[TRTLLM-9962][feat] Some optimizations for two-model spec dec #10208
Conversation
|
/bot run |
📝 WalkthroughWalkthroughChanges remove PyTorch compile decorators from speculative execution methods, introduce dynamic extra decoding steps calculation during warmup, make drafting loops conditional based on max_draft_len, and optimize draft token gathering through vectorization instead of per-request loops. Changes
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Pre-merge checks and finishing touches❌ Failed checks (2 warnings)
✅ Passed checks (1 passed)
✨ Finishing touches
🧪 Generate unit tests (beta)
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. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
tensorrt_llm/_torch/pyexecutor/model_engine.py (1)
867-876: Extra decoding steps handling in warmup looks sound; note the wrapper assertionThe new logic in
_create_warmup_requestthat:
- derives
num_extra_decoding_stepsvia_get_num_extra_decoding_steps(),- inflates
blocks_to_usebynum_gen_tokens * blocks_per_gen, and- passes
num_extra_decoding_stepsintoadd_dummy_requestsfor generation warmupcorrectly mirrors the additional KV usage of fused drafting loops (CDL) while remaining a no‑op for non‑wrapped models (where
_get_num_extra_decoding_stepsreturns 0). This should make both torch.compile and CUDA‑graph warmups substantially more faithful to the actual runtime behavior.One nuance to keep in mind is the assert inside
_get_num_extra_decoding_steps:if isinstance(self.model, BaseDraftingLoopWrapper): return self.model.max_total_draft_tokens else: assert not self.model_is_wrapped, ( f"Please add logic to determine num_extra_decoding_steps for drafting loop {type(self.model)}" ) return 0This is good for catching unsupported wrapped models early, but it does mean that any future drafting-loop wrapper types must either subclass
BaseDraftingLoopWrapperor extend_get_num_extra_decoding_steps, otherwise warmup will assert. If you anticipate additional wrapper types, it may be worth documenting this or relaxing the assertion to a logged warning withnum_extra_decoding_steps = 0.Also applies to: 898-905
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (4)
tensorrt_llm/_torch/pyexecutor/model_engine.pytensorrt_llm/_torch/pyexecutor/py_executor.pytensorrt_llm/_torch/speculative/drafting_loops.pytensorrt_llm/_torch/speculative/model_drafter.py
💤 Files with no reviewable changes (1)
- tensorrt_llm/_torch/pyexecutor/py_executor.py
🧰 Additional context used
📓 Path-based instructions (2)
**/*.py
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
**/*.py: Code developed for TensorRT-LLM should conform to Python 3.8+
Indent Python code with 4 spaces. Do not use tabs
Always maintain the namespace when importing in Python, even if only one class or function from a module is used
Python files should use snake_case naming:some_file.py
Python classes should use PascalCase naming:class SomeClass
Python functions and methods should use snake_case naming:def my_awesome_function():
Python local variables should use snake_case naming:my_variable = ...
Python variable names that start with a number should be prefixed with 'k':k_99th_percentile = ...
Python global variables should use upper snake_case with prefix 'G':G_MY_GLOBAL = ...
Python constants should use upper snake_case naming:MY_CONSTANT = ...
Avoid shadowing variables declared in an outer scope in Python
Initialize all externally visible members of a Python class in the constructor
For Python interfaces that may be used outside a file, prefer docstrings over comments
Python comments should be reserved for code within a function, or interfaces that are local to a file
Use Google style docstrings in Python for classes and functions, which can be parsed by Sphinx
Python attributes and variables can be documented inline with type and description
Avoid using reflection in Python when functionality can be easily achieved without reflection
When using try-except blocks in Python, limit the except to the smallest set of errors possible
When using try-except blocks in Python to handle multiple possible variable types (duck-typing), keep the body of the try as small as possible, using the else block for logic
Files:
tensorrt_llm/_torch/speculative/model_drafter.pytensorrt_llm/_torch/speculative/drafting_loops.pytensorrt_llm/_torch/pyexecutor/model_engine.py
**/*.{cpp,h,cu,cuh,py}
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
All TensorRT-LLM Open Source Software code should contain an NVIDIA copyright header that includes the year of its latest meaningful modification
Files:
tensorrt_llm/_torch/speculative/model_drafter.pytensorrt_llm/_torch/speculative/drafting_loops.pytensorrt_llm/_torch/pyexecutor/model_engine.py
🧠 Learnings (7)
📚 Learning: 2025-12-12T03:27:18.859Z
Learnt from: tongyuantongyu
Repo: NVIDIA/TensorRT-LLM PR: 9655
File: tensorrt_llm/_torch/pyexecutor/sampler.py:3031-3031
Timestamp: 2025-12-12T03:27:18.859Z
Learning: In tensorrt_llm/_torch/pyexecutor/sampler.py, when reviewing code that iterates through requests, ensure it does not convert excessive data into Python lists. Instead, the code should use torch.gather or indexing to gather only the data that will be used in the for loop before converting to Python lists. This minimizes data movement and improves performance.
