[https://nvbugs/5473781][fix] Cherry-pick "Fix llama 4 FP8 for PP>1 (#7220)" #7967
Conversation
Signed-off-by: Mike Iovine <[email protected]>
|
/bot run --disable-fail-fast |
📝 WalkthroughWalkthroughUpdates fusion gating conditions in Llama4DecoderLayer to disallow MLP/MOE fusion when pipeline parallelism is enabled, aligning pre/post fusion flags. Adds a TODO about PP>1 fusion. Expands integration tests to include an FP8 scenario with tp=4, pp=2, ep=1. Changes
Sequence Diagram(s)sequenceDiagram
autonumber
participant Config as RuntimeConfig
participant Layer as Llama4DecoderLayer
participant MLP as MLP/MOE Blocks
Note over Config,Layer: Initialization / Build-Time
Config->>Layer: enable_fusion, has_tp(), has_pp(), enable_attention_dp
Layer->>Layer: Compute PRE_MLP_FUSION = enable_fusion && has_tp() && !enable_attention_dp && !has_pp()
Layer->>Layer: Set POST_MLP_FUSION = PRE_MLP_FUSION
Layer->>Layer: Compute PRE_MOE_FUSION = enable_fusion && !enable_attention_dp && !has_pp()
Layer->>Layer: Set POST_MOE_FUSION = PRE_MOE_FUSION
Note over Layer,MLP: Execution-Time
alt Fusion enabled (no PP)
Layer->>MLP: Run fused MLP/MOE path
else Fusion disabled or PP>0
Layer->>MLP: Run unfused MLP/MOE path
end
Note over Layer: TODO: Fusion issue when PP > 1
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
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: 0
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
tensorrt_llm/_torch/models/modeling_llama.py (1)
834-846: Guard min-latency fusion flags under PP>1
In Llama4MinLatencyDecoderLayer.init (modeling_llama_min_latency.py), wrap each ofself.fusion_config.PRE_MLP_FUSION = mapping.has_tp() self.fusion_config.POST_MLP_FUSION = mapping.has_tp() self.fusion_config.PRE_MOE_FUSION = mapping.has_tp() self.fusion_config.POST_MOE_FUSION = mapping.has_tp()with
and not mapping.has_pp()(e.g.mapping.has_tp() and not mapping.has_pp()) to match Llama4DecoderLayer’s PP gating and prevent regressions.
🧹 Nitpick comments (5)
tensorrt_llm/_torch/models/modeling_llama.py (5)
411-446: DRY up fusion gating to a single predicate used by both MLP and MOE.The same condition is repeated; a single local variable improves readability and reduces drift.
Apply this refactor:
self.enable_fusion = os.environ.get( "TRTLLM_LLAMA_EAGER_FUSION_DISABLED", "0") == "0" @@ - if self.is_mlp_layer: + # Shared predicate: enable feed-forward fusion only when TP>1, no attention-DP, fusion enabled, and no PP + enable_ffn_fusion = ( + model_config.mapping.has_tp() + and not self.enable_attention_dp + and self.enable_fusion + and not model_config.mapping.has_pp() + ) + + if self.is_mlp_layer: self.feed_forward = GatedMLP( @@ - # TODO(TRTLLM-7809): Fix fusion with PP>1 - self.fusion_config.PRE_MLP_FUSION = model_config.mapping.has_tp( - ) and not self.enable_attention_dp and self.enable_fusion and not model_config.mapping.has_pp( - ) - self.fusion_config.POST_MLP_FUSION = self.fusion_config.PRE_MLP_FUSION + # TODO(TRTLLM-7809): Fix fusion with PP>1 + self.fusion_config.PRE_MLP_FUSION = enable_ffn_fusion + self.fusion_config.POST_MLP_FUSION = enable_ffn_fusion @@ - self.fusion_config.PRE_MOE_FUSION = model_config.mapping.has_tp( - ) and not self.enable_attention_dp and self.enable_fusion and not model_config.mapping.has_pp( - ) - self.fusion_config.POST_MOE_FUSION = self.fusion_config.PRE_MOE_FUSION + self.fusion_config.PRE_MOE_FUSION = enable_ffn_fusion + self.fusion_config.POST_MOE_FUSION = enable_ffn_fusion
353-358: Fix init return annotation.init should return None; current Tuple[...] hint is incorrect and confuses type checkers.
- ) -> Tuple[torch.Tensor, torch.Tensor]: + ) -> None:
615-620: Fix init return annotation (duplicate in LlamaDecoderLayer).Same issue as above.
- ) -> Tuple[torch.Tensor, torch.Tensor]: + ) -> None:
457-458: Broaden next_attn type to the shared Attention base.next_attn may be Llama4Attention or LlamaAttention; annotate to the base type.
- self.next_attn: LlamaAttention = None + self.next_attn: Optional[Attention] = None
1-1: Add SPDX Apache-2.0 header.Per repo guidelines, prepend the NVIDIA Apache-2.0 SPDX header with current year.
+# SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved. +# SPDX-License-Identifier: Apache-2.0 +# +# Licensed under the Apache License, Version 2.0 (the "License"); +# you may not use this file except in compliance with the License. +# You may obtain a copy of the License at +# +# http://www.apache.org/licenses/LICENSE-2.0 +# +# Unless required by applicable law or agreed to in writing, software +# distributed under the License is distributed on an "AS IS" BASIS, +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +# See the License for the specific language governing permissions and +# limitations under the License.
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (2)
tensorrt_llm/_torch/models/modeling_llama.py(2 hunks)tests/integration/defs/accuracy/test_llm_api_pytorch.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:
tests/integration/defs/accuracy/test_llm_api_pytorch.pytensorrt_llm/_torch/models/modeling_llama.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:
tests/integration/defs/accuracy/test_llm_api_pytorch.pytensorrt_llm/_torch/models/modeling_llama.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:
tests/integration/defs/accuracy/test_llm_api_pytorch.pytensorrt_llm/_torch/models/modeling_llama.py
🧠 Learnings (1)
📚 Learning: 2025-07-28T17:06:08.621Z
Learnt from: moraxu
PR: NVIDIA/TensorRT-LLM#6303
File: tests/integration/test_lists/qa/examples_test_list.txt:494-494
Timestamp: 2025-07-28T17:06:08.621Z
Learning: In TensorRT-LLM testing, it's common to have both CLI flow tests (test_cli_flow.py) and PyTorch API tests (test_llm_api_pytorch.py) for the same model. These serve different purposes: CLI flow tests validate the traditional command-line workflow, while PyTorch API tests validate the newer LLM API backend. Both are legitimate and should coexist.
Applied to files:
tests/integration/defs/accuracy/test_llm_api_pytorch.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
🔇 Additional comments (3)
tests/integration/defs/accuracy/test_llm_api_pytorch.py (1)
699-701: Good addition: cover FP8 with PP>1 (tp4, pp2).IDs are consistent; world-size guard below will skip when topology isn’t available. Looks good.
tensorrt_llm/_torch/models/modeling_llama.py (2)
422-427: Disable MLP fusion when PP>1 — correct fix for FP8+PP.This gating prevents fused pre/post-MLP paths under pipeline parallelism and aligns with the reported issue. ✅
440-444: Mirror MOE fusion gating to disallow under PP>1 — consistent with MLP.This keeps behavior coherent across MLP/MOE in PP setups. ✅
|
PR_Github #19829 [ run ] triggered by Bot |
|
PR_Github #19829 [ run ] completed with state |
|
/bot run --disable-fail-fast |
|
/bot kill |
|
PR_Github #19957 [ run ] triggered by Bot |
|
The release branch is not open yet. There will be another MI of release 1.0 -> main and the main -> release 1.1. Thus, we do not need to cherry pick this change into the release branch; it will be included in the next mass catch up. |
Pull request was closed
|
PR_Github #19957 [ run ] completed with state |
|
PR_Github #19959 [ ] completed with state |
4 participants
Description
This fix got into main but didn't accidentally did not make the release branch cutoff. It needs to be cherry-picked into the release branch.
Test Coverage
N/A
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.
Summary by CodeRabbit