[https://nvbugs/5485325][fix] Fix illegal memory access when max_nun_token equals to max_seq_len when MTP>=1 #7295
Conversation
📝 WalkthroughWalkthroughCompute a cuda_graph_warmup_block inside _get_token_num_for_estimation and use it to bound num_cache_blocks; defer creation of Changes
Sequence Diagram(s)sequenceDiagram
autonumber
participant App as Application
participant Util as _util._get_token_num_for_estimation
participant Engine as PyTorchModelEngine
Note over Util: Token estimation (updated)
App->>Util: _get_token_num_for_estimation(config)
Util->>Util: compute base num_cache_blocks
Util->>Util: compute cuda_graph_warmup_block = (max_seq_len+1)//tokens_per_block + _max_cuda_graph_batch_size - 1
Util->>Util: num_cache_blocks = max(cuda_graph_warmup_block, num_cache_blocks)
Util-->>App: return tokens_per_block * num_cache_blocks * beam_width
Note over Engine: Backend lifecycle & release_batch (updated)
App->>Engine: __init__(config)
alt torch_compile disabled
Engine-->>App: no _torch_compile_backend created
else torch_compile enabled (later)
Engine->>Engine: create _torch_compile_backend
end
App->>Engine: release_batch(batch, bs, draft_len)
alt batch == None (no KV cache space)
Engine-->>App: log info "No KV cache space stop capturing! batch size={bs}, draft_len={draft_len}"
Engine-->>App: return
else has KV cache space
Engine-->>App: proceed with release
end
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Suggested reviewers
📜 Recent review detailsConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro 💡 Knowledge Base configuration:
You can enable these sources in your CodeRabbit configuration. 📒 Files selected for processing (2)
🚧 Files skipped from review as they are similar to previous changes (1)
🧰 Additional context used📓 Path-based instructions (2)**/*.py📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
Files:
**/*.{c,cc,cpp,cxx,h,hh,hpp,hxx,cu,cuh,py}📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
Files:
🧠 Learnings (3)📓 Common learnings📚 Learning: 2025-08-14T21:04:50.248ZApplied to files:
📚 Learning: 2025-08-15T06:46:54.897ZApplied to files:
🧬 Code graph analysis (1)tensorrt_llm/_torch/pyexecutor/_util.py (1)
⏰ 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)
✨ 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. CodeRabbit Commands (Invoked using PR/Issue comments)Type Other keywords and placeholders
Status, Documentation and Community
|
yizhang-nv
force-pushed
the
fix-mtp-seq-qe-max-num
branch
from
a4801c3 to
ba431c4
Compare
![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 (2)
tensorrt_llm/_torch/pyexecutor/model_engine.py (1)
782-787: Prefer “continue” and enrich the log for easier troubleshootingReturning here aborts capturing for all remaining batch sizes; continuing lets the loop try smaller batch sizes or other draft lengths that may still fit. Also, include free_blocks/tokens_per_block to make capacity issues actionable in logs.
- if batch is None: - # No KV cache space! - logger.info( - f"No KV cache space stop capturing! batch size={bs}, draft_len={draft_len}" - ) - return + if batch is None: + # No KV cache space for this setting; try next draft_len / smaller bs. + logger.info( + f"No KV cache space; skip capture for this setting. " + f"batch_size={bs}, draft_len={draft_len}, " + f"free_blocks={kv_cache_manager.get_num_free_blocks()}, " + f"tokens_per_block={kv_cache_manager.tokens_per_block}" + ) + continuetensorrt_llm/_torch/pyexecutor/_util.py (1)
10-11: Remove duplicate ModelConfig importModelConfig is imported twice (absolute and relative); keep one to avoid confusion.
-from tensorrt_llm._torch.model_config import ModelConfig @@ -from ..model_config import ModelConfig +from ..model_config import ModelConfigAlso applies to: 20-21
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
💡 Knowledge Base configuration:
- MCP integration is disabled by default for public repositories
- Jira integration is disabled by default for public repositories
- Linear integration is disabled by default for public repositories
You can enable these sources in your CodeRabbit configuration.
