Implement generate_vbe_metadata cpu #3715
Open
Conversation
This file contains 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
|
This pull request was exported from Phabricator. Differential Revision: D69162870 |
✅ Deploy Preview for pytorch-fbgemm-docs ready!
To edit notification comments on pull requests, go to your Netlify site configuration. |
spcyppt
force-pushed
the
export-D69162870
branch
from
7529dfe to
b2d0bcd
Compare
spcyppt
added a commit
to spcyppt/FBGEMM
that referenced
this pull request
Feb 21, 2025
Summary: X-link: facebookresearch/FBGEMM#796 This diff implements `generate_vbe_metadata` for cpu, such that the function returns the same output for CPU, CUDA and MTIA. To support VBE on CPU with existing fixed-batch-size CPU kernel, we need to recompute offsets, which is previously done in python. This diff implements offsets recomputation in C++ such that all manipulations are done in C++. Note that reshaping offsets and grad_input to work with existing fixed-batch-size CPU kernels are done in Autograd instead of wrapper to avoid multiple computations. VBE CPU tests are in the next diff. Differential Revision: D69162870
|
This pull request was exported from Phabricator. Differential Revision: D69162870 |
1 similar comment
|
This pull request was exported from Phabricator. Differential Revision: D69162870 |
spcyppt
force-pushed
the
export-D69162870
branch
from
b2d0bcd to
aac9690
Compare
spcyppt
added a commit
to spcyppt/FBGEMM
that referenced
this pull request
Feb 21, 2025
Summary: Pull Request resolved: pytorch#3715 X-link: facebookresearch/FBGEMM#796 This diff implements `generate_vbe_metadata` for cpu, such that the function returns the same output for CPU, CUDA and MTIA. To support VBE on CPU with existing fixed-batch-size CPU kernel, we need to recompute offsets, which is previously done in python. This diff implements offsets recomputation in C++ such that all manipulations are done in C++. Note that reshaping offsets and grad_input to work with existing fixed-batch-size CPU kernels are done in Autograd instead of wrapper to avoid multiple computations. VBE CPU tests are in the next diff. Differential Revision: D69162870
|
This pull request was exported from Phabricator. Differential Revision: D69162870 |
spcyppt
added a commit
to spcyppt/FBGEMM
that referenced
this pull request
Feb 21, 2025
Summary: Pull Request resolved: pytorch#3715 X-link: facebookresearch/FBGEMM#796 This diff implements `generate_vbe_metadata` for cpu, such that the function returns the same output for CPU, CUDA and MTIA. To support VBE on CPU with existing fixed-batch-size CPU kernel, we need to recompute offsets, which is previously done in python. This diff implements offsets recomputation in C++ such that all manipulations are done in C++. Note that reshaping offsets and grad_input to work with existing fixed-batch-size CPU kernels are done in Autograd instead of wrapper to avoid multiple computations. VBE CPU tests are in the next diff. Differential Revision: D69162870
spcyppt
force-pushed
the
export-D69162870
branch
from
aac9690 to
ae43025
Compare
Summary: Annotate tensors in the schema. Reference doc: https://docs.google.com/document/d/1r_2-jPbqvoD069aEL_3pTiBOuYf1atu9L2ZQhA1Vut8/edit?tab=t.0 Differential Revision: D70210509
Summary: X-link: pytorch/torchrec#2751 X-link: facebookresearch/FBGEMM#793 **Backend**: D68054868 --- As the number of arguments in TBE keeps growing, some of the optimizers run into number of arguments limitation (i.e., 64) during pytorch operation registration. **For long-term growth and maintenance, we hence redesign TBE API by packing some of the arguments into list. Note that not all arguments are packed.** We pack the arguments as a list for each type. For **common** arguments, we pack - weights and arguments of type `Momentum` into TensorList - other tensors and optional tensors to list of optional tensors `aux_tensor` - `int` arguments into `aux_int` - `float` arguments into `aux_float` - `bool` arguments into `aux_bool`. Similarly for **optimizer-specific** arguments, we pack - arguments of type `Momentum` that are *__not__ optional* into TensorList - *optional* tensors to list of optional tensors `optim_tensor` - `int` arguments into `optim_int` - `float` arguments into `optim_float` - `bool` arguments into `optim_bool`. We see issues with pytorch registration across packing SymInt in python-C++, so we unroll and pass SymInt arguments individually. **This significantly reduces number of arguments.** For example, `split_embedding_codegen_lookup_rowwise_adagrad_with_counter_function`, which currently has 61 arguments only have 26 arguments with this API design. Please refer to the design doc on which arguments are packed and signature. Design doc: https://docs.google.com/document/d/1dCBg7dcf7Yq9FHVrvXsAmFtBxkDi9o6u0r-Ptd4UDPE/edit?tab=t.0#heading=h.6bip5pwqq8xb Full signature for each optimizer lookup function will be provided shortly. Reviewed By: sryap Differential Revision: D68055168
Summary: X-link: facebookresearch/FBGEMM#796 This diff implements `generate_vbe_metadata` for cpu, such that the function returns the same output for CPU, CUDA and MTIA. To support VBE on CPU with existing fixed-batch-size CPU kernel, we need to recompute offsets, which is previously done in python. This diff implements offsets recomputation in C++ such that all manipulations are done in C++. Note that reshaping offsets and grad_input to work with existing fixed-batch-size CPU kernels are done in Autograd instead of wrapper to avoid multiple computations. VBE CPU tests are in the next diff. Differential Revision: D69162870
spcyppt
added a commit
to spcyppt/FBGEMM
that referenced
this pull request
Feb 26, 2025
Summary: X-link: facebookresearch/FBGEMM#796 This diff implements `generate_vbe_metadata` for cpu, such that the function returns the same output for CPU, CUDA and MTIA. To support VBE on CPU with existing fixed-batch-size CPU kernel, we need to recompute offsets, which is previously done in python. This diff implements offsets recomputation in C++ such that all manipulations are done in C++. Note that reshaping offsets and grad_input to work with existing fixed-batch-size CPU kernels are done in Autograd instead of wrapper to avoid multiple computations. VBE CPU tests are in the next diff. Differential Revision: D69162870
spcyppt
force-pushed
the
export-D69162870
branch
2 times, most recently
from
90727e6 to
3a59542
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Summary:
X-link: https://github.com/facebookresearch/FBGEMM/pull/796
This diff implements
generate_vbe_metadatafor cpu, such that the function returns the same output for CPU, CUDA and MTIA.To support VBE on CPU with existing fixed-batch-size CPU kernel, we need to recompute offsets, which is previously done in python. This diff implements offsets recomputation in C++ such that all manipulations are done in C++.
Note that reshaping offsets and grad_input to work with existing fixed-batch-size CPU kernels are done in Autograd instead of wrapper to avoid multiple computations.
VBE CPU tests are in the next diff.
Reviewed By: sryap
Differential Revision: D69162870