[None][feat] KV Cache Connector API #7228
Conversation
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
…t inside py_executor_creator, pass connector manager to kv cache manager Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
Signed-off-by: jthomson04 <[email protected]>
|
PR_Github #16714 [ run ] completed with state |
Signed-off-by: richardhuo-nv <[email protected]>
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 3
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (2)
tensorrt_llm/_torch/pyexecutor/py_executor_creator.py (2)
239-239: Fix logging call (args vs format string).Passing a second positional arg without a %s will raise at runtime.
- logger.info("ATTENTION RUNTIME FEATURES: ", attn_runtime_features) + logger.info("ATTENTION RUNTIME FEATURES: %s", attn_runtime_features)
187-191: Remove redundant LoadFormat import and unify its usage
- At lines 187–188, delete the inner
from tensorrt_llm.llmapi.llm_args import LoadFormat—reuse the top-levelLoadFormatimportspec_config.load_formatis declared asUnion[str, LoadFormat]; comparing it directly to"dummy"may misfire when it’s an enum. Normalize to the enum (e.g.if spec_config.load_format == LoadFormat.DUMMY) or coerce string values into theLoadFormatenum for consistency
♻️ Duplicate comments (1)
tensorrt_llm/_torch/pyexecutor/py_executor_creator.py (1)
386-398: Disallow transceiver + connector early (repeat of prior feedback).Connector is incompatible with KvCacheTransceiver; fail fast.
if executor_config.scheduler_config.capacity_scheduler_policy != CapacitySchedulerPolicy.GUARANTEED_NO_EVICT: raise NotImplementedError( "KV connector is only supported with guaranteed no evict scheduler policy." ) + # Preemptively disallow transceiver + connector + if executor_config.cache_transceiver_config is not None: + raise NotImplementedError( + "KV Cache Connector is not supported together with KvCacheTransceiver." + )
🧹 Nitpick comments (2)
tensorrt_llm/_torch/pyexecutor/py_executor_creator.py (2)
15-17: Optional: prefer module import to preserve namespace.Using direct symbol imports leaks many names; consider importing the module and qualifying usage.
-from tensorrt_llm.bindings.executor import (CapacitySchedulerPolicy, - ContextChunkingPolicy, - ExecutorConfig) +import tensorrt_llm.bindings.executor as executor_bindingsFollow-ups (outside this hunk):
- Replace references: CapacitySchedulerPolicy → executor_bindings.CapacitySchedulerPolicy, ContextChunkingPolicy → executor_bindings.ContextChunkingPolicy, ExecutorConfig → executor_bindings.ExecutorConfig.
386-389: Avoid logging full connector config (may contain sensitive fields).Log module/class identifiers instead.
- logger.info( - f"Initializing kv connector with config: {kv_connector_config}") + logger.info( + "Initializing KV connector: module=%s worker=%s scheduler=%s", + kv_connector_config.connector_module, + kv_connector_config.connector_worker_class, + kv_connector_config.connector_scheduler_class, + )
📜 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 (1)
tensorrt_llm/_torch/pyexecutor/py_executor_creator.py(8 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/py_executor_creator.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/py_executor_creator.py
🧠 Learnings (1)
📚 Learning: 2025-08-26T09:37:10.463Z
Learnt from: jiaganc
PR: NVIDIA/TensorRT-LLM#7031
File: tensorrt_llm/bench/dataclasses/configuration.py:90-104
Timestamp: 2025-08-26T09:37:10.463Z
Learning: In TensorRT-LLM, the `get_pytorch_perf_config()` method returns `self.pytorch_config` which can contain default `cuda_graph_config` values, so `llm_args` may already have this config before the extra options processing.
Applied to files:
tensorrt_llm/_torch/pyexecutor/py_executor_creator.py
🪛 Ruff (0.12.2)
tensorrt_llm/_torch/pyexecutor/py_executor_creator.py
408-408: Line too long (131 > 120)
(E501)
⏰ 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 (10)
tensorrt_llm/_torch/pyexecutor/py_executor_creator.py (10)
3-4: LGTM: required imports for connector initialization.
19-19: LGTM: KvCacheConnectorConfig import.
34-34: LGTM: KvCacheConnectorManager import.
