gemini-code-assist[bot] commented on PR #18844: URL: https://github.com/apache/tvm/pull/18844#issuecomment-3977052004
## Summary of Changes Hello, I'm Gemini Code Assist[^1]! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request significantly refactors the TVM codebase by eliminating several layers of legacy indirection within the node and runtime layers. The changes focus on removing outdated header files, reorganizing core components for better logical placement, and modernizing serialization processes to directly utilize the Foreign Function Interface (FFI). This cleanup simplifies the architecture, reduces redundant abstractions, and aligns the system with more direct and efficient FFI interactions, ultimately improving maintainability and clarity. ### Highlights * **Removed Legacy Packed Function Header**: The `include/tvm/runtime/packed_func.h` header and its associated includes have been completely removed, streamlining the runtime layer. * **Relocated Attribute Registry**: The `src/node/attr_registry.h` file has been moved to `src/ir/attr_registry.h`, improving the logical organization of core IR components. * **Modernized Serialization**: The legacy `tvm::SaveJSON` and `tvm::LoadJSON` functions have been replaced with direct calls to `ffi::ToJSONGraph` and `ffi::FromJSONGraph`, and their forwarding headers removed, simplifying the serialization mechanism. * **Streamlined Node Creation**: The C++ `node.MakeNode` function has been deprecated, and the Python `make_node` function now directly redirects to `ffi.MakeObjectFromPackedArgs`, reducing indirection in object instantiation. 🧠**New Feature in Public Preview:** You can now enable **Memory** to help **Gemini Code Assist** learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. **Click [here](https://codeassist.google/code-review/login) to enable Memory in your admin console.** <details> <summary><b>Changelog</b></summary> * **include/tvm/ir/serialization.h** * Removed the serialization header file. * **include/tvm/node/serialization.h** * Removed the forwarding serialization header file. * **include/tvm/runtime/packed_func.h** * Removed the packed function header file. * **python/tvm/ir/attrs.py** * Updated imports to use `_tvm_ffi_api` directly. * Modified the `make_node` function to call `_tvm_ffi_api.MakeObjectFromPackedArgs` directly, including special handling for `ir.DictAttrs`. * **src/contrib/msc/core/ir/plugin.h** * Updated the include path for `attr_registry.h`. * **src/ir/attr_registry.h** * Renamed from `src/node/attr_registry.h`. * Updated internal header guards and file comments to reflect the new path. * **src/ir/op.cc** * Updated the include path for `attr_registry.h`. * **src/ir/serialization.cc** * Modified `SaveJSON` and `LoadJSON` functions to be `static`. * Updated the parameter types of `SaveJSON` and `LoadJSON` to `ffi::Any`. * **src/node/attr_registry.h** * Renamed to `src/ir/attr_registry.h`. * Updated internal header guards and file comments to reflect the new path. * **src/node/reflection.cc** * Removed the reflection utility source file. * **src/relax/backend/adreno/annotate_custom_storage.cc** * Removed the include for `tvm/ir/serialization.h`. * **src/relax/backend/adreno/fold_vdevice_scope_change.cc** * Removed the include for `tvm/ir/serialization.h`. * **src/relax/transform/alter_op_impl.cc** * Replaced the `tvm/ir/serialization.h` include with `tvm/ffi/extra/serialization.h`. * Updated `DeepCopyIndexMap` to use direct FFI serialization calls (`ffi::ToJSONGraph` and `ffi::FromJSONGraph`). * **src/relax/transform/convert_layout.cc** * Replaced the `tvm/ir/serialization.h` include with `tvm/ffi/extra/serialization.h`. * Updated `LayoutConvertMutator` to use direct FFI serialization calls. * **src/relax/transform/specialize_primfunc_based_on_callsite.cc** * Removed the include for `tvm/ir/serialization.h`. * **src/runtime/contrib/cutlass/fp16_group_gemm_sm100.cu** * Removed the include for `tvm/runtime/packed_func.h`. * **src/runtime/contrib/cutlass/fp8_groupwise_scaled_gemm_sm100.cu** * Removed the include for `tvm/runtime/packed_func.h`. * **src/runtime/contrib/cutlass/fp8_groupwise_scaled_gemm_sm90.cu** * Removed the include for `tvm/runtime/packed_func.h`. * **src/runtime/contrib/cutlass/fp8_groupwise_scaled_group_gemm_sm100.cu** * Removed the include for `tvm/runtime/packed_func.h`. * **src/runtime/hexagon/rpc/simulator/rpc_server.cc** * Removed the include for `tvm/runtime/packed_func.h`. * **src/s_tir/meta_schedule/database/database.cc** * Updated `Workload::AsJSON` and `Workload::FromJSON` to use direct FFI JSON serialization and deserialization. * **src/s_tir/meta_schedule/database/database_utils.cc** * Updated `JSONDumps` to use direct FFI JSON serialization. * **src/s_tir/meta_schedule/mutator/mutate_compute_location.cc** * Updated `InitializeWithTuneContext` and `FindCandidates` to use direct FFI JSON serialization and deserialization. * **src/s_tir/meta_schedule/mutator/mutate_parallel.cc** * Updated `InitializeWithTuneContext` and `Apply` to use direct FFI JSON serialization and deserialization. * **src/s_tir/meta_schedule/mutator/mutate_thread_binding.cc** * Updated `InitializeWithTuneContext` to use direct FFI JSON serialization. * **src/s_tir/meta_schedule/utils.h** * Replaced `tvm/ir/serialization.h` include with `tvm/ffi/extra/json.h` and `tvm/ffi/extra/serialization.h`. * Updated `DeepCopyIRModule` to use direct FFI serialization. * **src/s_tir/schedule/primitive/blockize_tensorize.cc** * Updated the `DeepCopy` function to use direct FFI serialization. * **src/s_tir/schedule/primitive/layout_transformation.cc** * Updated `AttrsAsJSON` and `AttrsFromJSON` to use direct FFI JSON serialization and deserialization for layout transformation attributes. * **src/s_tir/schedule/trace.cc** * Updated `TranslateInputRVs` to use direct FFI JSON deserialization. * **src/s_tir/schedule/utils.h** * Replaced `tvm/ir/serialization.h` include with `tvm/ffi/extra/json.h` and `tvm/ffi/extra/serialization.h`. * Updated the include path for `attr_registry.h`. * **src/script/printer/utils.h** * Replaced `tvm/ir/serialization.h` include with `tvm/ffi/extra/json.h` and `tvm/ffi/extra/serialization.h`. * Updated `Docsify` to use direct FFI JSON serialization. * **src/target/tag.cc** * Updated the include path for `attr_registry.h`. * **src/target/target_kind.cc** * Updated the include path for `attr_registry.h`. </details> <details> <summary><b>Activity</b></summary> * The author provided a test plan indicating that all 16 tests in `tests/python/ir/` are expected to pass. * The author provided a test plan indicating that 349 tests in `tests/python/tir-base/` are expected to pass, with 2 skipped. * The author expects a full clean build to complete without any errors. </details> <details> <summary><b>Using Gemini Code Assist</b></summary> <br> The full guide for Gemini Code Assist can be found on our [documentation page](https://developers.google.com/gemini-code-assist/docs/review-github-code), here are some quick tips. <b>Invoking Gemini</b> You can request assistance from Gemini at any point by creating a comment using either `/gemini <command>` or `@gemini-code-assist <command>`. Below is a summary of the supported commands on the current page. Feature | Command | Description --- | --- | --- Code Review | `/gemini review` | Performs a code review for the current pull request in its current state. Pull Request Summary | `/gemini summary` | Provides a summary of the current pull request in its current state. Comment | @gemini-code-assist | Responds in comments when explicitly tagged, both in pull request comments and review comments. Help | `/gemini help` | Displays a list of available commands. <b>Customization</b> To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a `.gemini/` folder in the base of the repository. Detailed instructions can be found [here](https://developers.google.com/gemini-code-assist/docs/customize-gemini-behavior-github). <b>Limitations & Feedback</b> Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with :thumbsup: and :thumbsdown: on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up [here](https://google.qualtrics.com/jfe/form/SV_2cyuGuTWsEw84yG). <b>You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the [Gemini Code Assist IDE Extension](https://cloud.google.com/products/gemini/code-assist).</b> </details> [^1]: Review the [Privacy Notices](https://policies.google.com/privacy), [Generative AI Prohibited Use Policy](https://policies.google.com/terms/generative-ai/use-policy), [Terms of Service](https://policies.google.com/terms), and learn how to configure Gemini Code Assist in GitHub [here](https://developers.google.com/gemini-code-assist/docs/customize-gemini-behavior-github). Gemini can make mistakes, so double check it and [use code with caution](https://support.google.com/legal/answer/13505487). -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
