microsoft/onnxruntime

Pull Requests Commits

Cuda Graph on by default (#26929)

umangb-09 committed 139 days ago

Verified 0a93edb0

[QNN-EP] Apply Softmax layout transformation for GPU (#26233)

qti-mattsinc committed 139 days ago

Verified 65e694ea

[QNN-EP] Support pad op pre-opset11 (#26248)

qti-mattsinc committed 139 days ago

Verified 15efe1c1

Minimize binary footprint of the FL nuget (#26961)

baijumeswani committed 139 days ago

Verified a5831a3d

Integration with ONNX==1.20.1 (#26579)

titaiwangms committed 139 days ago

Verified de398c41

[QNN EP] Default to HTP instead of using last provided device (#26696)

qti-mattsinc committed 139 days ago

Verified addc8120

[js/node] fix incorrect output name mapping (#26939)

fs-eire committed 139 days ago

Verified ce4a7c58

coreml: Fix Gather fp16 support (#26442)

borg323 committed 139 days ago

Verified 412233f9

Allow present_key to be empty when past_key is provided in Attention (#26303)

justinchuby committed 139 days ago

Verified 1d149f5a

[MLAS] Fix building on FreeBSD / powerpc64le (#26895)

pkubaj committed 139 days ago

Verified e0d5b585

[CUDA] Update Flash Attention Implementation and APIs (#26937)

tianleiwu committed 140 days ago

Verified 5d24c954

[EP ABI] Add APIs to create control flow kernels for plugin EPs (#26927)

adrianlizarraga committed 141 days ago

Verified 1ed8fd90

[CUDA] GQA CUDA Kernel Fusion and Performance Optimization (#26920)

tianleiwu committed 141 days ago

Verified 39d8520b

Allow choosing memory type for CudaPinned allocator (#26892)

AndreyOrb committed 142 days ago

Verified 5b88e4ed

[QNN EP] Add Case-2 LPBQ pattern support for Gemm and Matmul nodes (#25865)

quic-tirupath committed 142 days ago

Verified c1da27c2

Update protobuf references from 3.20.3 to 4.25.8 (#26910)

eserscor committed 142 days ago

Verified e6110d03

Add OrtExternalResourceImporter API for D3D12 shared resource import (#26828)

nieubank committed 142 days ago

Verified c54be3c7

[OVEP] Update openvino version to 2025.4.1 (#26943)

preetha-intel committed 142 days ago

Verified a14231a5

[QNN EP] Update tensor naming logic in DebugNodeInputsOutputs (#26660)

yuhuchua-qti committed 142 days ago

Verified 3924b6e8

[ICM CXX] Address Pad kernel shortcomings (#26864)

yuslepukhin committed 142 days ago

Verified 5d596088

[webgpu] Simplify the signature of CanApplyFlashAttention (#26926)

qjia7 committed 143 days ago

Verified 56c984ff

[webgpu] Support `Split-K` in more situations (#26806)

Jiawei-Shao committed 143 days ago

Verified 4a858a82

[CUDA] Speed up flash attention build (#26924)

tianleiwu committed 143 days ago

Verified 541d5daf

GPG changed default install location on 12/31/2025 for Windows (#26917)

eserscor committed 143 days ago

Verified 751af64a

[QNN EP] Fix the build error when the specified QNN library is missing (#26872)

kuanyul-qti committed 144 days ago

Verified 517e06c6

[EP ABI] Add OrtKernelInfo APIs for kernel-based plugin EPs (#26803)

adrianlizarraga committed 144 days ago

Verified 27e6002d

[QNN EP] Add Static bias encoding support for Conv op (#26710)

quic-ashwshan committed 145 days ago

Verified 1e393751

Bump qs and body-parser in /onnxruntime/test/wasm (#26914)

dependabot[bot] committed 145 days ago

Verified 342f7193

Build onnxruntime with `caller-framework WinAI` for FoundryLocal (#26915)

baijumeswani committed 145 days ago

Verified 0409ba6f

[webgpu] Support any batch size for dp4a matmul path (#26884)

qjia7 committed 145 days ago

Verified 34a0b152

Newer Older