onnxruntime
[OVEP] OpenVINO EP Features and bug-fixes for ORT-1.24 - Follow up
#27007
Merged
Merged
[OVEP] OpenVINO EP Features and bug-fixes for ORT-1.24 - Follow up #27007
adrianlizarraga
merged
283 commits into
microsoft:main
from
intel:intel/ovep-1.24
[OVEP] Fix for building OVEP without vcpkg flag (#637)
2313d11c
Merge branch 'master' into sync_msft_4_4_25
a9a3d10f
Merge pull request #638 from intel/sync_msft_4_4_25
06368151
[OVEP] Updated Documentation for python wheels (#640)
1e85f1d3
Device type refactoring (#630)
80dfee9f
Enable adaptive stripping and eliminate dependency of weight sharing …
2e4d5410
Add Config for Release build
c0c347ca
Merge branch 'master' into syncing_msft_8_4_25
8517c64e
Merge pull request #643 from intel/syncing_msft_8_4_25
b999a1bb
Bug fix in provider key verification (#644)
a8527b90
Fix the LUID check (#647)
4e63ef68
Merge branch 'master' into sync_msft_10_4_25
f0216fd1
Merge pull request #649 from intel/sync_msft_10_4_25
a5e6e05a
Update OV version for Intel Internal CI
6fc0ed0c
Merge pull request #652 from intel/jatin_update_int_ci_2025_1
9294c4c7
[OVEP] Update ov version in ort (#653)
c2558f3a
Removing support for internal ci(Intel Internal) (#662)
ad5f824d
Merge branch 'master' into sync_msft_17_4_25
254bda08
Merge pull request #663 from intel/sync_msft_17_4_25
21c7ab36
Support for triggering Internal Ci(Intel Internal) (#665)
4c0acd8a
incorporate requested changes for PR:24394 (#661)
74013352
Add support for session option ep.stop_context_sharing (#655)
269f6fe2
Merge branch 'master' into sync_msft_24_4_25
74518847
Merge pull request #675 from intel/sync_msft_24_4_25
e59c069e
Revert "Add support for session option ep.stop_context_sharing (#655)…
9ee331a9
Release model proto after we have the serialized string to reduce pea…
a077c79a
Merge branch 'master' into sync_msft_2_5_25
9dd6a1ad
Merge pull request #683 from intel/sync_msft_2_5_25
e3540092
add channel wise quantization option for QDQ, and opt for intel NPU (…
8d2f3c41
Don't include initializers in compute capability (#686)
76d31224
Merge branch 'master' into sync_msft_16_5_25
e6dd15bc
Merge pull request #692 from intel/sync_msft_16_5_25
080f66ba
[OVEP] Fixed coverity issues (#693)
8d57482d
[GPU] Enable qdq_stripping path for GPU (#694)
599cd259
Optimize CPU time spent in inference path (#682)
1a22a572
Support workload type for dynamic shaped models (#690)
b602f472
Merge branch 'master' into sync_msft_2_6_25
ad5413d5
Merge pull request #699 from intel/sync_msft_2_6_25
be8fdedf
Sahar/tdr failure (#698)
be8f8be2
feat: ORT GenAI Stateful Compilation changes (#676)
660adfcf
OVEP unittest (#685)
5ec3c7ee
Merge branch 'master' into ovep-develop
81c5f594
Merge pull request #708 from intel/sync_msft_13_6_25
f86768bb
Support for bounded dynamic model (#701)
ca06b7a3
[OVEP] Fix for appropriate device not selected (#696)
9b245a45
Improve the condition for device=GPU and qdq stripping enabled (#705)
409b2243
Optimize CPU time spent in inference path (continued) (#695)
cbef6175
feat: Enable EpContext OVIR Encapsulation (#704)
6d04a2ec
Add operator hardSwish into OVEP (#709)
278f6a7d
[CVS-169168] Change name of metadata file and add filepath validation…
90869ffe
fix: Fix logic in OnnxToOvNetworkBindings for stateful models (#719)
e141e2ff
Merge branch 'master' into sync_msft_25_6_25
7176c12e
Merge pull request #721 from intel/sync_msft_25_6_25
a95ed23e
[OVEP] Fix UnsupportedOpMode logic for Reshape
7f86fad4
Fix metadata name when ep.context_file_path is not provided (#722)
dfc8ff07
EPCtx changes for dynamic model with reshape_input provider optional …
0c1fcfd6
Merge branch 'master' into ovep-develop
f3672dfd
Merge pull request #724 from intel/sync_msft_2_7_27
c3284bc0
fix: enable test & lint fixes (#725)
13e77924
Reduce the peak memory even with CPU fallback by moving the fallback …
80daa9b5
Add QDQ scale propagation pass (#713)
