[None][test] restrict max_num_tokens in disagg mtp config #10442

ruodil · 2026-01-06T08:00:52Z

Summary by CodeRabbit

Tests
- Updated test configuration parameters across multiple performance test scenarios to align token generation settings with Multi-Token Predictor configurations.

_{✏️ Tip: You can customize this high-level summary in your review settings.}

Description

otherwise it will cause assertion error:Configuration validation failed: config error: gen_max_tokens != gen_max_batch_size * (mtp_size + 1)
https://gitlab-master.nvidia.com/dlswqa/trtllm-ci/-/jobs/250090329#L2776

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 the stage-list parameter 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.md
and the scripts/test_to_stage_mapping.py helper.

kill

kill

Kill all running builds associated with pull request.

skip

skip --comment COMMENT

Skip 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-pipeline

Reuse 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.

coderabbitai · 2026-01-06T08:05:42Z

📝 Walkthrough

Walkthrough

Multiple test configuration YAML files for performance testing with disaggregated settings are updated. The max_num_tokens parameter in worker_config.gen is recalculated and updated across files with inline comments documenting the formula based on respective mtp_size values. Changes are configuration-only with no control flow or logic modifications.

Changes

Cohort / File(s)	Change Summary
Performance Test Configuration Updates `tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx__gen__bs_eplb0_mtp-Default.yaml` (12 files)	Updated `worker_config.max_num_tokens` from 128 to calculated values (ranging from 4 to 64) with inline comments explaining the formula: `mtp_size=N ⇒ max_num_tokens = batch_size * (N + 1)`. Files affected include configurations with mtp1, mtp2, and mtp3 settings across varying context, pipeline parallel, batch size, and dependency depth parameters. No other configuration fields modified.

Estimated code review effort

🎯 1 (Trivial) | ⏱️ ~3 minutes

Pre-merge checks

❌ Failed checks (1 inconclusive)

Check name	Status	Explanation	Resolution
Description check	❓ Inconclusive	The description provides the core reason for the changes (fixing a configuration validation error) but lacks details on Test Coverage and incomplete PR Checklist.	Add explicit test coverage details explaining how these config changes are tested, and clarify which checklist items were reviewed and are applicable to this configuration-only change.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title clearly and specifically describes the main change: restricting max_num_tokens in disagg MTP configurations to fix validation failures.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.

📜 Recent review details

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between ab58d7c and e03e328.

📒 Files selected for processing (12)

tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx1_pp4_gen6_tep8_bs1_eplb0_mtp3-Default.yaml
tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx1_pp8_gen1_dep16_bs1_eplb0_mtp3-Default.yaml
tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx1_pp8_gen1_dep8_bs4_eplb0_mtp2-Default.yaml
tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx1_pp8_gen1_tep8_bs2_eplb0_mtp3-Default.yaml
tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx3_pp8_gen1_dep16_bs8_eplb0_mtp2-Default.yaml
tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx3_pp8_gen1_dep32_bs2_eplb0_mtp3-Default.yaml
tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx5_pp4_gen1_dep16_bs8_eplb0_mtp3-Default.yaml
tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx5_pp4_gen1_dep32_bs2_eplb0_mtp3-Default.yaml
tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx7_pp4_gen1_dep16_bs16_eplb0_mtp1-Default.yaml
tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx8_pp4_gen1_dep16_bs32_eplb0_mtp1-Default.yaml
tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx8_pp4_gen1_dep32_bs4_eplb0_mtp3-Default.yaml
tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx8_pp4_gen1_dep32_bs8_eplb0_mtp3-Default.yaml

🧰 Additional context used

🧠 Learnings (1)

📓 Common learnings

Learnt from: djns99
Repo: NVIDIA/TensorRT-LLM PR: 6915
File: cpp/tensorrt_llm/kernels/cutlass_kernels/moe_gemm/moe_kernels.cu:4010-4012
Timestamp: 2025-08-14T23:23:27.449Z
Learning: For MOE (Mixture of Experts) code reviews in TensorRT-LLM, avoid repeatedly suggesting finalize fusion validation checks and safety assertions. The user djns99 has indicated these suggestions are repetitive and unwanted across multiple MOE-related changes.

Learnt from: samuellees
Repo: NVIDIA/TensorRT-LLM PR: 6974
File: tensorrt_llm/serve/scripts/benchmark_dataset.py:558-566
Timestamp: 2025-08-18T08:42:02.640Z
Learning: In TensorRT-LLM's RandomDataset (tensorrt_llm/serve/scripts/benchmark_dataset.py), when using --random-token-ids option, sequence length accuracy is prioritized over semantic correctness for benchmarking purposes. The encode/decode operations should use skip_special_tokens=True and add_special_tokens=False to ensure exact target token lengths.

Learnt from: pengbowang-nv
Repo: NVIDIA/TensorRT-LLM PR: 7192
File: tests/integration/test_lists/test-db/l0_dgx_b200.yml:56-72
Timestamp: 2025-08-26T09:49:04.956Z
Learning: In TensorRT-LLM test configuration files, the test scheduling system handles wildcard matching with special rules that prevent duplicate test execution even when the same tests appear in multiple yaml files with overlapping GPU wildcards (e.g., "*b200*" and "*gb200*").

