For https://github.com/pytorch/pytorch/issues/114850, we will port distributed tests to Intel GPU.
We could enable Intel GPU with following methods and try the best to keep the original code styles:
- instantiate_device_type_tests()
- use "torch.accelerator.current_accelerator()" to determine the accelerator backend
- enabled XPU for some test path
- Unify some common code under torch/testing/_internal for multiple backend, for example:
- requires_nccl_version
- _dynamo_dist_per_rank_init
- DynamoDistributedSingleProcTestCase
- DistTestCases
- FSDPTestMultiThread
Pull Request resolved: https://github.com/pytorch/pytorch/pull/158533
Approved by: https://github.com/guangyey, https://github.com/d4l3k
Co-authored-by: Yu, Guangye <106960996+guangyey@users.noreply.github.com>
Usually `world_size=torch.cuda.device_count()` for FSDPTest-based tests
But distributed test class `TestFullyShardAllGatherExtensionsMultiProcess` [forces to use `world_size=2`](0a6e1d6b9b/test/distributed/_composable/fsdp/test_fully_shard_extensions.py (L170)) even for 1 GPU.
Then NCCL fails with errors:
```
HIP_VISIBLE_DEVICES=0 python distributed/_composable/fsdp/test_fully_shard_extensions.py -v -k test_all_gather_extensions_train_parity
...
ncclInvalidUsage: This usually reflects invalid usage of NCCL library.
Duplicate GPU detected : rank 1 and rank 0 both on CUDA device c000
Duplicate GPU detected : rank 0 and rank 1 both on CUDA device c000
```
The test method [has `@skip_if_lt_x_gpu(2)` decorator](0a6e1d6b9b/test/distributed/_composable/fsdp/test_fully_shard_extensions.py (L209)), but test fails during test class initialization before decorator activation
This PR will skip FSDPtest-based tests if `world_size > torch.cuda.device_count()`
```
HIP_VISIBLE_DEVICES=0 python distributed/_composable/fsdp/test_fully_shard_extensions.py -v -k test_all_gather_extensions_train_parity
...
dist init r=0, world=2
dist init r=1, world=2
SKIPPED [15.5507s] (Need at least 2 CUDA devices)
```
Pull Request resolved: https://github.com/pytorch/pytorch/pull/155836
Approved by: https://github.com/jeffdaily
Pytest considers all symbols starting with `test_` as a test case/function and runs them.
The `test_compiled_fsdp` is a decorator but due to the import discovered by pytest.
Rename it to avoid.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/144544
Approved by: https://github.com/Skylion007
Motivation: Generalize unit tests so that can be executed for cuda and non cuda devices.
Depedency : #133209 Merged now.
There was a #135242 for these changes and closed due to in correct commits. I have incoroprated the changes as suggested in comments.
@kwen2501 @zeshengzong Please review the changes.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139184
Approved by: https://github.com/kwen2501
Co-authored-by: Yu, Guangye <guangye.yu@intel.com>
Motivation: Generalize unit tests so that can be executed for cuda and non cuda devices.
Depedency : #133209 Merged now.
There was a #135242 for these changes and closed due to in correct commits. I have incoroprated the changes as suggested in comments.
@kwen2501 @zeshengzong Please review the changes.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/139184
Approved by: https://github.com/kwen2501
Co-authored-by: Yu, Guangye <guangye.yu@intel.com>
**Overview**
This PR moves `torch/distributed/_composable/fsdp` to `torch/distributed/fsdp/_fully_shard` and makes public APIs available from `torch.distributed.fsdp`, e.g.:
```
from torch.distributed.fsdp import fully_shard
```
This is targeting 2.6 release. I rewrote some of the documentation with (hopefully) improved phrasing.
