microsoft/onnxruntime

Pull Requests Commits

fix patch files

edgchen1 committed 301 days ago

265cd876

patch cpuinfo with fp16 detection fallback

edgchen1 committed 301 days ago

5d8b20a4

Revert "add MlasIsDynamicQGemmAvailable() function, use that instead of checking CPUIDInfo::HasArm_SME2()."

edgchen1 committed 301 days ago

b124ff43

add note about has_fp16_ workaround, remove debug print

edgchen1 committed 307 days ago

9b186b21

debug - print has_fp16_ value

edgchen1 committed 308 days ago

cb7bbd6b

edgchen1 committed 309 days ago

e2ddd856

bump cpuinfo to 403d652dca4c1046e8145950b1c0997a9f748b57 to get asan issue fix

edgchen1 committed 309 days ago

a18cdcac

add MlasIsDynamicQGemmAvailable() function, use that instead of checking CPUIDInfo::HasArm_SME2().

edgchen1 committed 309 days ago

6a9cfc75

Upgrade cpuinfo version. Update ARM64 Windows feature detection to use cpuinfo functions which are now implemented.

edgchen1 committed 309 days ago

ab320744

[js/rn] Migrate to JSI implementation (#25764)

hans00 committed 310 days ago

Verified d55ade03

Consistent with the configuration in the packaged cmake (#26104)

WangFengtu1996 committed 310 days ago

Verified e8243313

[Lora] Adjust device dispatch according to the new OrtDevice defs (#26551)

yuslepukhin committed 310 days ago

Verified fe373724

webgpu: fix dispatch size issue of Transpose operator (#26501)

xhcao committed 310 days ago

Verified 1936d646

[web/webgpu] add `validationMode` to webgpu EP specific options. (#26581)

fs-eire committed 311 days ago

Verified b39e1443

fix zero size input for where op on webgpu (#26576)

Guenther Schmuelling committed 313 days ago

Verified 3f4ab763

webgpu csum - axis can be int32 or int64 (#26578)

Guenther Schmuelling committed 313 days ago

Verified 4f600bba

[webgpu] Add implementation of BiasGelu (#26560)

fs-eire committed 313 days ago

Verified 683e78c3

qgemm: optimize avxvnni QGEMM inner kernel for M=1 (#22952)

r-devulap committed 313 days ago

Verified d6a372aa

Bump js-yaml from 4.1.0 to 4.1.1 in /js (#26577)

dependabot[bot] committed 313 days ago

Verified a7df7b19

Revert to use setup-dotnet@v3 for x86 release pipeline (#26559)

chilo-ms committed 313 days ago

Verified 1f006796

Adding SME1 Convolution Kernel to convole_kleidiai.cpp (#26402)

JonathanC-ARM committed 314 days ago

Verified 8fe48049

[webgpu] Fix `GatherBlockQuantized` on Intel ADL/TGL platforms (#26526)

daijh committed 314 days ago

Verified d6219b6b

Adds automatic loading of the external `tensorrt_plugins` library in the TensorRT-RTX Execution Provider to support custom transformer plugins (#26558)

anujj committed 314 days ago

Verified 907ede2e

webgpu qmoe (#26489)

Guenther Schmuelling committed 314 days ago

Verified 59557984

[webgpu] revise class ComputeContext in WebGPU EP (#26552)

fs-eire committed 315 days ago

Verified 43568cb7

[QNN-EP] Use same power votes as burst for SHP (#26465)

quic-calvnguy committed 316 days ago

Verified c30905d6

[webgpu] Fused SplitPackedQKV with FusedQKRotaryEmbedding (#26447)

xiaofeihan1 committed 317 days ago

Verified cf8476b1

Example plugin EPs - readme and minor updates (#26498)

edgchen1 committed 317 days ago

Verified ed46e3a8

[Build] update cuda 13 package: fatbin compress mode and cuda archs (#26516)

tianleiwu committed 317 days ago

Verified 8fceca0d

[WebNN] Correct behavior of Softmax < opset 13 (#26391)

Honry committed 318 days ago

Verified a4e44a47

Older