCollectiveTensor, tracing out a `wait_tensor()` call before dispatching to any of the other ops in the function we are tracing
(4) however: `trigger_wait()` was getting called in a way where we would ignore its output (and return `self.elem` directly), which would cause the `wait_tensor` ops to get DCE'd.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/125677
Approved by: https://github.com/wanchaol, https://github.com/yifuwang
ghstack dependencies: #125676
This PR removes the legacy impls of c10d_functional ops which are now irrelevant. For backward compatibility purpose, c10d_functional ops now call into _c10d_functional ops.
We also changed c10d_functional ops to be CompositeExplicitAutograd, so that when traced, only _c10d_functional ops appear in the graph. After this, we'll be able to remove the Inductor IR for the legacy functional collectives.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/124979
Approved by: https://github.com/wanchaol
## Summary
After this PR, the functional collective Python APIs will stop honoring `TORCH_DISABLE_NATIVE_FUNCOL` and only use native funcol ops. Specifically, this PR:
- Removed `use_native_funcol()`.
- Removed the code path in the Python APIs when `use_native_funcol()` is `False`.
- Changed the CI tests that runs on both native funcol and legacy funcol through the Python API to only run with native funcol.
## Test Changes
`test_functional_api.py`
- Removed the tests where only one of output_split_sizes or input_split_sizes is specified. This behavior is unreliable has been removed from the native funcol.
- Removed `TestWaitiness` which tests an implementation detail of the legacy funcol. We have equivalent tests for native funcol in `test/distributed/test_c10d_functional_native.py` b7fac76fc2/test/distributed/test_c10d_functional_native.py (L114-L116)
`test/distributed/_tensor/test_dtensor.py`
`test/distributed/_tensor/test_dtensor_compile.py`
`test/distributed/test_device_mesh.py`
`test/distributed/_tensor/experimental/test_tp_transform.py`
`test/distributed/_tensor/test_matrix_ops.py`
`test/distributed/test_inductor_collectives.py`
- All these tests were double running with both native funcol and legacy funcol. Changed to only run with native funcol.
`test/distributed/test_c10d_functional_native.py`
- Removed the `run_with_native_funcol` decorators.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/123777
Approved by: https://github.com/wanchaol
ghstack dependencies: #123776
This adds the differentiable collective -- all_to_all_single_grad. This is the initial proof of concept PR and I will be adding the remaining collectives in follow up PRs.
This adds a new function called `all_to_all_single_autograd` which is the autograd variant of `all_to_all_single`. For backwards compatibility + initial testing we wanted to make the autograd variant separate to avoid regressions.
This uses `autograd::Function` to register an Autograd op that calls the original `_c10d_functional::all_to_all_single` via the dispatcher. This works with compile and inductor as opposed to the previous Python implementation that had issues. As this uses the existing `_c10d_functional` ops we don't need to register any meta functions or lowering.
To avoid cudaStream issues this explicitly calls `wait_tensor` in the backward method to ensure it runs under the same stream as the async operation. This hurts performance but can be alleviated potentially using `compile`.
Related work: https://github.com/pytorch/torchrec/blob/main/torchrec/distributed/comm_ops.py
Test plan:
```
pytest test/distributed/test_functional_api.py -k test_all_to_all_single_compile
pytest test/distributed/test_functional_api.py
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/123599
Approved by: https://github.com/yifuwang
Fixes https://github.com/pytorch/pytorch/issues/122404
Previously, when rewriting c10d collectives, if the group argument is
unspecified or None, we create a world pg variable out of thin air and
pass it to the rewrite target. The approach was problematic, as it
assumes the symbol `torch` is available in the scope (see #122404).
After #120560, dynamo can now trace dist.group.WORLD. If the group
argument is unspecified, we can just set it with dist.group.WORLD in the
rewrite target.
Testing
pytest test/distributed/test_inductor_collectives.py -k test_dynamo_rewrite_dist_allreduce
Also verified with the repro provided in #122404
Pull Request resolved: https://github.com/pytorch/pytorch/pull/122561
Approved by: https://github.com/wconstab
ghstack dependencies: #120560
This does not introduce a new test but is tested by checking that all the classes we already have still behave as before now that they don't explicitly disable torch_function.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/120632
Approved by: https://github.com/ezyang
Summary: In non-strict mode of torch.export() we didn't set those `is_compiling()` to `True` which is needed by some models.
Test Plan: Unit tests and manual testing.
Differential Revision: D53624452
Pull Request resolved: https://github.com/pytorch/pytorch/pull/119602
Approved by: https://github.com/suo
There are some tests in this file that are impl specific, e.g. verifying generated code via `FileCheck`. These tests are covered for native funcol in test_c10d_functional_native.py, therefore marking them with `@run_with_legacy_funcol`.
