[None][fix] Fix int4 awq for sm120/121#11561
Conversation
|
/bot --skip-test |
GitHub Bot Help
Provide a user friendly way for developers to interact with a Jenkins server. Run See details below for each supported subcommand. Details
Launch build/test pipelines. All previously running jobs will be killed.
kill
Kill all running builds associated with pull request. skip
Skip testing for latest commit on pull request. 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. |
|
/bot run --skip-test |
|
PR_Github #36127 [ run ] triggered by Bot. Commit: |
|
PR_Github #36127 [ run ] completed with state |
pamelap-nvidia
force-pushed
the
sm120_int4
branch
2 times, most recently
from
e749bc8 to
73b677c
Compare
|
/bot run |
📝 WalkthroughWalkthroughThis pull request differentiates SM architecture version support for quantized GEMM operations, introducing separate maximum SM versions for W4A8 (FP8-based) versus W4A16 (FP16/BF16-based) mixed-dtype operations. Additionally, weight preprocessing for SM >= 90 is refactored to apply consistently, and SM120/Blackwell candidate selection is gated on WEIGHT_ONLY flags. Changes
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes 🚥 Pre-merge checks | ✅ 3 | ❌ 2❌ Failed checks (2 warnings)
✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Tip 💬 Introducing Slack Agent: The best way for teams to turn conversations into code.Slack Agent is built on CodeRabbit's deep understanding of your code, so your team can collaborate across the entire SDLC without losing context.
Built for teams:
One agent for your entire SDLC. Right inside Slack. 👉 Get your free trial and get 200 agent minutes per Slack user (a $50 value). 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: 1
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (4)
tests/unittest/_torch/thop/parallel/test_w4a16_linear.py (1)
1-3:⚠️ Potential issue | 🟠 Major | ⚡ Quick winAdd the required NVIDIA copyright/license header.
This modified Python source file is missing the required file header.
As per coding guidelines: "All source files (.cpp, .h, .cu, .py) should contain an NVIDIA copyright header with the year of latest modification and Apache 2.0 license notice."
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@tests/unittest/_torch/thop/parallel/test_w4a16_linear.py` around lines 1 - 3, Add the required NVIDIA copyright/license header at the top of the Python source file (the module that currently starts with "import pytest" and "import torch"): insert the standard NVIDIA header block including the year of latest modification and the Apache 2.0 license notice as specified by project guidelines, placing it above all imports so the file begins with the header comment followed by the existing import lines.cpp/tensorrt_llm/kernels/cutlass_kernels/cutlass_heuristic.cpp (1)
2-2:⚠️ Potential issue | 🟠 Major | ⚡ Quick winUpdate the modified-file copyright year.
This file was changed in this PR, but the header still ends at 2023.
As per coding guidelines: "Include NVIDIA copyright header on all new files; update year on modified files."
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@cpp/tensorrt_llm/kernels/cutlass_kernels/cutlass_heuristic.cpp` at line 2, Update the copyright header in the file to include the current year (extend the year range on the copyright header line from 2023 to the current year) so the modified-file header is correct; locate the copyright header at the top of cpp/tensorrt_llm/kernels/cutlass_kernels/cutlass_heuristic.cpp and change the year range in that header comment to include the current year (e.g., 2023-2026).tests/unittest/_torch/thop/parallel/test_finegrained_mixed_dtype_gemm.py (1)
1-3:⚠️ Potential issue | 🟠 Major | ⚡ Quick winAdd the required NVIDIA copyright/license header.
This modified Python source file is missing the required file header.
As per coding guidelines: "All source files (.cpp, .h, .cu, .py) should contain an NVIDIA copyright header with the year of latest modification and Apache 2.0 license notice."
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@tests/unittest/_torch/thop/parallel/test_finegrained_mixed_dtype_gemm.py` around lines 1 - 3, Add the required NVIDIA copyright and Apache-2.0 license header as a top-of-file comment block in tests/unittest/_torch/thop/parallel/test_finegrained_mixed_dtype_gemm.py (place it before the existing imports: pytest, torch, and the utils imports), using the latest modification year and the standard NVIDIA header text; ensure the header is a commented Python block (each line beginning with #) so it does not affect execution.tests/unittest/_torch/thop/parallel/test_w4a8_linear.py (1)
1-3:⚠️ Potential issue | 🟠 Major | ⚡ Quick winAdd the required NVIDIA copyright/license header.
