[None][feat] add an example of KV cache host offloading #7767
Conversation
Signed-off-by: junq <22017000+QiJune@users.noreply.github.com>
Signed-off-by: junq <22017000+QiJune@users.noreply.github.com>
Signed-off-by: junq <22017000+QiJune@users.noreply.github.com>
📝 WalkthroughWalkthroughAdds a new example script demonstrating KV cache host offloading behavior and updates documentation to link to it. No production code or public API changes. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
actor User
participant CLI as CLI/Script
participant Builder as LLM Builder
participant KV as KV Cache Manager
participant Gen as Generator
User->>CLI: Run llm_kv_cache_offloading.py [--enable_offloading]
CLI->>Builder: Build LLM with KvCacheConfig<br/>GPU cache=256 tokens, page_size=16,<br/>host cache=1 GiB if enabled else 0
Builder->>KV: Initialize caches (GPU[, Host])
rect rgb(235, 245, 255)
note right of CLI: First Round
CLI->>Gen: generate(Prompt A, max_tokens=max_seq_len)
Gen->>KV: Allocate/use pages
KV-->>Gen: Cache state updated
CLI->>Gen: generate(Prompt B)
Gen->>KV: Allocate/evict/reuse pages
end
rect rgb(240, 255, 240)
note right of CLI: Second Round
CLI->>Gen: generate(Prompt A)
alt Offloading enabled
KV-->>Gen: Reuse via host-backed pages (hits)
else Offloading disabled
KV-->>Gen: Miss → reallocate (evicted)
end
CLI->>Gen: generate(Prompt B)
Gen->>KV: Similar reuse/evict behavior
end
CLI-->>User: Print prompts, outputs, and cache reuse cues
CLI->>Builder: shutdown()
Estimated code review effort🎯 2 (Simple) | ⏱️ ~10 minutes Pre-merge checks and finishing touches❌ Failed checks (2 warnings)
✅ Passed checks (1 passed)
✨ Finishing touches
🧪 Generate unit tests
Tip 👮 Agentic pre-merge checks are now available in preview!Pro plan users can now enable pre-merge checks in their settings to enforce checklists before merging PRs.
Please see the documentation for more information. Example: reviews:
pre_merge_checks:
custom_checks:
- name: "Undocumented Breaking Changes"
mode: "warning"
instructions: |
Pass/fail criteria: All breaking changes to public APIs, CLI flags, environment variables, configuration keys, database schemas, or HTTP/GraphQL endpoints must be documented in the "Breaking Change" section of the PR description and in CHANGELOG.md. Exclude purely internal or private changes (e.g., code not exported from package entry points or explicitly marked as internal).Please share your feedback with us on this Discord post. 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: 2
🧹 Nitpick comments (6)
examples/llm-api/llm_kv_cache_offloading.py (6)
49-51: Follow import guideline: keep module namespaces.Avoid importing classes directly; import modules/submodules and reference via namespace.
Apply:
-import argparse - -from tensorrt_llm import LLM, SamplingParams -from tensorrt_llm.llmapi import KvCacheConfig +import argparse +import tensorrt_llm as trtllm +from tensorrt_llm import llmapiAnd update usages below:
- sampling_params = SamplingParams(max_tokens=max_seq_len) + sampling_params = trtllm.SamplingParams(max_tokens=max_seq_len) @@ - llm = LLM( + llm = trtllm.LLM( @@ - kv_cache_config=KvCacheConfig( + kv_cache_config=llmapi.KvCacheConfig(
53-60: Add a concise docstring and type hints to main().Improves clarity and aligns with docstring guidelines.
-def main(args): +def main(args: argparse.Namespace) -> None: + """Run the KV cache offloading demo. + Args: + args: Parsed CLI arguments (see --enable_offloading, --model). + """
66-75: Name alignment: “page size” vs tokens_per_block.Use a name aligned with the API to avoid confusion.
- kv_cache_page_size = 16 + kv_cache_tokens_per_block = 16 @@ - tokens_per_block=kv_cache_page_size, + tokens_per_block=kv_cache_tokens_per_block,
79-87: Make the model configurable (8B default is heavy).Allow overriding via CLI; keeps the example runnable on modest GPUs.