Applied to files:
tensorrt_llm/_torch/speculative/model_drafter.py
📚 Learning: 2025-12-12T03:27:18.859Z
Learnt from: tongyuantongyu
Repo: NVIDIA/TensorRT-LLM PR: 9655
File: tensorrt_llm/_torch/pyexecutor/sampler.py:3031-3031
Timestamp: 2025-12-12T03:27:18.859Z
Learning: In tensorrt_llm/_torch/pyexecutor/sampler.py, when reviewing code that iterates through requests, ensure it does not access torch.Tensor objects (CPU or GPU) inside the loop. Instead, the code should use .tolist() to convert batched data tensors to Python lists beforehand, and then access the list in the for loop. This is a critical performance consideration.
Applied to files:
tensorrt_llm/_torch/speculative/model_drafter.py
📚 Learning: 2025-08-14T21:04:50.248Z
Learnt from: thorjohnsen
Repo: NVIDIA/TensorRT-LLM PR: 6910
File: cpp/tensorrt_llm/batch_manager/kvCacheManager.cpp:0-0
Timestamp: 2025-08-14T21:04:50.248Z
Learning: In KV cache onboarding logic during prefill in cpp/tensorrt_llm/batch_manager/kvCacheManager.cpp, when calculating which blocks fall within the attention window, use getTokensPerBlock() to advance token indices rather than block->getUniqueTokens().size(), because the calculation needs to consider the post-prefill state where blocks will be filled to capacity, not their current token count.
Applied to files:
tensorrt_llm/_torch/speculative/model_drafter.pytensorrt_llm/_torch/pyexecutor/model_engine.py
📚 Learning: 2025-08-19T12:45:11.997Z
Learnt from: amitz-nv
Repo: NVIDIA/TensorRT-LLM PR: 7033
File: tensorrt_llm/_torch/pyexecutor/model_engine.py:0-0
Timestamp: 2025-08-19T12:45:11.997Z
Learning: In tensorrt_llm/_torch/pyexecutor/model_engine.py, DoRA (Delta Orthogonal Rank Adaptation) functionality was removed from the PyTorch flow to eliminate issues with inverted DoRA detection logic. The original is_dora condition was checking if scaling_vec_pointer == 0, which was potentially incorrect.
Applied to files:
tensorrt_llm/_torch/pyexecutor/model_engine.py
📚 Learning: 2025-08-19T12:45:35.429Z
Learnt from: amitz-nv
Repo: NVIDIA/TensorRT-LLM PR: 7033
File: tensorrt_llm/_torch/pyexecutor/model_engine.py:2086-2092
Timestamp: 2025-08-19T12:45:35.429Z
Learning: DoRA (Delta Orthogonal Rank Adaptation) functionality has been removed from the PyTorch flow in tensorrt_llm/_torch/pyexecutor/model_engine.py. The is_dora field is computed but not used downstream in the PyTorch flow, so converting it to a tensor would be wasteful overhead.
Applied to files:
tensorrt_llm/_torch/pyexecutor/model_engine.py
📚 Learning: 2025-08-26T06:07:02.166Z
Learnt from: shaharmor98
Repo: NVIDIA/TensorRT-LLM PR: 7231
File: tensorrt_llm/_torch/pyexecutor/_util.py:504-509
Timestamp: 2025-08-26T06:07:02.166Z
Learning: In tensorrt_llm/_torch/pyexecutor/_util.py, when calling model_engine.set_lora_model_config(), pass model_binding_config.mlp_hidden_size directly without multiplying by mapping.tp_size, as the mlp_hidden_size from get_bindings_model_config() is already the per-TP rank value needed for LoRA weight packaging.
