ORT 1.28.0 release cherry-pick round 1#29771

Merged

tianleiwu merged 16 commits into

tlwu/rel-1.28.0-cherry-pick-round1

Jul 18, 2026

Commits on Jul 17, 2026

Fix libcudart.so.13 hard dependency in pybind module breaking import on CPU-only Linux (#29590)

Copilot
and
tianleiwu
committed
[CUDA] Fix XQA GroupQueryAttention cudaErrorInvalidValue on Blackwell (sm_120) (#29706)
tianleiwu
committed
[CUDA] Do not link nvrtc (#29705)
tianleiwu
committed
[CUDA] Update cuda arch list for packages of cuda 12.8 (#29711)
tianleiwu
committed
[CUDA] Add cuDNN-free ArgMax/ArgMin/ReduceSum and fix LogSoftmax on plugin EP (#29620)
tianleiwu
committed
Enable Spectre-mitigated MSVC libs for BinSkim builds (#29624)

tianleiwu
and
Copilot
committed
[BUILD] CUDA_QUANT_PREPROCESS off by default and Adjust CI (#29687)
tianleiwu
committed
[CUDA] Fix null allocator passed to plugin EP kernel PrePack (#29658)
tianleiwu
committed
Add Windows ARM64 CUDA plugin package and align CUDA metadata/artifact naming (#28896)

tianleiwu
and
Copilot
committed
Enable fpA_intB GEMM in CUDA builds and add configurable options (#29622)
tianleiwu
committed
[Build] Use GPU pool to unblock CI temporarily (#29731)

tianleiwu
and
GitHub Copilot
committed
Add OrtErrorCode::ORT_DEVICE_RESET. (#29748)

edgchen1
authored and
tianleiwu
committed
Fix plugin EP allocator deleter lifetime

yen-shi
authored and
tianleiwu
committed
Add Java setup to Windows x64 QNN CI Pipeline (#29754)

mcollinswisc
authored and
tianleiwu
committed
[Build] install java in Windows nuget packaging
tianleiwu
committed

Commits on Jul 18, 2026

Fix windows build for sm90 or later
tianleiwu
committed