**Follow-Ups**
- [x] Add some explanation in the docs about FSDP1 vs. FSDP2
- [ ] Move unit tests from `test/distributed/_composable/fsdp` to `test/distributed/fsdp/fully_shard/`
Pull Request resolved: https://github.com/pytorch/pytorch/pull/141868
Approved by: https://github.com/kwen2501, https://github.com/wconstab, https://github.com/weifengpy
Co-authored-by: Svetlana Karslioglu <svekars@meta.com>
Adding `destroy_pg_upon_exit` property to allow derived Test classes to control whether auto destroy is desired.
(Otherwise, derived test classes will need to rewrite the `_run()` method, leading to duplicated code of `_run()` and if one needs to add things to `_run` in the future, more code change is needed.)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/141192
Approved by: https://github.com/wconstab
Faced with an annoying string of warnings like this when running tests,
<img width="1644" alt="Screenshot 2024-11-15 at 11 23 21 AM" src="https://github.com/user-attachments/assets/91ff4e1d-3c29-4510-9a61-46e7df68a212">
My choices seem to be (1) call destroy_process_group() at the end of
each test fn, (2) do this in some wrapper, (3) do it in the base test
class.
Since tests in MultiProcessTestCase are responsible for calling
init_process_group themselves, they should also be responsible for
calling destroy (or at least method (3) would be asymmetric and may
result in double-destroy).
But it doesn't feel worth it to go add a destroy call manually to each
test, and try/except for a possible second destroy call seems like a
happy middle ground.
Note: tests that want to ensure that destroy runs cleanly can and should
still call destroy _inside_ the test, and this change does not affect
that.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140820
Approved by: https://github.com/fegin
Faced with an annoying string of warnings like this when running tests,
<img width="1644" alt="Screenshot 2024-11-15 at 11 23 21 AM" src="https://github.com/user-attachments/assets/91ff4e1d-3c29-4510-9a61-46e7df68a212">
My choices seem to be (1) call destroy_process_group() at the end of
each test fn, (2) do this in some wrapper, (3) do it in the base test
class.
Since tests in MultiProcessTestCase are responsible for calling
init_process_group themselves, they should also be responsible for
calling destroy (or at least method (3) would be asymmetric and may
result in double-destroy).
But it doesn't feel worth it to go add a destroy call manually to each
test, and try/except for a possible second destroy call seems like a
happy middle ground.
Note: tests that want to ensure that destroy runs cleanly can and should
still call destroy _inside_ the test, and this change does not affect
that.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/140820
Approved by: https://github.com/fegin
ghstack dependencies: #140460, #140815
## Motivation
The FSDP common code for FSDP UT execution is mostly written with cuda device in mind. However other devices such the intel Gaudi supports most of the functionality. We are generalizing the base content so that the UT content can be used for non-cuda device execution.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/133209
Approved by: https://github.com/kwen2501
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
This PR creates these `GroupedSchedulerNode`s:
- One for each all-gather code block (cast + copy-in + all-gather)
- One for each all-gather-wait code block (all-gather-wait + copy-out)
- One for each reduce-scatter code block (copy-in + reduce-scatter)
- One for each reduce-scatter-wait code block (reduce-scatter-wait)
This serves two goals:
- Prevent outside ops from being fused into these op groups, in order to have more predicable memory usage.
- Make it easier to specify the dependency e.g. from `i+1` all-gather group node to the `i` all-gather-wait group node, to enforce FSDP2 comm ordering (i.e. "serialization of comms").
The actual "reorder-for-FSDP-compute-comm-overlap" PR will come next.
Test commands:
- `pytest -rA test/distributed/test_compute_comm_reordering.py::TestComputeCommReorderingMultiProc`
- `pytest -rA test/distributed/_composable/fsdp/test_fully_shard_compile.py::TestFullyShardCompile::test_transformer_backend_inductor`
- `pytest -rA test/distributed/_composable/fsdp/test_fully_shard_compile.py::TestFullyShardCompile::test_nested_fully_shard_backend_inductor`
Pull Request resolved: https://github.com/pytorch/pytorch/pull/131510
Approved by: https://github.com/yifuwang
This PR adds two APIs `set_modules_to_forward_prefetch` and `set_modules_to_backward_prefetch` to enable explicit forward/backward all-gather prefetching, respectively.