This modified Python source file is missing the required file header.
As per coding guidelines: "All source files (.cpp, .h, .cu, .py) should contain an NVIDIA copyright header with the year of latest modification and Apache 2.0 license notice."
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@tests/unittest/_torch/thop/parallel/test_w4a8_linear.py` around lines 1 - 3, Add the required NVIDIA copyright/license header at the top of the file (above the existing imports for pytest, torch and Parameter): insert the standard NVIDIA header block including the correct latest modification year and the Apache 2.0 license notice used across the codebase, ensuring it appears before any code or imports so the file begins with the license header.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@tensorrt_llm/quantization/functional.py`:
- Around line 968-971: The branch that sets do_weight_interleave based on sm_ ==
100 or sm_ == 103 is unreachable because sm_ is always remapped to 80 when sm_
>= 90; to fix, change the logic so the SM100/103 override is evaluated before
the remap or narrow the remap condition (e.g., only remap when 90 <= sm_ < 100),
ensuring the check against sm_ == 100 or sm_ == 103 (the code using variable sm_
and setting do_weight_interleave) can actually execute.
---
Outside diff comments:
In `@cpp/tensorrt_llm/kernels/cutlass_kernels/cutlass_heuristic.cpp`:
- Line 2: Update the copyright header in the file to include the current year
(extend the year range on the copyright header line from 2023 to the current
year) so the modified-file header is correct; locate the copyright header at the
top of cpp/tensorrt_llm/kernels/cutlass_kernels/cutlass_heuristic.cpp and change
the year range in that header comment to include the current year (e.g.,
2023-2026).
In `@tests/unittest/_torch/thop/parallel/test_finegrained_mixed_dtype_gemm.py`:
- Around line 1-3: Add the required NVIDIA copyright and Apache-2.0 license
header as a top-of-file comment block in
tests/unittest/_torch/thop/parallel/test_finegrained_mixed_dtype_gemm.py (place
it before the existing imports: pytest, torch, and the utils imports), using the
latest modification year and the standard NVIDIA header text; ensure the header
is a commented Python block (each line beginning with #) so it does not affect
execution.
In `@tests/unittest/_torch/thop/parallel/test_w4a16_linear.py`:
- Around line 1-3: Add the required NVIDIA copyright/license header at the top
of the Python source file (the module that currently starts with "import pytest"
and "import torch"): insert the standard NVIDIA header block including the year
of latest modification and the Apache 2.0 license notice as specified by project
guidelines, placing it above all imports so the file begins with the header
comment followed by the existing import lines.
In `@tests/unittest/_torch/thop/parallel/test_w4a8_linear.py`:
- Around line 1-3: Add the required NVIDIA copyright/license header at the top
of the file (above the existing imports for pytest, torch and Parameter): insert
the standard NVIDIA header block including the correct latest modification year
and the Apache 2.0 license notice used across the codebase, ensuring it appears
before any code or imports so the file begins with the license header.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Enterprise
Run ID: 71af6f98-55b2-4986-a61c-11676022e347
📒 Files selected for processing (6)