Other tests are marked with `@run_with_both_funcol_impls`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/120025
Approved by: https://github.com/wanchaol
ghstack dependencies: #119982
```
Between the time we switch to the native funcol by default and the time when
we are confident that we can remove the legacy implementation, we want to
ensure that the legacy funcol remains covered by unit tests. This is to
prepare for any potential (but unlikely) reverts. The following utilities
help achieve this goal.
run_with_{native,legacy}_funcol - mark a test to run with only
{native,legacy} funcol. These decorators are for impl specific tests (e.g.
verifying generated code with FileCheck).
run_with_both_funcol_impls - parametrize a test to run with both legacy and
native funcol.
run_with_both_funcol_impls_with_arg - same as run_with_both_funcol_impls, but
passes `enable_native_funcol` to the test so impl specific checks can be
carried out.
```
This PR also marks some tests we want to cover in this fashion. More tests will be marked in subsequent PRs.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/119950
Approved by: https://github.com/wanchaol
ghstack dependencies: #119881
### Summary
@LucasLLC recently implemented `broadcast` in funcol. This is not yet available in the native funcol ops. This PR adds support for broadcast for native funcol.
- Added `_c10d_functional::broadcast` and `_c10d_functional::broadcast_`
- Integrated with python functol broadcast and `AsyncCollectiveTensor`
- Implemented Inductor lowering. Verified correctness and buffer reuse behavior
- Validated dynamo traceability
- Validated AOTInductor compile-ability
Pull Request resolved: https://github.com/pytorch/pytorch/pull/119229
Approved by: https://github.com/wanchaol
ghstack dependencies: #119104
### Summary
Run the relevant tests in `test/distributed/_tensor/test_dtensor_compile.py` and `test/distributed/test_device_mesh.py` with native funcol enabled, in addition to with them being disabled.
All tests excepts `test_tp_compile_comm_reordering` pass. This is expected because the native funcols have slightly different IRs, so the reordering pass needs to be adjusted. This test is disabled for now.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/118437
Approved by: https://github.com/LucasLLC
ghstack dependencies: #118910, #118911
### Summary
- Added `group_name` as the third field in `dim_group_infos`.
- `DeviceMeshTest` now runs both w/ and w/0 `_USE_NATIVE_C10D_FUNCTIONAL=1` in CI.
### Other fixes
- Convert `reduceOp` to lower case before passing it into c10d_functional ops.
- Added a finalizer to handle unwaited collectives (this mirrors the treatment for Python functional collective ops).
Pull Request resolved: https://github.com/pytorch/pytorch/pull/118423
Approved by: https://github.com/wanchaol, https://github.com/LucasLLC, https://github.com/wconstab
This diff introduces an env var `_USE_NATIVE_C10D_FUNCTIONAL` that tells `_functional_collective` to use native `c10d_functional` ops. The Python version and the native version will co-exist until we completely switch to the native version after more testing and verification.
NOTE: `DeviceMesh` support for native `c10d_functional` will be added in a subsequent PR.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/113057
Approved by: https://github.com/LucasLLC, https://github.com/wconstab, https://github.com/wanchaol
Summary:
- Ported `all_to_all_single` to native c10d_functional
- Added Inductor support for the native `all_to_all_single` via the new collective IR's `create_out_of_place()`
- Since the new collective IR derives from `FallbackKernel` which implements a generic `free_unbacked_symbols`, no additional unbacked symbol handling for all_to_all_single is required
Pull Request resolved: https://github.com/pytorch/pytorch/pull/113438
Approved by: https://github.com/yf225, https://github.com/ezyang
This diff aims to directly import DeviceMesh from torch.distributed.device_mesh instead of importing it from dist._tensor. This is done to avoid a circular dependency issue. The code changes in each file of the diff are as follows:
- torch/distributed/_functional_collectives.py: import DeviceMesh from torch.distributed instead of dist._tensor.
Overall, this diff aims to improve the code by avoiding circular dependencies and improving the import statements.
==
The above summary is generated by LLM with minor manual fixes. The following summary is by me.
The original import causes some issues when compiling DDP with compiled_autograd. The root cause of compilation failure is not identified but it is good to fix the lazy initialization, which indirectly fixes the compilation issues for DDP.
Differential Revision: [D51857246](https://our.internmc.facebook.com/intern/diff/D51857246/)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/115649
Approved by: https://github.com/wconstab, https://github.com/wz337
ghstack dependencies: #115523, #115302, #115648