```
def set_modules_to_forward_prefetch(self, modules: List[FSDPModule]): -> None
def set_modules_to_backward_prefetch(self, modules: List[FSDPModule]): -> None
```
**Motivation**
FSDP2 implements _reasonable defaults_ for forward and backward prefetching. In forward, it uses implicit prefetching and allows two all-gather output tensors to be alive at once (so that the current all-gather copy-out can overlap with the next all-gather). In backward, it uses explicit prefetching based on the reverse post-forward order.
However, there may be cases where with expert knowledge, we can reduce communication bubbles by moving all-gathers manually. One way to expose such behavior is to expose _prefetching limits_, i.e. integers that configure how many outstanding all-gathers/all-gather output tensors can be alive at once. IMIHO, this leans toward _easy_, not _simple_ (see [PyTorch design principles](https://pytorch.org/docs/stable/community/design.html#principle-2-simple-over-easy)).
The crux of the problem is that there may be special cases where manual intervention can give better performance. Exposing a prefetching limit and allowing users to pass a value >1 just smooths over the problem since such a limit would generally apply over the entire model even though it possibly should not. Then, expert users will see a specific all-gather that they want to deviate from this limit, and there is little we can do.
Thus, we instead choose to expose the most primitive extension point: namely, every `FSDPModule` gives an opportunity to prefetch other all-gathers in forward and in backward. How to leverage this extension point is fully up to the user. Implementing the prefetch limit can be done using this extension point (e.g. record the post-forward order yourself using forward hooks, iterate over that order, and call the `set_modules_to_forward_prefetch` / `set_modules_to_backward_prefetch` APIs).
Differential Revision: [D58700346](https://our.internmc.facebook.com/intern/diff/D58700346)
Pull Request resolved: https://github.com/pytorch/pytorch/pull/128884
Approved by: https://github.com/ckluk2, https://github.com/weifengpy
This CI showcases FSDP2 works with `torch.compile` root model, since FSDP1 can do the same
compiling root Transformer without AC: `pytest test/distributed/_composable/fsdp/test_fully_shard_training.py -k test_train_parity_multi_group`
compiling root Transformer with AC: `pytest test/distributed/_composable/fsdp/test_fully_shard_training.py -k test_train_parity_with_activation_checkpointing`
Pull Request resolved: https://github.com/pytorch/pytorch/pull/127832
Approved by: https://github.com/awgu
**Overview**
This PR adds pre/post-all-gather extensions to FSDP2.
- The pre/post-all-gather extensions are specified at the tensor-level on the `sharded_param._local_tensor` (i.e. the tensor wrapped by the sharded `DTensor`). If the user has a tensor-subclass parameter on the module passed to FSDP that preserves the subclass through the sharding ops (e.g. `new_zeros`, `chunk`, etc.), then the `sharded_param._local_tensor` will naturally be of that subclass.
- The pre-all-gather function has signature:
```
def fsdp_pre_all_gather(self) -> Tuple[Tuple[torch.Tensor, ...], Any]
```
- The first return value is a `Tuple[torch.Tensor, ...]` of the all-gather inputs. It is a tuple since a subclass could contribute >1 inner tensors.
- The second return value is any optional metadata needed to pass through to the post-all-gather.
- The post all-gather function has signature:
```
def fsdp_post_all_gather(
self,
all_gather_outputs: Tuple[torch.Tensor, ...],
metadata: Any,
param_dtype: torch.dtype,
*,
out: Optional[torch.Tensor] = None,
) -> Union[Tuple[torch.Tensor, Tuple[torch.Tensor, ...]], None]:
```
- The `all_gather_outputs` are exactly the all-gathered versions of the `fsdp_pre_all_gather` 1st return value (representing the all-gather inputs). We make sure to unflatten these back to ND for the user.
