[None][feat] Stack PRs for sweep perfing#13205

Draft

Wanli-Jiang wants to merge 37 commits intoNVIDIA:mainfrom

Wanli-Jiang:user/williamj/stack-pr-v2

Commits on Apr 22, 2026

[None][chore] Update flashinfer-python from 0.6.6 to 0.6.8rc1

yihwang-nv
authored and
Wanli-Jiang
committed
[None][chore] Update Docker to upgrade nvidia-cutlass-dsl to 4.4.2

yihwang-nv
authored and
Wanli-Jiang
committed
[None][chore] Update Docker to upgrade nvidia-cutlass-dsl to 4.4.2

yihwang-nv
authored and
Wanli-Jiang
committed
[None][chore] Update CI image tags to PR-13064 staging images

yihwang-nv
authored and
Wanli-Jiang
committed
[None][fix] Remove stale nvidia-cutlass-dsl before pip install in CI

yihwang-nv
authored and
Wanli-Jiang
committed
Revert "[None][chore] Update CI image tags to PR-13064 staging images"

yihwang-nv
authored and
Wanli-Jiang
committed
[None][fix] Clean stale nvidia-cutlass-dsl before pip install in CI

yihwang-nv
authored and
Wanli-Jiang
committed
[None][fix] Guard cutlass-dsl imports and clean site-packages in Check Test List

yihwang-nv
authored and
Wanli-Jiang
committed
[None][chore] Regenerate security_scanning/poetry.lock content-hash

yihwang-nv
authored and
Wanli-Jiang
committed
[None][fix] Clean stale packages in SLURM install and update CI images

yihwang-nv
authored and
Wanli-Jiang
committed
[None][chore] Remove clean_site_packages.py and all references

yihwang-nv
authored and
Wanli-Jiang
committed
[None][chore] Update flashinfer-python from 0.6.8rc1 to 0.6.8

yihwang-nv
authored and
Wanli-Jiang
committed
[None][fix] Remove stale PersistentTileSchedulerParams hooks for cutlass-dsl 4.4.2

yihwang-nv
authored and
Wanli-Jiang
committed
[None][fix] Route custom pipelines through sm100 _make_sync_object for cutlass-dsl 4.4.2

yihwang-nv
authored and
Wanli-Jiang
committed
[None][fix] Accept loc/ip kwargs in custom pipeline overrides for cutlass-dsl 4.4.2

yihwang-nv
authored and
Wanli-Jiang
committed
[None][chore] Revert security_scanning/poetry.lock to origin/main

yihwang-nv
authored and
Wanli-Jiang
committed
Revert docker image since the staging one is removed
Wanli-Jiang
committed
[None][feat] Add NemotronHPuzzleConfig modeling
Wanli-Jiang
committed
Upgrade flashinfer to nightly build
Wanli-Jiang
committed
Fix reviewer's comment
Wanli-Jiang
committed
Update mamba SSD to use flashinfer kernel
Wanli-Jiang
committed
Fix python deps for cutlass-dsl
Wanli-Jiang
committed
Fix error when rebasing commit
Wanli-Jiang
committed
[None][feat] Optimize causal_conv1d prefill and decode kernels
Wanli-Jiang
committed
Fix unittests of nemotron-h
Wanli-Jiang
committed
[TRTLLM-11585][feat] Add CUTEDSL moe backend for nemotron-h
Wanli-Jiang
committed
[TRTLLM-11585][refactor] Rename gather MoE SwiGLU kernel to act_fusion
Wanli-Jiang
committed
Fix cutedsl upgrade issues
Wanli-Jiang
committed
Refactor the routing part in trtllmgen

ChristinaZ
authored and
Wanli-Jiang
committed
Revise based on review

ChristinaZ
authored and
Wanli-Jiang
committed
Update the deepseek routing

ChristinaZ
authored and
Wanli-Jiang
committed
[None][fix] Fix Mamba cache correctness under MTP + CUDA-graph padding
Wanli-Jiang
committed
Use updated docker image
Wanli-Jiang
committed
Revert "[None][fix] Fix Mamba cache correctness under MTP + CUDA-graph padding"
Wanli-Jiang
committed
[None][fix] Fix Mamba cache correctness under MTP + CUDA-graph padding
Wanli-Jiang
committed
Revert "[None][fix] Fix Mamba cache correctness under MTP + CUDA-graph padding"
Wanli-Jiang
committed
[None][fix] Fix Mamba cache correctness under MTP + CUDA-graph padding
Wanli-Jiang
committed