⏰ 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 (12)

tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx8_pp4_gen1_dep16_bs32_eplb0_mtp1-Default.yaml (1)

52-53: LGTM! Calculation correctly implements the MTP constraint.

The max_num_tokens value of 64 correctly satisfies the formula max_batch_size * (mtp_size + 1) = 32 * 2 = 64, which resolves the configuration validation error mentioned in the PR description.

tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx8_pp4_gen1_dep32_bs4_eplb0_mtp3-Default.yaml (1)

52-53: LGTM! Calculation correctly implements the MTP constraint.

The max_num_tokens value of 16 correctly satisfies the formula max_batch_size * (mtp_size + 1) = 4 * 4 = 16, which resolves the configuration validation error.

tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx8_pp4_gen1_dep32_bs8_eplb0_mtp3-Default.yaml (1)

52-53: LGTM! Calculation correctly implements the MTP constraint.

The max_num_tokens value of 32 correctly satisfies the formula max_batch_size * (mtp_size + 1) = 8 * 4 = 32, which resolves the configuration validation error.

tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx1_pp8_gen1_dep8_bs4_eplb0_mtp2-Default.yaml (1)

52-53: LGTM! Calculation correctly implements the MTP constraint.

The max_num_tokens value of 12 correctly satisfies the formula max_batch_size * (mtp_size + 1) = 4 * 3 = 12, which resolves the configuration validation error.

tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx7_pp4_gen1_dep16_bs16_eplb0_mtp1-Default.yaml (1)

52-53: LGTM! Calculation correctly implements the MTP constraint.

The max_num_tokens value of 32 correctly satisfies the formula max_batch_size * (mtp_size + 1) = 16 * 2 = 32, which resolves the configuration validation error.

tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx1_pp4_gen6_tep8_bs1_eplb0_mtp3-Default.yaml (1)

52-53: LGTM! Formula correctly applied.

The calculation correctly implements the formula max_num_tokens = max_batch_size * (mtp_size + 1) where max_batch_size=1 (line 51) and mtp_size=3 (line 82), resulting in 1 * (3 + 1) = 4. This addresses the configuration validation assertion mentioned in the PR description.

tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx3_pp8_gen1_dep16_bs8_eplb0_mtp2-Default.yaml (1)

52-53: LGTM! Formula correctly applied.

The calculation correctly implements the formula where max_batch_size=8 (line 51) and mtp_size=2 (line 82), resulting in 8 * (2 + 1) = 24.

tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx1_pp8_gen1_tep8_bs2_eplb0_mtp3-Default.yaml (1)

52-53: LGTM! Formula correctly applied.

The calculation correctly implements the formula where max_batch_size=2 (line 51) and mtp_size=3 (line 83), resulting in 2 * (3 + 1) = 8.

tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx5_pp4_gen1_dep32_bs2_eplb0_mtp3-Default.yaml (1)

52-53: LGTM! Formula correctly applied.

The calculation correctly implements the formula where max_batch_size=2 (line 51) and mtp_size=3 (line 82), resulting in 2 * (3 + 1) = 8.

tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx5_pp4_gen1_dep16_bs8_eplb0_mtp3-Default.yaml (1)

52-53: LGTM! Formula correctly applied.

The calculation correctly implements the formula where max_batch_size=8 and mtp_size=3 (from num_nextn_predict_layers), resulting in 8 * (3 + 1) = 32. Verification of all 45+ disagg MTP test configurations confirms the constraint max_num_tokens = max_batch_size * (mtp_size + 1) is consistently applied across the test suite.

tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx3_pp8_gen1_dep32_bs2_eplb0_mtp3-Default.yaml (1)

52-53: LGTM! Correct formula implementation.

The max_num_tokens calculation correctly applies the formula max_batch_size * (mtp_size + 1) = 2 * (3 + 1) = 8, which aligns with the max_batch_size (line 51) and num_nextn_predict_layers (line 82). This fix resolves the configuration validation assertion error mentioned in the PR objectives.

tests/integration/defs/perf/disagg/test_configs/disagg/perf/deepseek-r1-fp4_128k8k_ctx1_pp8_gen1_dep16_bs1_eplb0_mtp3-Default.yaml (1)

52-53: LGTM! Correct formula implementation.

The max_num_tokens calculation correctly applies the formula max_batch_size * (mtp_size + 1) = 1 * (3 + 1) = 4, which aligns with the max_batch_size (line 51) and num_nextn_predict_layers (line 82). This fix resolves the configuration validation assertion error mentioned in the PR objectives.

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.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

ruodil · 2026-01-06T09:03:07Z

/bot skip --comment"skip test as just modifying cases"

tensorrt-cicd · 2026-01-06T09:11:07Z

PR_Github #30718 Bot args parsing error: usage: /bot skip --comment COMMENT
/bot skip: error: the following arguments are required: --comment

jiaganc · 2026-01-06T10:12:44Z

Also update max_num_tokens in PP8 configs?

Signed-off-by: Ruodi Lu <[email protected]>

ruodil · 2026-01-08T05:36:40Z

/bot --reuse-pipeline

github-actions · 2026-01-08T05:36:47Z