- The `metadata` is the `fsdp_pre_all_gather` 2nd return value, untouched.
- The `param_dtype` is the parameter dtype based on the passed-in `MixedPrecisionPolicy`. Namely, if no policy is passed in, then `param_dtype` is the original dtype, and otherwise, it is the `MixedPrecisionPolicy.param_dtype`.
- If `out` is not specified, then the return value has type `Tuple[torch.Tensor, Tuple[torch.Tensor, ...]]`. The first tuple item is the unsharded parameter (e.g. re-wrapping into some subclass). The second tuple item is a tuple of unsharded inner tensors that FSDP should free during reshard. These should be derived from the all-gather outputs.
- The `out` argument is required due to FSDP's `resize_` usage. We require an in-place variant for the backward all-gather. Here, `out` will be exactly the object returned as the first tuple item in the out-of-place variant mentioned before. The unsharded inner tensors will be allocated before calling `fsdp_post_all_gather`. When `out` is specified, the `fsdp_post_all_gather` should return `None`. If the post-all-gather does not do any out-of-place ops, then the `out` variant can just be a no-op since the unsharded inner tensors will be the same as the all-gather outputs, which FSDP directly writes to after all-gather. (E.g., this is the case for both float8 and `NF4Tensor`.)
- We check for `fsdp_pre_all_gather` and `fsdp_post_all_gather` directly via `hasattr` to accommodate monkey patching so that we do not strictly require the user to use a tensor subclass. The monkey patch must happen after the local tensors have been finalized (after applying FSDP and after any meta-device init).
- For now, we require that all gradients in one FSDP parameter group share the same dtype. This is fine for float8 and `NF4Tensor` use cases. If this requirement is too strict, then in the future we can issue 1 reduce-scatter per dtype per group.
**Design Notes**
- We assume that the `sharded_param._local_tensor` is padded on dim-0.
- This assumption should not block immediate use cases, and when we pad the `DTensor._local_tensor` by default, this assumption will always be true.
- This assumption allows us to call `sharded_param._local_tensor.fsdp_pre_all_gather()`; i.e. it tells us from which tensor object to invoke `fsdp_pre_all_gather()`.
- Suppose we want to compose with CPU offloading. Then, CPU offloading's H2D copy should run first, i.e. `sharded_param._local_tensor.to("cuda").fsdp_pre_all_gather()`, where `_local_tensor.to("cuda")` should return an instance of the subclass so that it still defines `fsdp_pre_all_gather()`. Note that in this case, the subclass instance on GPU is a temporary, which means caching values on it would not be possible. One possibility would be to have `.to("cuda")` move any cached values too.
- `fsdp_post_all_gather` can either return an unsharded parameter that aliases with the all-gather output or does not alias, but there is no way to know a priori.
- If the unsharded parameter aliases with the all-gather output, then we should _not_ free the all-gather output in `unshard`.
- If the unsharded parameter does not alias with the all-gather output, then we prefer to free the all-gather output in `unshard` to avoid holding the unneeded temporary.
- One approach is for eager-mode to check for this alias (by comparing data pointers). However, this might be adversarial to full-graph compilation. The compromise for simplicity can be to always free the all-gather output in `reshard`.
Pull Request resolved: https://github.com/pytorch/pytorch/pull/122908
Approved by: https://github.com/weifengpy, https://github.com/wanchaol
ghstack dependencies: #119302
Summary:
without this the `ProcessGroupNCCL` lib would try to infer the device id and emit a warning.
This doesn't change the behavior just makes it explicit.
> ProcessGroupNCCL.cpp:3720] [PG 0 Rank 1] using GPU 1 to perform barrier as devices used by this process are currently unknown. This can potentially cause a hang if this rank to GPU mapping is incorrect.Specify device_ids in barrier() to force use of a particular device.
Test Plan: CI
Differential Revision: D55998175
Pull Request resolved: https://github.com/pytorch/pytorch/pull/123866
Approved by: https://github.com/awgu
