Add support for fused Q Up-Proj GEMM/RoPE/Quant. - #3303

Merged

sudhakarsingh27 merged 16 commits into

NVIDIA:mainfrom

chaseblock:qrope_fusion

Aug 7, 2026

Commits on Aug 5, 2026

Add support for fused Q Up-Proj GEMM/RoPE/Quant.
chaseblock
committed
c19f0df
[pre-commit.ci] auto fixes from pre-commit.com hooks

pre-commit-ci[bot]
authored and
chaseblock
committed
cd730aa
Remove unused function from fused mla q uproj, add error handling
chaseblock
committed
d0e3915
[pre-commit.ci] auto fixes from pre-commit.com hooks

pre-commit-ci[bot]
authored and
chaseblock
committed
9854af3
Adjust comment SBHD/BSHD
chaseblock
committed
5f90db7
Merge branch 'main' into qrope_fusion
sraman-rgb
authored
b993fde
Add cudnn_frontend version check; eliminate code dup in
chaseblock
committed
e2ebeac
Add missing docstrings
chaseblock
committed
417d233
Add unit test for q up proj fusion
chaseblock
committed
dc89f01
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot]
committed
24e0b03
Add test for bwd path of fused gemm+rope+quant.
chaseblock
committed
e20f595
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot]
committed
9770829
Consolidate fused q proj tests and add to unit test list
chaseblock
committed
a14039a
[pre-commit.ci] auto fixes from pre-commit.com hooks
pre-commit-ci[bot]
committed
d6785bf

Commits on Aug 6, 2026

Remove backward test from fused mla q uproj.
chaseblock
committed
11c0f1b
Merge branch 'main' into qrope_fusion
sudhakarsingh27
authored
8c01e02