e2ec2b38
Added to remove MAC CI Warnings (#726)
3238a902
fix: Undo perf_runner changes (#727)
1695972e
Merge branch 'master' into sync_msft_4_7_25
3106e3bb
Merge pull request #730 from intel/sync_msft_4_7_25
aa1730ca
Merge branch 'master' into sync_msft_7_7_25
efd14d99
Merge pull request #731 from intel/sync_msft_7_7_25
66eceb92
Merge branch 'master' into sync_msft_8_7_25
c9a53485
Merge pull request #733 from intel/sync_msft_8_7_25
8264e368
Enable dynamic path for NPU when enable_causallm is true (#732)
54151b1f
Merge branch 'master' into sync_msft_9_7_25
b4b94929
Merge pull request #734 from intel/sync_msft_9_7_25
f89f64e3
Allow zero-element tensors to get set (#737)
8d36ad26
Cluster Change to avoid Dangling DQLinear (#739)
89ccd817
Fix the model copies and redefinitions for CPU fallback (#728)
05126ff4
Revert "Cluster Change to avoid Dangling DQLinear (#739)" (#743)
776bedf7
Sync ORT main 16 07 25 (#744)
bc3dc45c
Merge branch 'master' into sync_msft_17_7_25
bda52f6d
Revert "Sync ORT main 16 07 25 (#744)" (#745)
87a7ac04
Merge pull request #746 from intel/sync_msft_17_7_25
acb29b6b
Added support for 2025.2 and enabled SimplifiedLayerNormalization op …
ea8cf217
Merge branch 'master' into sync_ort_main_19_7_25
8642e8b7
Merge pull request #749 from intel/sync_ort_main_19_7_25
a616abab
[OVEP] Fix coverity issues (#753)
217f2852
[OVEP] feat: Integrate new ABI with Legacy OVEP Plugin (#747)
138c6e05
Merge branch 'master' into sync_msft_22_7_25
dfdda93e
Merge pull request #756 from intel/sync_msft_22_7_25
a90ebe51
Merge branch 'master' into sync_msft_23_7_25
01c762b1
Merge pull request #757 from intel/sync_msft_23_7_25
e1b2c287
ov_factory: Use 'GPU_DEVICE_ID' property to match with ORT device_id …
3faf7d97
Merge branch 'master' into sync_msft_24_7_25
99e516bb
Merge pull request #760 from intel/sync_msft_24_7_25
9c7a1517
[OVEP] Fix for upsample optype (#761)
2306b4a7
Merge branch 'master' into synccc_msft_29_7_25
1833f047
Merge pull request #762 from intel/synccc_msft_29_7_25
420ec3a4
Merge branch 'master' into sync_msft_31072025
5a583f93
Merge pull request #764 from intel/sync_msft_31072025
ba8e3e78
[OVEP] Remove checks from load_config (#765)
f4da9f13
Add self-detecting on-the-fly bfloat16->float16 conversion pass (#741)
ed9e4250
[OVEP] Mild weight sharing- quantization paramters are kept as initia…
47a231ad
Merge pull request #766 from intel/niharika/mild_weight_sharing
c5330070
Merge branch 'master' into sync_msft_05082025
ddc64b90
Merge pull request #769 from intel/sync_msft_05082025
71f88774
Merge branch 'master' into sync_msft_07082025
1170738d
Merge pull request #770 from intel/sync_msft_07082025
055300fc
Merge branch 'master' into sync_msft_08082025
7f7091e0
Merge pull request #772 from intel/sync_msft_08082025
c0c1ed7b
Cluster Change to avoid Dangling DQLinear
e4f8acb0
Error in subgraph
aa31709b
Merge pull request #774 from intel/sahar/psu_lora
e77892e1
Merge branch 'master' into sync_msft_12082025
7ce9c96f
Merge pull request #775 from intel/sync_msft_12082025
07bf6161
Fix to set precision from config (#778)
725744a2
Fix the load_config not work when set INFERENCE_PRECISION_HINT (#777)
609dfbf8
Fix failing case where input onnx model is used with shared context e…
a780d5b3
[OVEP] Support for providing layout to input/output to OpenVINO (#767)
a6359eed
Merge branch 'master' into sync_msft_18082025
f05d6691
Merge pull request #780 from intel/sync_msft_18082025
78e46e2a
Merge branch 'master' into sync_msft_20082025
6dd04a58
Merge pull request #781 from intel/sync_msft_20082025
aea572d6
OVInferRequest::SetTensor: Set tensor upon cached_binding shape misma…
e6346544
Updated load_config mapping to make it a passthrough to OV properties…
0bad3d7e
Merge branch 'master' into sync_msft_25082025
cb59e2dc
Merge pull request #789 from intel/sync_msft_25082025
e812aea5
Fix model copying for QDQ stripping (#784)
8ecdbd06
