[TRTLLM-4279] fix: Add a protection test for checking trtllm custom ops #6515
Conversation
…e impl. Signed-off-by: Yang Li <[email protected]>
Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com> Signed-off-by: Yang Li <[email protected]>
Signed-off-by: Yang Li <[email protected]>
📝 WalkthroughWalkthroughThe changes remove the registration of two out-variant matrix multiplication functions from the Torch library fragment in a C++ source file. Additionally, a new Python test suite is introduced to verify the presence of fake kernel implementations for custom PyTorch operators, specifically targeting the "trtllm" namespace. Changes
Sequence Diagram(s)sequenceDiagram
participant Tester as TestCustomOp (Python)
participant Torch as PyTorch Operator Registry
participant TRTLLM as trtllm Custom Ops
Tester->>Torch: Discover custom ops in "trtllm" namespace
loop For each custom op
Tester->>Torch: Check for fake kernel implementation
alt Fake kernel missing and not waived
Tester->>Tester: Mark test as failed
else Fake kernel present or waived
Tester->>Tester: Continue
end
end
Tester->>Tester: Report test results
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~15 minutes Suggested reviewers
Note ⚡️ Unit Test Generation is now available in beta!Learn more here, or try it out under "Finishing Touches" below. 📜 Recent review detailsConfiguration used: .coderabbit.yaml 📒 Files selected for processing (1)
🚧 Files skipped from review as they are similar to previous changes (1)
✨ Finishing Touches
🧪 Generate unit tests
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
SupportNeed help? Create a ticket on our support page for assistance with any issues or questions. Note: Be mindful of the bot's finite context window. It's strongly recommended to break down tasks such as reading entire modules into smaller chunks. For a focused discussion, use review comments to chat about specific files and their changes, instead of using the PR comments. CodeRabbit Commands (Invoked using PR comments)
Other keywords and placeholders
Documentation and Community
|
yali-arch
changed the title
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (3)
tests/unittest/_torch/test_custom_ops.py (3)
4-4: Consider replacing wildcard import with specific imports.The wildcard import from
typingcan make code less readable and may introduce unexpected names into the namespace. Consider importing only the specific types needed.-from typing import * # noqa: F403 +from typing import List, Set
20-84: Well-structured base class with comprehensive operator discovery.The implementation correctly handles the challenges of discovering lazy-loaded C++ custom operators using schemas and provides thorough cleanup of global registries. The separation of concerns between discovery and testing is good design.
Consider adding error handling for cases where
library_utils.lookup_opmight fail:for schema in ops_schemas: if not schema.name.startswith(ns_prefix): continue - op = library_utils.lookup_op(schema.name) - ops.append(op) - print(f" {op._name}") + try: + op = library_utils.lookup_op(schema.name) + ops.append(op) + print(f" {op._name}") + except Exception as e: + print(f" Failed to lookup {schema.name}: {e}")
133-192: Core test effectively addresses the PR objective.The test logic is sound and will prevent new custom operators from being added without fake implementations. The separation of waived operations (legitimately don't need fakes) from those marked for future fixes is appropriate.
The extensive
to_fixlist (44 operators) represents significant technical debt. Consider:
- Adding GitHub issue references for tracking progress
- Adding a mechanism to periodically validate that operators in
to_fixstill exist in the codebase# TODO: add fake impl for these ops in follow-up PRs. + # Tracked in GitHub issues: [add issue numbers] to_fix = {
📜 Review details
Configuration used: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (2)
cpp/tensorrt_llm/thop/cublasScaledMM.cpp(0 hunks)tests/unittest/_torch/test_custom_ops.py(1 hunks)
💤 Files with no reviewable changes (1)
- cpp/tensorrt_llm/thop/cublasScaledMM.cpp
🧰 Additional context used
🧠 Learnings (2)
📓 Common learnings
Learnt from: moraxu
PR: NVIDIA/TensorRT-LLM#6303
File: tests/integration/test_lists/qa/examples_test_list.txt:494-494
Timestamp: 2025-07-28T17:06:08.621Z
Learning: In TensorRT-LLM testing, it's common to have both CLI flow tests (test_cli_flow.py) and PyTorch API tests (test_llm_api_pytorch.py) for the same model. These serve different purposes: CLI flow tests validate the traditional command-line workflow, while PyTorch API tests validate the newer LLM API backend. Both are legitimate and should coexist.
tests/unittest/_torch/test_custom_ops.py (1)
Learnt from: moraxu
PR: #6303
File: tests/integration/test_lists/qa/examples_test_list.txt:494-494
Timestamp: 2025-07-28T17:06:08.621Z
Learning: In TensorRT-LLM testing, it's common to have both CLI flow tests (test_cli_flow.py) and PyTorch API tests (test_llm_api_pytorch.py) for the same model. These serve different purposes: CLI flow tests validate the traditional command-line workflow, while PyTorch API tests validate the newer LLM API backend. Both are legitimate and should coexist.
🔇 Additional comments (4)
tests/unittest/_torch/test_custom_ops.py (4)
14-17: LGTM!The decorator correctly handles Windows incompatibility with
torch.compileusing the appropriate unittest skip mechanism.
87-93: LGTM!The test class is properly structured with appropriate inheritance and decorator usage. The docstring clearly explains the purpose.
