-
-
Notifications
You must be signed in to change notification settings - Fork 8.7k
[Compile][Platform] Make PiecewiseBackend pluggable and extendable #18076
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Merged
+305
−200
Merged
Changes from all commits
Commits
File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
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
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
Original file line number | Diff line number | Diff line change |
---|---|---|
@@ -0,0 +1,71 @@ | ||
# SPDX-License-Identifier: Apache-2.0 | ||
|
||
from typing import Any, Callable, Protocol | ||
|
||
import torch.fx as fx | ||
|
||
from vllm.compilation.backends import VllmBackend | ||
from vllm.config import VllmConfig | ||
|
||
|
||
class AbstractPiecewiseBackend(Protocol): | ||
""" | ||
PiecewiseBackend interface that allows platforms to extend | ||
piecewise static graph. | ||
""" | ||
|
||
def __init__(self, graph: fx.GraphModule, vllm_config: VllmConfig, | ||
graph_pool: Any, piecewise_compile_index: int, | ||
total_piecewise_compiles: int, sym_shape_indices: list[int], | ||
compiled_graph_for_general_shape: Callable, | ||
vllm_backend: VllmBackend, **kwargs): | ||
""" | ||
Initializes the PiecewiseBackend class with compilation and | ||
execution-related configurations. | ||
This class handles piecewise compilation, graph capturing, | ||
and dispatching for specific input shapes. | ||
Args: | ||
graph (fx.GraphModule): The graph represented in fx. | ||
vllm_config (VllmConfig): Global configuration for vLLM. | ||
graph_pool (Any): | ||
Graph memory pool handle, e.g., | ||
`torch.cuda.graph_pool_handle()`. | ||
piecewise_compile_index (int): | ||
Index of the current piecewise subgraph. | ||
total_piecewise_compiles (int): | ||
Total number of piecewise-compiled graphs. | ||
sym_shape_indices (list[int]): | ||
Indices of symbolic shape. | ||
compiled_graph_for_general_shape (Callable): | ||
Callable that executes the graph compiled for general shapes. | ||
vllm_backend (VllmBackend): | ||
Backend compiler that manages compilation and graph runtime | ||
for vLLM. | ||
Keyword Args: | ||
kwargs: Additional keyword arguments reserved for future | ||
extensions or custom platforms. | ||
""" | ||
raise NotImplementedError | ||
|
||
def __call__(self, *args) -> Any: | ||
"""Executes the compiled graph for given input args. | ||
If this is the first invocation, executes the general compiled graph | ||
and initiates the compilation process tracking. For subsequent calls, | ||
dynamically dispatches execution to either a compiled graph or a static | ||
graph based on the input shape. | ||
Args: | ||
*args: Variable length input arguments to be passed into the | ||
graph. The symbolic shape is expected to be in position | ||
`sym_shape_indices[0]`. | ||
Returns: | ||
Any: Output of the executed graph. This can be from the general | ||
compiled graph, a specialized compiled version for the given shape, | ||
or a replayed static graph. | ||
""" | ||
raise NotImplementedError |
Oops, something went wrong.
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.
Uh oh!
There was an error while loading. Please reload this page.