Fix all-gather duplicate params and wrong dtype #7462
Merged
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
The following assertion error arises when torch autocast is enabled.
[rank3]: File "/opt/deepspeed/deepspeed/runtime/zero/partitioned_param_coordinator.py", line 337, in fetch_sub_module
[rank3]: self.__inflight_param_registry.pop(param).wait(handle_dependency=not fast_fetch)
[rank3]: File "/opt/deepspeed/deepspeed/runtime/zero/partition_parameters.py", line 787, in wait
[rank3]: handle.wait(handle_dependency)
[rank3]: File "/opt/deepspeed/deepspeed/utils/nvtx.py", line 20, in wrapped_fn
[rank3]: ret_val = func(*args, **kwargs)
[rank3]: File "/opt/deepspeed/deepspeed/runtime/zero/partition_parameters.py", line 750, in wait
[rank3]: assert param.ds_status == ZeroParamStatus.INFLIGHT, f"expected param {param.ds_summary()} to be inflight"
[rank3]: AssertionError: expected param {'id': 685, 'status': 'AVAILABLE', 'numel': 131334144, 'ds_numel': 131334144, 'shape': (32064, 4096), 'ds_shape': (32064, 4096), 'requires_grad': True, 'grad_shape': None, 'persist': False, 'active_sub_modules': set(), 'ds_tensor.shape': torch.Size([16416768])} to be inflight
This is due to multiple all-gather ops in the same coalesced all-gather
sharing the same list of params (of mixed dtypes).
Make each all-gather exchange only params of a certain dtype. Also pass the
allgather dtype that matches the params.
Signed-off-by: Junjie Mao <[email protected]>
tjruwase
approved these changes
Aug 2, 2025
tohtana
approved these changes
Aug 3, 2025
LYMDLUT
pushed a commit
to LYMDLUT/DeepSpeed
that referenced
this pull request
Aug 20, 2025
The following assertion error arises when torch autocast is enabled.
[rank3]: File
"/opt/deepspeed/deepspeed/runtime/zero/partitioned_param_coordinator.py",
line 337, in fetch_sub_module
[rank3]:
self.__inflight_param_registry.pop(param).wait(handle_dependency=not
fast_fetch)
[rank3]: File
"/opt/deepspeed/deepspeed/runtime/zero/partition_parameters.py", line
787, in wait
[rank3]: handle.wait(handle_dependency)
[rank3]: File "/opt/deepspeed/deepspeed/utils/nvtx.py", line 20, in
wrapped_fn
[rank3]: ret_val = func(*args, **kwargs)
[rank3]: File
"/opt/deepspeed/deepspeed/runtime/zero/partition_parameters.py", line
750, in wait
[rank3]: assert param.ds_status == ZeroParamStatus.INFLIGHT, f"expected
param {param.ds_summary()} to be inflight"
[rank3]: AssertionError: expected param {'id': 685, 'status':
'AVAILABLE', 'numel': 131334144, 'ds_numel': 131334144, 'shape': (32064,
4096), 'ds_shape': (32064, 4096), 'requires_grad': True, 'grad_shape':
None, 'persist': False, 'active_sub_modules': set(), 'ds_tensor.shape':
torch.Size([16416768])} to be inflight
This is due to multiple all-gather ops in the same coalesced all-gather
sharing the same list of params (of mixed dtypes).
Make each all-gather exchange only params of a certain dtype. Also pass
the allgather dtype that matches the params.
Signed-off-by: Junjie Mao <[email protected]>
Co-authored-by: Junjie Mao <[email protected]>
Co-authored-by: Olatunji Ruwase <[email protected]>
Signed-off-by: lym <[email protected]>
mauryaavinash95
pushed a commit
to DataStates/DeepSpeed
that referenced
this pull request
Oct 4, 2025
The following assertion error arises when torch autocast is enabled.
[rank3]: File
"/opt/deepspeed/deepspeed/runtime/zero/partitioned_param_coordinator.py",
line 337, in fetch_sub_module
[rank3]:
self.__inflight_param_registry.pop(param).wait(handle_dependency=not
fast_fetch)
[rank3]: File
"/opt/deepspeed/deepspeed/runtime/zero/partition_parameters.py", line
787, in wait
[rank3]: handle.wait(handle_dependency)
[rank3]: File "/opt/deepspeed/deepspeed/utils/nvtx.py", line 20, in
wrapped_fn
[rank3]: ret_val = func(*args, **kwargs)
[rank3]: File
"/opt/deepspeed/deepspeed/runtime/zero/partition_parameters.py", line
750, in wait
[rank3]: assert param.ds_status == ZeroParamStatus.INFLIGHT, f"expected
param {param.ds_summary()} to be inflight"
[rank3]: AssertionError: expected param {'id': 685, 'status':
'AVAILABLE', 'numel': 131334144, 'ds_numel': 131334144, 'shape': (32064,
4096), 'ds_shape': (32064, 4096), 'requires_grad': True, 'grad_shape':
None, 'persist': False, 'active_sub_modules': set(), 'ds_tensor.shape':
torch.Size([16416768])} to be inflight
This is due to multiple all-gather ops in the same coalesced all-gather
sharing the same list of params (of mixed dtypes).
Make each all-gather exchange only params of a certain dtype. Also pass
the allgather dtype that matches the params.
Signed-off-by: Junjie Mao <[email protected]>
Co-authored-by: Junjie Mao <[email protected]>
Co-authored-by: Olatunji Ruwase <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The following assertion error arises when torch autocast is enabled.
[rank3]: File "/opt/deepspeed/deepspeed/runtime/zero/partitioned_param_coordinator.py", line 337, in fetch_sub_module
[rank3]: self.__inflight_param_registry.pop(param).wait(handle_dependency=not fast_fetch)
[rank3]: File "/opt/deepspeed/deepspeed/runtime/zero/partition_parameters.py", line 787, in wait
[rank3]: handle.wait(handle_dependency)
[rank3]: File "/opt/deepspeed/deepspeed/utils/nvtx.py", line 20, in wrapped_fn
[rank3]: ret_val = func(*args, **kwargs)
[rank3]: File "/opt/deepspeed/deepspeed/runtime/zero/partition_parameters.py", line 750, in wait
[rank3]: assert param.ds_status == ZeroParamStatus.INFLIGHT, f"expected param {param.ds_summary()} to be inflight"
[rank3]: AssertionError: expected param {'id': 685, 'status': 'AVAILABLE', 'numel': 131334144, 'ds_numel': 131334144, 'shape': (32064, 4096), 'ds_shape': (32064, 4096), 'requires_grad': True, 'grad_shape': None, 'persist': False, 'active_sub_modules': set(), 'ds_tensor.shape': torch.Size([16416768])} to be inflight
This is due to multiple all-gather ops in the same coalesced all-gather sharing the same list of params (of mixed dtypes).
Make each all-gather exchange only params of a certain dtype. Also pass the allgather dtype that matches the params.