Applied to files:
tensorrt_llm/_torch/pyexecutor/model_engine.py
📚 Learning: 2025-12-12T03:27:08.565Z
Learnt from: tongyuantongyu
Repo: NVIDIA/TensorRT-LLM PR: 9655
File: tensorrt_llm/_torch/pyexecutor/sampler.py:3031-3031
Timestamp: 2025-12-12T03:27:08.565Z
Learning: In files under tensorrt_llm/_torch/pyexecutor, avoid accessing torch.Tensor objects inside for-loops when iterating over requests. Convert batched tensors to Python lists beforehand using tensor.tolist(), and then iterate over those lists. This improves performance by reducing tensor-bound operations inside hot loops. Apply this pattern to similar code paths that process batches to access simple Python data structures (lists) inside loops.
Applied to files:
tensorrt_llm/_torch/pyexecutor/model_engine.py
🧬 Code graph analysis (3)
tensorrt_llm/_torch/speculative/model_drafter.py (1)
tensorrt_llm/functional.py (1)
index_select(2216-2275)
tensorrt_llm/_torch/speculative/drafting_loops.py (2)
tensorrt_llm/_torch/speculative/interface.py (1)
is_eagle3(61-62)tensorrt_llm/_torch/speculative/eagle3.py (1)
Eagle3SpecMetadata(114-276)
tensorrt_llm/_torch/pyexecutor/model_engine.py (1)
tensorrt_llm/_torch/attention_backend/trtllm.py (3)
tokens_per_block(712-716)max_seq_len(692-702)max_seq_len(705-709)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (1)
- GitHub Check: Pre-commit Check
🔇 Additional comments (1)
tensorrt_llm/_torch/speculative/drafting_loops.py (1)
121-144: Guarding drafting loop onmax_draft_len > 1looks correctThe new
if self.max_draft_len > 1:guard and cachedis_eagle3flag preserve behavior while avoiding unnecessaryprepare_for_generation/save_metadata_statework when only a single draft token is requested. The Eagle3 index update on the first extra iteration is still applied only when additional iterations actually run, which is consistent with the previous semantics.
|
PR_Github #29478 [ run ] triggered by Bot. Commit: |
|
PR_Github #29478 [ run ] completed with state
|
|
/bot run |
|
PR_Github #29483 [ run ] triggered by Bot. Commit: |
|
PR_Github #29483 [ run ] completed with state
|
|
/bot run |
|
PR_Github #29564 [ run ] triggered by Bot. Commit: |
|
PR_Github #29564 [ run ] completed with state
|
ziyixiong-nv
force-pushed
the
dev-fxiong-trtllm9962
branch
from
8bc701f to
8170aa0
Compare
|
/bot run |
|
PR_Github #29717 [ run ] triggered by Bot. Commit: |
|
PR_Github #29717 [ run ] completed with state
|
|
/bot run |
|
PR_Github #29815 [ run ] triggered by Bot. Commit: |
|
PR_Github #29815 [ run ] completed with state
|
ziyixiong-nv
force-pushed
the
dev-fxiong-trtllm9962
branch
from
8170aa0 to
e159a09
Compare
|
/bot run |
|
PR_Github #29851 [ run ] triggered by Bot. Commit: |
ziyixiong-nv
force-pushed
the
dev-fxiong-trtllm9962
branch
from
21fa197 to
ea482c6
Compare
|
/bot run |
|
PR_Github #29885 [ run ] triggered by Bot. Commit: |
|
PR_Github #29885 [ run ] completed with state
|
Signed-off-by: ziyixiong-nv <[email protected]>
Signed-off-by: ziyixiong-nv <[email protected]>
Signed-off-by: ziyixiong-nv <[email protected]>
ziyixiong-nv
force-pushed
the
dev-fxiong-trtllm9962
branch
from
ea482c6 to
7d6632b
Compare
|
/bot run |
|
PR_Github #30011 [ run ] triggered by Bot. Commit: |
|
PR_Github #30011 [ run ] completed with state |
5 participants
Summary by CodeRabbit
Release Notes
✏️ Tip: You can customize this high-level summary in your review settings.
Description
This PR applies several optimizations for two-model spec dec.
Test Coverage
PR Checklist
Please review the following before submitting your PR:
PR description clearly explains what and why. If using CodeRabbit's summary, please make sure it makes sense.
PR Follows TRT-LLM CODING GUIDELINES to the best of your knowledge.
Test cases are provided for new code paths (see test instructions)
Any new dependencies have been scanned for license and vulnerabilities
CODEOWNERS updated if ownership changes
Documentation updated as needed
Update tava architecture diagram if there is a significant design change in PR.
The reviewers assigned automatically/manually are appropriate for the PR.
Please check this after reviewing the above items as appropriate for this PR.
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.