415-421: LGTM: instantiate connector scheduler only on rank 0.This satisfies the manager invariant and avoids cross-rank assertions.
439-445: LGTM: pass KvCacheConnectorManager into KvCacheCreator.Plumbs connector into KV manager build path.
449-449: LGTM: estimation path integrates with connector-aware managers.
485-487: LGTM: avoid activating connector during estimation.Passing None during estimating prevents premature hooks.
503-504: LGTM: finalize manager build post-estimation.
530-531: LGTM: connector attached for the final executor.
215-217: Audit completed: no internal call sites found
Defaultkv_connector_config=Nonepreserves backward compatibility.
|
@Shixiaowei02 could you approve this MR? It's a duplicate of #6488, which was moved to a new branch since John Thompson is no longer at Nvidia. |
|
/bot run |
|
PR_Github #16885 [ run ] triggered by Bot |
|
PR_Github #16885 [ run ] completed with state |
There was a problem hiding this comment.
Actionable comments posted: 3
📜 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/py_executor.py(10 hunks)tensorrt_llm/executor/worker.py(6 hunks)
🚧 Files skipped from review as they are similar to previous changes (1)
- tensorrt_llm/_torch/pyexecutor/py_executor.py
🧰 Additional context used
🧠 Learnings (1)
📚 Learning: 2025-08-26T09:37:10.463Z
Learnt from: jiaganc
PR: NVIDIA/TensorRT-LLM#7031
File: tensorrt_llm/bench/dataclasses/configuration.py:90-104
Timestamp: 2025-08-26T09:37:10.463Z
Learning: In TensorRT-LLM, the `get_pytorch_perf_config()` method returns `self.pytorch_config` which can contain default `cuda_graph_config` values, so `llm_args` may already have this config before the extra options processing.
Applied to files:
tensorrt_llm/executor/worker.py
🧬 Code graph analysis (1)
tensorrt_llm/executor/worker.py (1)
tensorrt_llm/llmapi/llm_args.py (1)
KvCacheConnectorConfig(407-419)
⏰ 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 (3)
tensorrt_llm/executor/worker.py (3)
21-21: Consolidated import looks good.Single import of KvCacheConnectorConfig, PybindMirror, TorchLlmArgs resolves the prior duplication.
135-135: create_py_executor signature includes kv_connector_config
The functioncreate_py_executorintensorrt_llm/_torch/pyexecutor/py_executor_creator.pydefines thekv_connector_configparameter, so supplying this kwarg is supported.
689-689: No action required:kv_connector_configalready defaults toNone, preserving backward compatibility.
|
/bot skip --comment "CI passes before automatic merge of main" |
|
/bot reuse-pipeline |
|
PR_Github #16920 [ reuse-pipeline ] triggered by Bot |
|
PR_Github #16920 [ reuse-pipeline ] completed with state |
|
/bot skip --comment "Getting ridiculous..." |
|
PR_Github #16928 [ skip ] triggered by Bot |
|
PR_Github #16928 [ skip ] completed with state |
|
/bot skip --comment "Previously passed" |
|
PR_Github #16932 [ skip ] triggered by Bot |
|
PR_Github #16932 [ skip ] completed with state |
juney-nvidia
removed
the
Community want to contribute
Signed-off-by: jthomson04 <[email protected]> Signed-off-by: richardhuo-nv <[email protected]> Co-authored-by: jthomson04 <[email protected]> Co-authored-by: Iman Tabrizian <[email protected]> Co-authored-by: Sharan Chetlur <[email protected]>
|
@richardhuo-nv This document is restricted. Could you make it publicly readable? |
I don't own this doc either. Let me try to generate a snapshot of the doc. |
https://drive.google.com/file/d/1L_Dave6C1ullxJNifmBBzw6kjWMvKO54/view?usp=sharing |
Summary by CodeRabbit
New Features
Documentation
Tests
Description
Supersedes #6488. Rebased onto latest main.
This is a first implementation of a KV connector API in TRTLLM. See https://docs.google.com/document/d/104jQ_Qhx5-rb1d-8ZBSN-jYQESaRjNJVhJoT3X7_KGw/edit?tab=t.0#heading=h.dh5skq4lgud6 for details. This is a very limited initial implementation, so a couple caveats:
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.
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.