95-128: Excellent validation test for the testing infrastructure.This test effectively validates that the fake kernel detection mechanism works correctly by creating a test operator without a fake implementation and confirming it's properly detected. The use of
torch.library.opcheckandlibrary_utils.has_fake_kernelprovides comprehensive validation.
195-196: LGTM!Standard unittest execution pattern.
|
/bot run |
|
PR_Github #13673 [ run ] triggered by Bot |
|
PR_Github #13673 [ run ] completed with state |
|
/bot run |
|
PR_Github #13695 [ run ] triggered by Bot |
|
PR_Github #13695 [ run ] completed with state |
Signed-off-by: Yang Li <[email protected]>
|
/bot run |
|
PR_Github #13755 [ run ] triggered by Bot |
|
PR_Github #13755 [ run ] completed with state |
…ps (NVIDIA#6515) Signed-off-by: Yang Li <[email protected]> Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com> Signed-off-by: Lanyu Liao <[email protected]>
…ps (NVIDIA#6515) Signed-off-by: Yang Li <[email protected]> Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com>
4 participants
Summary by CodeRabbit
Tests
Refactor
Description
It has been observed that the problem of missing fake implementations for new custom operations occurs from time to time, and such issues are often identified only after a certain period.
As a result, a new test file tensorrt_llm/_torch/custom_ops/torch_custom_ops.py is added to check the presence of fake implementations for each custom operation.
There are some CPP custom ops that don't have fake impl. They are tracked in the test but will be fixed in future PRs:
Test Coverage
GitHub Bot Help
/bot [-h] ['run', 'kill', 'skip', 'reuse-pipeline'] ...Provide a user friendly way for developers to interact with a Jenkins server.
Run
/bot [-h|--help]to print this help message.See details below for each supported subcommand.
Details
run [--reuse-test (optional)pipeline-id --disable-fail-fast --skip-test --stage-list "A10-PyTorch-1, xxx" --gpu-type "A30, H100_PCIe" --test-backend "pytorch, cpp" --add-multi-gpu-test --only-multi-gpu-test --disable-multi-gpu-test --post-merge --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx" --detailed-log --debug(experimental)]Launch build/test pipelines. All previously running jobs will be killed.
--reuse-test (optional)pipeline-id(OPTIONAL) : Allow the new pipeline to reuse build artifacts and skip successful test stages from a specified pipeline or the last pipeline if no pipeline-id is indicated. If the Git commit ID has changed, this option will be always ignored. The DEFAULT behavior of the bot is to reuse build artifacts and successful test results from the last pipeline.--disable-reuse-test(OPTIONAL) : Explicitly prevent the pipeline from reusing build artifacts and skipping successful test stages from a previous pipeline. Ensure that all builds and tests are run regardless of previous successes.--disable-fail-fast(OPTIONAL) : Disable fail fast on build/tests/infra failures.--skip-test(OPTIONAL) : Skip all test stages, but still run build stages, package stages and sanity check stages. Note: Does NOT update GitHub check status.--stage-list "A10-PyTorch-1, xxx"(OPTIONAL) : Only run the specified test stages. Examples: "A10-PyTorch-1, xxx". Note: Does NOT update GitHub check status.--gpu-type "A30, H100_PCIe"(OPTIONAL) : Only run the test stages on the specified GPU types. Examples: "A30, H100_PCIe". Note: Does NOT update GitHub check status.--test-backend "pytorch, cpp"(OPTIONAL) : Skip test stages which don't match the specified backends. Only support [pytorch, cpp, tensorrt, triton]. Examples: "pytorch, cpp" (does not run test stages with tensorrt or triton backend). Note: Does NOT update GitHub pipeline status.--only-multi-gpu-test(OPTIONAL) : Only run the multi-GPU tests. Note: Does NOT update GitHub check status.--disable-multi-gpu-test(OPTIONAL) : Disable the multi-GPU tests. Note: Does NOT update GitHub check status.--add-multi-gpu-test(OPTIONAL) : Force run the multi-GPU tests in addition to running L0 pre-merge pipeline.--post-merge(OPTIONAL) : Run the L0 post-merge pipeline instead of the ordinary L0 pre-merge pipeline.--extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx"(OPTIONAL) : Run the ordinary L0 pre-merge pipeline and specified test stages. Examples: --extra-stage "H100_PCIe-TensorRT-Post-Merge-1, xxx".--detailed-log(OPTIONAL) : Enable flushing out all logs to the Jenkins console. This will significantly increase the log volume and may slow down the job.--debug(OPTIONAL) : Experimental feature. Enable access to the CI container for debugging purpose. Note: Specify exactly one stage in thestage-listparameter to access the appropriate container environment. Note: Does NOT update GitHub check status.For guidance on mapping tests to stage names, see
docs/source/reference/ci-overview.mdand the
scripts/test_to_stage_mapping.pyhelper.kill
killKill all running builds associated with pull request.
skip
skip --comment COMMENTSkip testing for latest commit on pull request.
--comment "Reason for skipping build/test"is required. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.reuse-pipeline
reuse-pipelineReuse a previous pipeline to validate current commit. This action will also kill all currently running builds associated with the pull request. IMPORTANT NOTE: This is dangerous since lack of user care and validation can cause top of tree to break.