Sahar/psu lora fix 2 (#788)
e974fbee
Commit PSU Lora fix (#791)
ed6d8e03
Fix to resolve EPCtx filename confusion with wrapped OVIRs (#793)
6aa84204
Allow mmapping native binaries in ov 2025.3 (#794)
866a24cf
fix: lint fixes (#795)
7eafccb9
Update operator support status for OpenVINO 2025.2 (#792)
e019aa15
[OVEP] Fix cov issues (#796)
167f2ad8
Merge branch 'master' into sync_msft_28082025
676a4d22
Merge pull request #797 from intel/sync_msft_28082025
b9a1885f
remove onnxruntime/test/providers/openvino/openvino_plugin.cc (#798)
b7244f1a
OVEP-CI updating version (#799)
be346bb7
Sync with Microsoft ONNX Runtime - 01/09/2025 (#801)
2f1ad9d0
Revert "Sync with Microsoft ONNX Runtime - 01/09/2025 (#801)"
cc8b267c
Merge pull request #803 from intel/revert-801-sync_msft_01092025
161e6456
Re-enable setting default precision for OV devices (#802)
c3276dea
bf16 tensor handling related fix (#805)
edc51ea8
Merge branch 'master' into sync_msft_10_9_25
fba645c6
Merge pull request #810 from intel/sync_msft_10_9_25
3215fefb
[OVEP] Fix to increase provider value upto 2048 char (#807)
a5bd8eea
Make cache_dir and num_stream options session-local (#809)
02cf7e35
Merge branch 'master' into sync_msft_15092025
f3d7b4fa
Merge pull request #811 from intel/sync_msft_15092025
c0981133
Merge branch 'master' into sync_msft_17092025
42b1fbb7
Merge pull request #814 from intel/sync_msft_17092025
f7a56565
Fix performance degradation in Ubuntu (#815)
58e83ef5
Merge branch 'master' into sync_msft_26092025
89db8a48
Merge pull request #818 from intel/sync_msft_26092025
9d565328
Merge branch 'master' into sync_msft_01102025
ce3a0330
Merge pull request #820 from intel/sync_msft_01102025
2d82713c
Don't embed external initializers into the proto to avoid 2GB limit (…
f8b09041
Fix Regression in Model PSS and PSR, add check for zero-size initiali…
d1025547
Merge branch 'master' into sync_msft_10102025
9a5e182c
Merge pull request #826 from intel/sync_msft_10102025
26524797
When dynamic shapes are used the proto might be overriden and offsets…
b685871c
trigger stateful path for Phisilica model
65bbecc0
unify the code
1e132f36
[OVEP] fixed NPU exception message when CPU fallback is disabled (#832)
59f22e17
CVS-174886: Make onnxruntime tests pass on OpenVINO (#790)
5c40da54
[OVEP] Enable Session Option to Stop Context Sharing (#822)
914c07ab
Merge branch 'master' into sync_msft_16102025
37b99c10
Merge pull request #833 from intel/sync_msft_16102025
f7483e67
CVS-174008: Enable ETW Tracing for OVEP (#827)
19ebc1f2
address PR review
25c6976b
CVS-174585: Memory map shared weights when possible (#829)
397c61bb
Merge branch 'master' into sync_msft_27102025
8e9ecb72
Merge pull request #835 from intel/sync_msft_27102025
c7204c55
Run lintrunner and fix formatting issues (#836)
6d410655
CVS-175734- [OVEP GPU] add GQA in support list for GPU backend (#830)
eff6cac4
CVS-175737-[OVEP] Expose kvcache_rewind python api (#831)
b1f77501
update the keyword for matching key_value_input_names
513e1981
CVS-175447-[OVEP] Add a check for type mismatches in QDQ stripping (#…
20de366c
optimize the code
d7ee534b
revert original code which is functional
3c1c4c3e
remove useless change
7c1720d5
Merge branch 'master' into sync_msft_03112025
81772e90
Merge pull request #843 from intel/sync_msft_03112025
323cfeb2
Merge branch 'ovep-develop' into make_stateful_phisilica
2041402c
Merge pull request #821 from Kotomi-Du/make_stateful_phisilica
b9a73f32
[OVEP] support to run layout feature using python bindings (#846)
7aa53632
CVS-176081: Add support for nested maps to load_config parsing (#844)
fa68db19
[CVS-172796] fix bfloat16 conversion when single cast node to bfloat1…
70acefe2
ov_stateful_patch_utils: Remove NPUW WA for avoiding SinCos when cont…
57431973
CVS-175880 Implement single bin (#847)
10af800e
Merge branch 'master' into sync_msft_14112025
52c88dfd
