microsoft/onnxruntime

Pull Requests Commits

Refactor WebGPU MatMul compute dispatch

qjia7 committed 1 day ago

c8734854

docs: clarify WebGPU MatMul dispatch ownership

qjia7 committed 1 day ago

4ccbbbc7

Format Intel WebGPU Split-K profiles

qjia7 committed 1 day ago

c9b6a60f

Refactor WebGPU Split-K vendor policy

qjia7 committed 2 days ago

65d4725e

Address WebGPU MatMul review feedback

qjia7 committed 2 days ago

2d529c6f

Refine WebGPU MatMul vendor scheduling and tuning

qjia7 committed 2 days ago

a3e7c2dd

fix: handle zero-K WebGPU MatMul inputs

qjia7 committed 2 days ago

296503f9

docs: align MatMul scheduler validation details

qjia7 committed 2 days ago

4f0bef6c

fix: address WebGPU MatMul review and CI failures

qjia7 committed 2 days ago

614bd012

Merge remote-tracking branch 'origin/main' into codex/webgpu-matmul-algorithm-scheduler

qjia7 committed 2 days ago

ae4148a9

Harden shape inference for contrib ops (#32607)

shiyi9801 committed 2 days ago

Verified bca9d15b

[CUDA] Enable non-causal PagedAttention with small cache pages (#32619)

Copilot committed 2 days ago

Verified 51bda5f5

fix: validate forced WebGPU MatMul edge cases

qjia7 committed 3 days ago

50e2cc99

docs: move WebGPU MatMul scheduler design

qjia7 committed 3 days ago

0b03a49a

build: keep Vulkan configuration test-only

qjia7 committed 3 days ago

a75f43bb

Fix out-of-bounds read when a Loop/Scan body input is also an initializer (#32609)

wangw-1991 committed 4 days ago

Verified 09dfa6ad

Fix heap buffer overflow from reusing a packed sub-byte buffer for a full-byte tensor (#32611)

wangw-1991 committed 4 days ago

Verified 061eecbd

[CPU] Add opt-in MatMulNBits FP32 throughput mode (#31723)

mustjab committed 4 days ago

Verified e69f8415

Add `activation` mode to `com.microsoft::GatedRMSNorm` for SiLU/Swish/Sigmoid across CPU, CUDA, and WebGPU (#32512)

Copilot committed 4 days ago

Verified 0e18025d

Speedup on macos github actions: binary vcpkg cache (#32692)

eserscor committed 4 days ago

Verified e6cd18da

Remove WebGPU EP Python packaging pipeline (#29744)

edgchen1 committed 4 days ago

Verified 1100e116

Collapse A10 pools to a single pool (#32691)

eserscor committed 4 days ago

Verified b55c1c39

Update syntax for release mode in NPM (#32689)

eserscor committed 4 days ago

Verified 46a9c8f5

Fix binary vcpkg caching for github actions (#32682)

eserscor committed 4 days ago

Verified f90e813a

Limit sparse initializer dense allocation (#32684)

apsonawane committed 5 days ago

Verified 62f5193d

Execution plan to offload export on CPU for a MoE (#32261)

xadupre committed 5 days ago

Verified 6b3206e3

Load CUDA external data through pinned buffers (#32437)

xadupre committed 5 days ago

Verified b8df4d66

Combine npm and nuget packing pipelines (#32655)

eserscor committed 5 days ago

Verified ca66cc8f

test: skip unsupported forced MatMul paths

qjia7 committed 5 days ago

b6f7cbc2

test: guard Vulkan MatMul algorithm tests

qjia7 committed 5 days ago

145b259a

Older