[AINode] Integrate device manager framework #16998
Open
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
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## master #16998 +/- ##
============================================
+ Coverage 39.28% 39.33% +0.05%
Complexity 212 212
============================================
Files 5073 5079 +6
Lines 339848 340571 +723
Branches 43401 43477 +76
============================================
+ Hits 133495 133976 +481
- Misses 206353 206595 +242 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
Contributor
There was a problem hiding this comment.
Pull request overview
This pull request integrates a device manager framework to support AINode managing models under different process units (CPU, CUDA). The changes introduce a unified backend usage interface through DeviceManager, replace string-based device IDs with torch.device objects throughout the codebase, and update the Thrift API to return device types alongside device IDs.
Key changes:
- New device manager framework with backend adapters for CPU and CUDA
- Refactored device handling from string-based to
torch.deviceobjects - Updated Thrift API to return device type information alongside device IDs
Reviewed changes
Copilot reviewed 22 out of 22 changed files in this pull request and generated 15 comments.
Show a summary per file
| File | Description |
|---|---|
| iotdb-protocol/thrift-ainode/src/main/thrift/ainode.thrift | Changed TShowAIDevicesResp from list to map to include device type information |
| iotdb-core/node-commons/src/main/java/org/apache/iotdb/commons/schema/column/ColumnHeaderConstant.java | Added DEVICE_TYPE column header and updated showAIDevicesColumnHeaders |
| iotdb-core/datanode/src/main/java/org/apache/iotdb/db/queryengine/plan/execution/config/metadata/ai/ShowAIDevicesTask.java | Updated to handle deviceIdMap instead of deviceIdList |
| iotdb-core/ainode/pyproject.toml | Restricted Python version and adjusted numpy/pandas version constraints |
| iotdb-core/ainode/iotdb/ainode/core/util/gpu_mapping.py | Removed (functionality replaced by device manager) |
| iotdb-core/ainode/iotdb/ainode/core/rpc/handler.py | Integrated DeviceManager and updated device validation logic |
| iotdb-core/ainode/iotdb/ainode/core/model/model_loader.py | Updated to use DeviceManager for model device placement |
| iotdb-core/ainode/iotdb/ainode/core/manager/inference_manager.py | Changed to use torch.device instead of string device IDs |
| iotdb-core/ainode/iotdb/ainode/core/manager/device_manager.py | New unified device management interface |
| iotdb-core/ainode/iotdb/ainode/core/inference/pool_scheduler/basic_pool_scheduler.py | Updated device ID types from string to torch.device |
| iotdb-core/ainode/iotdb/ainode/core/inference/pool_scheduler/abstract_pool_scheduler.py | Updated type signatures for torch.device |
| iotdb-core/ainode/iotdb/ainode/core/inference/pool_controller.py | Refactored to use torch.device throughout |
| iotdb-core/ainode/iotdb/ainode/core/inference/pipeline/pipeline_loader.py | Updated device parameter type to torch.device |
| iotdb-core/ainode/iotdb/ainode/core/inference/pipeline/basic_pipeline.py | Integrated DeviceManager for device handling |
| iotdb-core/ainode/iotdb/ainode/core/inference/inference_request_pool.py | Updated device handling with DeviceManager |
| iotdb-core/ainode/iotdb/ainode/core/device/env.py | New distributed environment handling |
| iotdb-core/ainode/iotdb/ainode/core/device/device_utils.py | New device parsing utilities |
| iotdb-core/ainode/iotdb/ainode/core/device/backend/cuda_backend.py | New CUDA backend adapter |
| iotdb-core/ainode/iotdb/ainode/core/device/backend/cpu_backend.py | New CPU backend adapter |
| iotdb-core/ainode/iotdb/ainode/core/device/backend/base.py | New backend adapter protocol definition |
| iotdb-core/ainode/iotdb/ainode/core/device/backend/init.py | New package initialization |
| iotdb-core/ainode/iotdb/ainode/core/device/init.py | New package initialization |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
iotdb-core/ainode/iotdb/ainode/core/device/backend/cpu_backend.py
Outdated
Show resolved
Hide resolved
iotdb-core/ainode/iotdb/ainode/core/inference/pool_scheduler/basic_pool_scheduler.py
Show resolved
Hide resolved
iotdb-core/ainode/iotdb/ainode/core/inference/pool_controller.py
Outdated
Show resolved
Hide resolved
iotdb-core/ainode/iotdb/ainode/core/device/backend/cuda_backend.py
Outdated
Show resolved
Hide resolved
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
…o aidevicemanager
|
Contributor
|
LGTM |
Contributor
|
LGTM~ |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 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.
To support AINode managing models under different process units, we introduce device manager in this PR. We provide unified backend usage interfaces, which can be found at
iotdb/ainode/core/manager/device_manager.py.