microsoft/onnxruntime

Pull Requests Commits

Add KernelContext_GetSyncStream

tianleiwu committed 66 days ago

e3d1cc9e

test(cuda): rely on user stream ordering

tianleiwu committed 66 days ago

a1a2574c

fix(cuda): disable unsafe plugin concurrent runs

tianleiwu committed 66 days ago

3ce7ff36

fix(cuda): honor default user stream in plugin EP

tianleiwu committed 66 days ago

45b1e11f

tianleiwu committed 66 days ago

55683b67

[CUDA plugin] Fix graph-stream ownership intent and drop dangling scratch Stream*

tianleiwu committed 66 days ago

968fbe87

tianleiwu committed 66 days ago

ed129141

tianleiwu committed 66 days ago

2de6303e

tianleiwu committed 66 days ago

fed19f7e

tianleiwu committed 66 days ago

d257b39f

tianleiwu committed 66 days ago

81005448

Address review feedback: stream-order plugin scratch, clarify Shape/PrePack comments

tianleiwu committed 66 days ago

6bfd8241

cuda plugin user stream for cuda graph

tianleiwu committed 67 days ago

789c7712

Fix DropQDQ not dropping Q/DQ around opset-22 MaxPool (#29201)

tairenpiao committed 67 days ago

Verified c69373c4

Fix CPU GQA NaN output for right-padded batched prompts with rotary embeddings (#29069)

Copilot committed 67 days ago

Verified 701d88ff

[CUDA] Enable CUDA GQA QK-Norm and XQA decode (#29186)

tianleiwu committed 67 days ago

Verified 1472c16e

[Pipeline] Fix Python packaging pipeline failures (#29200)

hariharans29 committed 67 days ago

Verified 8c62bee6

Fix GroupQueryAttention right-padded rotary prefill CUDA test (#29218)

tianleiwu committed 67 days ago

Verified 14a6c9e9

Experimental C++ API update (#29142)

edgchen1 committed 67 days ago

Verified 8c856fb9

fix bound check on webgpu/slice (#28704)

Guenther Schmuelling committed 67 days ago

Verified bc26b968

Bump @babel/core from 7.29.0 to 7.29.6 in /js/react_native/e2e (#29156)

dependabot[bot] committed 67 days ago

Verified a0aec633

[CUDA] update doc for cuda contrib op GroupQueryAttention (#29173)

tianleiwu committed 67 days ago

Verified 8c08e943

Remove the dynamic WGSL generator (duktape/Node) path (#29141)

danielsongmicrosoft committed 67 days ago

Verified 50f0314b

Address github issue 29071 (#29158)

yuslepukhin committed 67 days ago

Verified 9ad4d58c

Build with abseil 20250814 (NVCC) (#28586)

mc-nv committed 67 days ago

Verified 5b361106

[CUDA] Enable XQA by default for FP16/BF16 GQA (#29046)

namgyu-youn committed 68 days ago

Verified 6be94ded

webgpu: fix GQA batched right-padded prefill with do_rotary (#29002)

qjia7 committed 68 days ago

Verified 5ad25165

Add int8/uint8 CPU support for SpaceToDepth and int8 for DepthToSpace (#29154)

ArsalanShakil committed 68 days ago

Verified 737d2661

Fix TransposeOptimizer type error on zero-point-less DequantizeLinear (#28716) (#29192)

tairenpiao committed 69 days ago

Verified d3746361

Fix XNNPACK Clip/Relu fusion when the producer output has other consumers (#28814)

tairenpiao committed 69 days ago

Verified 86cbd205

Older