microsoft/onnxruntime

Pull Requests Commits

Test CUDA arena savings from planned workspace

chilo-ms committed 6 hours ago

33bca061

Avoid workspace declaration for MatMulNBits GEMV

chilo-ms committed 9 hours ago

3888b4b9

Bridge planned workspace access to CUDA

chilo-ms committed 5 days ago

6bce3856

Test planned workspace across MatMulNBits chain

chilo-ms committed 5 days ago

00808c31

Plan CUDA workspace with activations

chilo-ms committed 6 days ago

725f01f6

Preallocate declared CUDA workspace

chilo-ms committed 6 days ago

202269bd

Verify Level-2 workspace reservations

chilo-ms committed 6 days ago

d05286bf

Report workspace estimate discrepancies

chilo-ms committed 6 days ago

e9e0e3d4

Unify workspace selection state

chilo-ms committed 7 days ago

da899f32

Resolve workspace estimate with resource cost

chilo-ms committed 8 days ago

c8068bce

Clarify initial resource cost calculation

chilo-ms committed 8 days ago

edbd5755

Document pending resource accounting state

chilo-ms committed 8 days ago

6dbbd9b0

Rename size-based resource accountant

chilo-ms committed 8 days ago

a5af386b

Combine profiled and estimated workspace accounting

chilo-ms committed 8 days ago

37560cf0

Report committed workspace estimates after graph partitioning

chilo-ms committed 8 days ago

3afd9306

Wire max-shape inference into workspace estimation (#31613)

chilo-ms committed 8 days ago

Verified 808ba88b

Weightless support for all initializers (#29607)

chilo-ms committed 8 days ago

Verified 53211cd3

[CUDA] Split matmul nbits gemv to multiple cu files (#31837)

tianleiwu committed 8 days ago

Verified 1d69bbd6

[CUDA] Fix CUDA 13 packaging build failures (#31748)

tianleiwu committed 8 days ago

Verified 66eae3e6

[ARM] MLAS: portable machine-code SVE elementwise kernels, FEXPA exp (#31145)

mirounga committed 8 days ago

Verified 1eb1eb96

Fix GatherBlockQuantized default zero point on CUDA (#31693)

Copilot committed 8 days ago

Verified ab7d4609

Make DML runtime fused graph kernel own its model path (#31664)

adrastogi committed 8 days ago

Verified 33d509c7

[Build] Split matmul_nbits cu files (#31834)

tianleiwu committed 8 days ago

Verified 9d52a149

[Build] Reduce nvcc_threads to 1 for linux cuda plugin ci (#31729)

tianleiwu committed 9 days ago

Verified e9c76ade

Add double support for Cos op on CPU EP (#28975)

elwhyjay committed 9 days ago

Verified 3162778d

Minimize telemetry context and modernize 1DS integration (#29872)

bmehta001 committed 9 days ago

Verified 44895032

[Build] Install complete Perl runtime in Linux packaging images (#31724)

tianleiwu committed 10 days ago

Verified 2e76898a

[MLAS] Add int8 extreme-value coverage for ARM64 SymmQgemm (#31606)

kjg0724 committed 10 days ago

Verified 4c1d32b3

Validate constant tensor byte size in DML OnnxTensorWrapper (#31665)

adrastogi committed 10 days ago

Verified 4dfcacc9

WebGPU: Support int32 and uint32 for CumSum (#31702)

miaobin committed 11 days ago

Verified 31137265

Older