📒 Files selected for processing (2)
tensorrt_llm/_torch/pyexecutor/_util.py(1 hunks)tensorrt_llm/_torch/pyexecutor/model_engine.py(1 hunks)
🧰 Additional context used
📓 Path-based instructions (2)
**/*.py
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
**/*.py: Code must target Python 3.8+
Indent Python code with 4 spaces; do not use tabs
Preserve module namespaces when importing; import modules/packages and access members via the module (e.g., from package.subpackage import foo; foo.SomeClass())
Python file names should be snake_case
Python class names should be PascalCase
Python functions/methods and local variables should be snake_case; variables beginning with a number should be prefixed with k_ (e.g., k_99th_percentile)
Global variables should be UPPER_SNAKE_CASE prefixed with G_ (e.g., G_MY_GLOBAL); constants should be UPPER_SNAKE_CASE
Avoid shadowing variables from outer scopes; initialize all externally visible members in init
Prefer docstrings for interfaces used outside a file; comments should be reserved for in-function or file-local interfaces
Use Google-style docstrings for classes and functions; attributes and variables may be documented inline with trailing string literals
Avoid reflection when simpler, explicit code suffices (e.g., avoid dict(**locals()) patterns)
In try/except, catch the narrowest exceptions possible
For duck-typing patterns, keep the try body minimal and move logic to else to avoid masking unrelated failures
Files:
tensorrt_llm/_torch/pyexecutor/_util.pytensorrt_llm/_torch/pyexecutor/model_engine.py
**/*.{c,cc,cpp,cxx,h,hh,hpp,hxx,cu,cuh,py}
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
Prepend the NVIDIA copyright header (current year) to all source files (.cpp, .h, .cu, .py, etc.)
Files:
tensorrt_llm/_torch/pyexecutor/_util.pytensorrt_llm/_torch/pyexecutor/model_engine.py
🧠 Learnings (3)
📓 Common learnings
Learnt from: thorjohnsen
PR: NVIDIA/TensorRT-LLM#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.
📚 Learning: 2025-08-14T21:04:50.248Z
Learnt from: thorjohnsen
PR: NVIDIA/TensorRT-LLM#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/pyexecutor/_util.py
📚 Learning: 2025-08-19T12:45:11.997Z
Learnt from: amitz-nv
PR: NVIDIA/TensorRT-LLM#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
🧬 Code graph analysis (2)
tensorrt_llm/_torch/pyexecutor/_util.py (2)
tensorrt_llm/_torch/attention_backend/trtllm.py (3)
max_seq_len(558-568)max_seq_len(571-575)tokens_per_block(578-582)tensorrt_llm/runtime/generation.py (1)
tokens_per_block(1180-1181)
tensorrt_llm/_torch/pyexecutor/model_engine.py (1)
tensorrt_llm/logger.py (1)
info(137-138)
🪛 Ruff (0.12.2)
tensorrt_llm/_torch/pyexecutor/_util.py
184-184: Local variable cuda_graph_warmup_block is assigned to but never used
Remove assignment to unused variable cuda_graph_warmup_block
(F841)
⏰ 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
yizhang-nv
changed the title
|
/bot run |
|
PR_Github #16681 [ run ] triggered by Bot |
|
PR_Github #16681 [ run ] completed with state |
Signed-off-by: Yi Zhang <[email protected]>
Signed-off-by: Yi Zhang <[email protected]>
yizhang-nv
force-pushed
the
fix-mtp-seq-qe-max-num
branch
from
9ed3158 to
7ca749d
Compare
|
/bot run |
|
PR_Github #16779 [ run ] triggered by Bot |
yizhang-nv
changed the title
|
PR_Github #16779 [ run ] completed with state |
|
Fixed by #7999 |
2 participants
Description
Reserve larger kv cache during warm up stage to properly allocate cuda graph
Test Coverage
PR Checklist
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.
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.
Summary by CodeRabbit
Refactor
Chores