Merge pull request #848 from intel/sync_msft_14112025
0f8a7546
Run lintrunner and fix formatting issues (#849)
51493cd9
CVS-175736 - [OVEP] Optimize Stateful Path: use output-to-input strat…
d0bac3eb
CVS-167480 : Report import failure error code (#715)
75c11ae3
CVS-175504: Additional single bin simplifications + fixes for bi-dire…
d951954e
CVS-176574 : Fix memory leaks for protobuf & DataOps (#852)
0d68ee56
CVS-175504 Fix mixing weight shared and non-shared models (#854)
dbd1ce04
ovep stateful: Enable explicit slice of prefill logits when NPUW_SLIC…
fa7ab09f
Updating OVEP to support 2025.4.0 (#853)
24c833c9
Merge branch 'master' into sync_msft_19112025
4b1976a2
CVS-175119-[OVEP] Fixed possibility of array index out of bounds in s…
a69cbf7c
Merge branch 'ovep-develop' into sync_msft_19112025
cb8c2709
Merge pull request #856 from intel/sync_msft_19112025
ade6a2e2
skipped failing testcase (MathOpTest.Clip_Default_int64) (#860)
2f7212c3
reset ort.eprp file changes (#862)
5a06f689
Merge branch 'master' into sync_msft_25112025
6274e3cd
Merge pull request #863 from intel/sync_msft_25112025
bb738ba9
Sync with Microsoft ONNX Runtime - 03/12/2025 (#867)
39d6db58
Revert "Sync with Microsoft ONNX Runtime - 03/12/2025 (#867)"
8291389e
Merge pull request #868 from intel/revert_sync_03122025
fa8f4649
CVS-176847 Reshape Input to apply to OVIR models as well (#864)
e0779f29
Merge branch 'master' into sync_msft_09122025
9164f62f
Merge pull request #879 from intel/sync_msft_09122025
f0685add
CVS-178008: Revert tests for ovep (#883)
13e537d4
Merge branch 'master' into sync_msft_15122025
f48e9e76
Merge pull request #885 from intel/sync_msft_15122025
5bedf088
CVS-177904: Address code review comments for backend_manager.cc (#881)
dbfade4c
Reorder KV cache using the new gather_by_axis API
e67a96b4
Do a ScatterElementsUpdate-based reorder during execution
349eff9a
Get variable update lengths from incoming indices
49ed42a6
Make changes to support new KVCache fusion
8dea771c
Add proper include
6438df8b
add reorder KV cache API
c76fb925
clean up code
13a9f01c
add post process for internal handled inputs
101102db
disable update_kvcache for npu + pass kv info
59bd56eb
refactor code
7ea12e74
minor change
ab95e39f
refactor with int32 indices, string_view parsing.
16658657
fix naming
0a733e5e
Merge branch 'master' into sync_msft_05012026
8c9afa68
Merge pull request #890 from intel/sync_msft_05012026
3f9e49c0
CVS-178009: Fix provider tests based on code review (#880)
ba9dd5eb
Merge branch 'ovep-develop' into update_kvcache_node
e1da324a
fix comment
5f74c2f8
Merge pull request #872 from Kotomi-Du/update_kvcache_node
ca044e39
[OVEP] Fix model import logic to check for OV tensor before stream (#…
564e1977
Fix tensor constness for OV 2026.0 (#895)
4d4d5fc6
CVS-174235 : OV SDK Version Check Removed for EpCtx Consumption Flow …
cbec7749
Throw exception on plugin ep creation failure (#894)
2ced6f49
Merge branch 'master' into sync_msft_09012026
f4fa3780
Merge pull request #897 from intel/sync_msft_09012026
2d61435e
CVS-175980: Disable pre-allocation of memory for dynamic models throu…
b51ec772
perftest: iterate over the metadata entries first in filter_ep_device…
bc3ccde7
Fix model import for OV versions older than 26.0 (#898)
c56cf710
[OVEP] Fix lint issues (#901)
5b8c7ed1
Merge branch 'master' into sync_msft_14012026
48c737cb
Merge pull request #902 from intel/sync_msft_14012026
cd20d5d4
Adopt the required lint on MSFT side to ovep-develop PR check (#899)
e3ed2125
Remove local val scripts
799e776b
adrianlizarraga
commented
on 2026-01-14
MayureshV1
approved these changes
on 2026-01-14
adrianlizarraga
dismissed these changes
on 2026-01-14
adrianlizarraga
commented
on 2026-01-15
skip kDmlExecutionProvider for QuantizeLinearOpTest.Int8_PositiveZero…
41538169
adrianlizarraga
approved these changes
on 2026-01-15
Reviewers
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub