Pull Requests microsoft/onnxruntime

Speedup on macos github actions: binary vcpkg cache

#32692 by eserscor was merged 2026-09-18 21:03

Collapse A10 pools to a single pool

#32691 by eserscor was merged 2026-09-18 19:38

Remove more dead pipeline yaml

#32690 by eserscor was closed 2026-09-18 17:39

Update syntax for release mode in NPM

#32689 by eserscor was merged 2026-09-18 17:55

[WebGPU] Upgrade Dawn to the latest

#32685 by mei1127 was closed 2026-09-18 02:34

Limit sparse initializer dense allocation

#32684 by apsonawane was merged 2026-09-18 16:48

Fix binary vcpkg caching for github actions

#32682 by eserscor was merged 2026-09-18 17:07

Remove DML packaging

#32677 by eserscor was merged 2026-09-17 18:59

Move back to main A10 image

#32676 by eserscor was merged 2026-09-17 18:07

Fix Objective-C API docs generation

#32674 by Copilot was closed 2026-09-17 16:12

Fix WebGPU subgroup config header dependency

#32672 by xadupre was merged 2026-09-17 22:19

Batch QMoE CPU QNBit expert GEMMs into grouped MLAS dispatches

#32668 by tianleiwu was merged 2026-09-17 17:18

[WebGPU] Enable subgroup matrix kernels on AMD GPUs

#32667 by gyagp was merged 2026-09-17 09:14

[Plugin EP] Preserve streams in single-tensor data transfers

#32666 by xiaofeihan1 was merged 2026-09-17 07:40

[Build] Use H100 instead of A10 pool for CI

#32664 by tianleiwu was closed 2026-09-17 03:34

Fix WebNN ConvTranspose shape arithmetic overflow

#32661 by edgchen1 was merged 2026-09-17 16:16

Add BF16 routing support to CUDA MoE

#32659 by kunal-vaishnavi was merged 2026-09-17 03:03

Combine npm and nuget packing pipelines

#32655 by eserscor was merged 2026-09-18 13:11

perf: run block-wise int4/int8 QMoE CPU experts on the MLAS QNBit GEMM kernels

#32644 by ykhrustalev was merged 2026-09-17 06:55

Reject non-control-flow nodes that carry subgraphs during session state finalization

#32641 by miaobin was merged 2026-09-17 17:21

Reject nodes that feed inputs to a zero-input operator schema

#32633 by shiyi9801 was merged 2026-09-17 17:00

Add Arm KleidiAI compatibility documentation

#32622 by damdoo001-arm was closed 2026-09-16 14:04

Optimize CPU uint8 GatherBlockQuantized dequantization

#32621 by Copilot was merged 2026-09-17 03:03

Add review guidance for deprecated components

#32620 by edgchen1 was merged 2026-09-17 16:29

Add GQA workspace estimation ep:CUDA memory

#32617 by titaiwangms was merged 2026-09-18 00:19

Improve CPU provider build time with Eigen PCH

#32616 by titaiwangms was merged 2026-09-17 03:03

Start transitioning pools to unified Windows pool

#32612 by eserscor was merged 2026-09-15 16:44

Fix heap buffer overflow from reusing a packed sub-byte buffer for a full-byte tensor

#32611 by wangw-1991 was merged 2026-09-19 06:55

Fix out-of-bounds read in EmbedLayerNormalization shape inference for scalar beta

#32610 by wangw-1991 was merged 2026-09-17 07:17

Fix out-of-bounds read when a Loop/Scan body input is also an initializer

#32609 by wangw-1991 was merged 2026-09-19 06:57