- llm = trtllm.LLM( - model="Qwen/Qwen3-8B", + llm = trtllm.LLM( + model=args.model, max_batch_size=max_batch_size, max_seq_len=max_seq_len, kv_cache_config=llmapi.KvCacheConfig(And extend the CLI:
parser = argparse.ArgumentParser( @@ parser.add_argument('--enable_offloading', action='store_true', help='Enable host RAM for KV cache offloading.') + parser.add_argument( + '--model', + default='Qwen/Qwen3-8B', + help='HF model identifier to load (use a smaller one for quick runs).')
91-117: Minor: consider clearer logs for reuse/eviction cues.Add brief hints to correlate with DEBUG logs.
- print("--- First Round ---") + print("--- First Round --- (expect A cached on GPU; B causes A to evict/offload)") @@ - print("\n--- Second Round ---") + print("\n--- Second Round --- (A then B to observe reuse from host if offloading)")
76-77: SamplingParams.max_tokens is "new tokens" — don't set it to max_seq_lenSamplingParams.max_tokens is the number of tokens to generate; setting it to max_seq_len can exceed the engine limit once prompt tokens are included. File: examples/llm-api/llm_kv_cache_offloading.py:76-77.
- Use a conservative or deduced value instead (e.g. SamplingParams(max_tokens=min(128, max_seq_len // 2)) or compute available_new = max_seq_len - prompt_len and use that).
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (2)
docs/source/features/kvcache.md(1 hunks)examples/llm-api/llm_kv_cache_offloading.py(1 hunks)
🧰 Additional context used
📓 Path-based instructions (3)
**/*.{h,hpp,hh,hxx,cpp,cxx,cc,cu,cuh,py}
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
Use only spaces, no tabs; indent with 4 spaces.
Files:
examples/llm-api/llm_kv_cache_offloading.py
**/*.py
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
**/*.py: Python code must target Python 3.8+.
Indent Python code with 4 spaces; do not use tabs.
Maintain module namespace when importing; prefer 'from package.subpackage import foo' then 'foo.SomeClass()' instead of importing the class directly.
Python filenames should be snake_case (e.g., some_file.py).
Python classes use PascalCase names.
Functions and methods use snake_case names.
Local variables use snake_case; prefix 'k' for variables that start with a number (e.g., k_99th_percentile).
Global variables use upper SNAKE_CASE prefixed with 'G' (e.g., G_MY_GLOBAL).
Constants use upper SNAKE_CASE (e.g., MY_CONSTANT).
Avoid shadowing variables from an outer scope.
Initialize all externally visible members of a class in the constructor.
Prefer docstrings for interfaces that may be used outside a file; comments for in-function or file-local interfaces.
Use Google-style docstrings for classes and functions (Sphinx-parsable).
Document attributes and variables inline so they render under the class/function docstring.
Avoid reflection when a simpler, explicit approach suffices (e.g., avoid dict(**locals()) patterns).
In try/except, catch the most specific exceptions possible.
For duck-typing try/except, keep the try body minimal and use else for the main logic.
Files:
examples/llm-api/llm_kv_cache_offloading.py
**/*.{cpp,cxx,cc,h,hpp,hh,hxx,cu,cuh,py}
📄 CodeRabbit inference engine (CODING_GUIDELINES.md)
Prepend the NVIDIA Apache-2.0 copyright header with current year to the top of all source files (e.g., .cpp, .h, .cu, .py).
Files:
examples/llm-api/llm_kv_cache_offloading.py
⏰ 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
|
I get one error message when I run this example. It does not appear related to your example code. The error is |
|
/bot skip --comment "example and doc change" |
|
PR_Github #18852 [ skip ] triggered by Bot |
|
PR_Github #18852 [ skip ] completed with state |
Signed-off-by: junq <22017000+QiJune@users.noreply.github.com>
Signed-off-by: junq <22017000+QiJune@users.noreply.github.com>
4 participants
Summary by CodeRabbit
New Features
Documentation
Description
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
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.
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.