[None][feat] Update TRTLLM MoE cubins #6726
Conversation
|
Important Review skippedMore than 25% of the files skipped due to max files limit. The review is being skipped to prevent a low-quality review. 192 files out of 300 files are above the max files limit of 100. Please upgrade to Pro plan to get higher limits. You can disable this status message by setting the ✨ Finishing Touches
🧪 Generate unit tests
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. 🪧 TipsChatThere are 3 ways to chat with CodeRabbit:
SupportNeed help? Create a ticket on our support page for assistance with any issues or questions. CodeRabbit Commands (Invoked using PR/Issue comments)Type Other keywords and placeholders
Status, Documentation and Community
|
|
/bot run |
|
PR_Github #14547 [ run ] triggered by Bot |
|
PR_Github #14547 [ run ] completed with state |
rosenrodt
force-pushed
the
trtllm-moe-tma-opt
branch
2 times, most recently
from
c8760ab to
18482f2
Compare
|
/bot run |
|
PR_Github #14580 [ run ] triggered by Bot |
cpp/tensorrt_llm/kernels/trtllmGenKernels/batchedGemm/KernelRunner.cpp
Outdated
Show resolved
Hide resolved
| // Next, greedily pour all remaining tokens to one expert to maximize CTA tile count. | ||
| // E.g., at this point tokens over 4 experts are [1, 1, 1, 1], and we have 4 tokens left. | ||
| // If each CTA handles 4 tokens/expert, the greedy strategy is to pour all remaining tokens | ||
| // to any one expert to get to the 5th CTA tile. Otherwise, we can only get 4 tiles in total. |
There was a problem hiding this comment.
I had better understanding of the correctness of the proposed logic by reformulating in my head -- pour all remaining tokens to the buckets of tileN size. These buckets, if full, could be attributed to any expert. If bucket is less than full, no new bucket is needed as we could've added those tokens to the existing expert's bucket with 1 token. Furthermore, the estimation does not consider that the same token has to go to different experts. Thus, the estimation is correct and still pessimistic.
There was a problem hiding this comment.
These buckets, if full, could be attributed to any expert
Yes. Attributing the full buckets to any expert should be the more accurate description. Let's change the description in the source if needed
There was a problem hiding this comment.
Could you add this sentence to your comment, please? It is really what helped me to understand in the past :)
There was a problem hiding this comment.
Revised. Not pushed yet as I need to run a few quick sanity tests locally after the rebase/rebuild cycle
There was a problem hiding this comment.
Sorry about the long delay. I just rebased and pushed the update
|
PR_Github #14580 [ run ] completed with state |
rosenrodt
force-pushed
the
trtllm-moe-tma-opt
branch
from
18482f2 to
9a62e44
Compare
|
/bot run |
|
PR_Github #14777 [ run ] triggered by Bot |
|
IIUC, besides the general benefit of using less mem BW, this optimization allows to bump the tileN without paying too much of mem BW cost. It should help for the cases with skewed expert distribution. Could you help to measure if there is a perf benefit, please? |
|
PR_Github #14777 [ run ] completed with state |
rosenrodt
force-pushed
the
trtllm-moe-tma-opt
branch
from
9a62e44 to
af5a05c
Compare
|
/bot run |
|
PR_Github #14819 [ run ] triggered by Bot |
|
PR_Github #14819 [ run ] completed with state |
|
/bot run |
|
PR_Github #14895 [ run ] triggered by Bot |
|
PR_Github #14895 [ run ] completed with state |
rosenrodt
force-pushed
the
trtllm-moe-tma-opt
branch
3 times, most recently
from
596a567 to
2c8a3b1
Compare
|
/bot run |
|
PR_Github #18043 [ run ] triggered by Bot |
|
PR_Github #18043 [ run ] completed with state |
|
/bot run |
|
PR_Github #18063 [ run ] triggered by Bot |
|
PR_Github #18063 [ run ] completed with state |
rosenrodt
force-pushed
the
trtllm-moe-tma-opt
branch
from
2c8a3b1 to
519278b
Compare
|
/bot run |
|
PR_Github #18129 [ run ] triggered by Bot |
|
PR_Github #18129 [ run ] completed with state |
|
/bot run |
|
PR_Github #18186 [ run ] triggered by Bot |
|
PR_Github #18186 [ run ] completed with state |
Signed-off-by: Anthony Chang <27950904+rosenrodt@users.noreply.github.com>
remove limitations Signed-off-by: Anthony Chang <27950904+rosenrodt@users.noreply.github.com>
Signed-off-by: Anthony Chang <27950904+rosenrodt@users.noreply.github.com>
rosenrodt
force-pushed
the
trtllm-moe-tma-opt
branch
from
519278b to
f558680
Compare
|
/bot run |
|
PR_Github #18235 [ run ] triggered by Bot |
|
PR_Github #18235 [ run ] completed with state |
|
/bot run |
1 similar comment
|
/bot run |
|
PR_Github #18279 [ run ] triggered by Bot |
|
/bot kill |
|
Superceded by #8156. Closing. |
6 participants
@coderabbitai summary
Description
_tile_tokens_dim()into inside the thop. See https://github.com/NVIDIA/TensorRT-LLM/pull/6645/files#r2255699861Test 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.