cpp/tensorrt_llm/kernels/cutlass_kernels/cutlass_heuristic.cpptensorrt_llm/_torch/custom_ops/torch_custom_ops.pytensorrt_llm/quantization/functional.pytests/unittest/_torch/thop/parallel/test_finegrained_mixed_dtype_gemm.pytests/unittest/_torch/thop/parallel/test_w4a16_linear.pytests/unittest/_torch/thop/parallel/test_w4a8_linear.py
|
PR_Github #46416 [ run ] triggered by Bot. Commit: |
|
PR_Github #46416 [ run ] completed with state
|
pamelap-nvidia
changed the title
|
/bot run |
|
PR_Github #46578 [ run ] triggered by Bot. Commit: |
|
PR_Github #46578 [ run ] completed with state
|
|
/bot run |
|
PR_Github #46862 [ run ] triggered by Bot. Commit: |
|
PR_Github #46862 [ run ] completed with state |
|
I think @yuxianq covered most of my points. So other than those, LGTM! Could we just run ≠the l0_6000 or just GPT-OSS E2E test before merging? |
Signed-off-by: Pamela <179191831+pamelap-nvidia@users.noreply.github.com>
The Python preprocess_weights_for_mixed_gemm had its sm_ >= 90 -> sm_ = 80 adjustment trapped inside an elif tied to the 2-D shape check, so vanilla 2-D Linear weights on SM120/121 kept sm_ = 120 and skipped both row permutation (gated sm_ < 100) and column interleave (gated sm_ < 90). The CUTLASS Sm80 finegrained kernel then read a mismatched layout, producing 80%+ mismatch with ground truth. Move the adjustment out of the elif so it applies to both 2-D and 3-D inputs, mirroring the C++ preprocessor. Also lift the FinegrainedMixedDtypeGemm activation-aware version gate so W4A16 is enabled up through SM121 while W4A8 stays gated at SM103 (no FP8-activation dispatch on the SM120 Sm80 fallback). Tests skip correctly under the new gate. Signed-off-by: Pamela <179191831+pamelap-nvidia@users.noreply.github.com>
Signed-off-by: Pamela <179191831+pamelap-nvidia@users.noreply.github.com>
Signed-off-by: Pamela <179191831+pamelap-nvidia@users.noreply.github.com>
Signed-off-by: Pamela <179191831+pamelap-nvidia@users.noreply.github.com>
Signed-off-by: Pamela <179191831+pamelap-nvidia@users.noreply.github.com>
pamelap-nvidia
force-pushed
the
sm120_int4
branch
from
82fa869 to
2823247
Compare
|
/bot run --extra-stage "GB10-PyTorch-Post-Merge-1,RTXPro6000D-PyTorch-Post-Merge-1,RTXPro6000D-4_GPUs-PyTorch-Post-Merge-1,RTXPro6000D-4_GPUs-PyTorch-Post-Merge-2" |
|
PR_Github #48029 [ run ] triggered by Bot. Commit: |
|
PR_Github #48029 [ run ] completed with state
|
|
/bot run --extra-stage "GB10-PyTorch-Post-Merge-1,RTXPro6000D-PyTorch-Post-Merge-1,RTXPro6000D-4_GPUs-PyTorch-Post-Merge-1,RTXPro6000D-4_GPUs-PyTorch-Post-Merge-2" |
|
PR_Github #48218 [ run ] triggered by Bot. Commit: |
|
PR_Github #48218 [ run ] completed with state
|
|
/bot run --extra-stage "GB10-PyTorch-Post-Merge-1,RTXPro6000D-PyTorch-Post-Merge-1,RTXPro6000D-4_GPUs-PyTorch-Post-Merge-1,RTXPro6000D-4_GPUs-PyTorch-Post-Merge-2" --disable-fail-fast |
|
PR_Github #48392 [ run ] triggered by Bot. Commit: |
|
PR_Github #48392 [ run ] completed with state
|
|
/bot run --extra-stage "GB10-PyTorch-Post-Merge-1,RTXPro6000D-PyTorch-Post-Merge-1,RTXPro6000D-4_GPUs-PyTorch-Post-Merge-1,RTXPro6000D-4_GPUs-PyTorch-Post-Merge-2" --disable-fail-fast |
|
PR_Github #48607 [ run ] triggered by Bot. Commit: |
|
PR_Github #48607 [ run ] completed with state
|
|
/bot run --extra-stage "GB10-PyTorch-Post-Merge-1,RTXPro6000D-PyTorch-Post-Merge-1,RTXPro6000D-4_GPUs-PyTorch-Post-Merge-1,RTXPro6000D-4_GPUs-PyTorch-Post-Merge-2" --disable-fail-fast |
|
PR_Github #48655 [ run ] triggered by Bot. Commit: |
|
PR_Github #48655 [ run ] completed with state |
5 participants
Summary by CodeRabbit
Bug Fixes
Tests
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
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 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.