[NPU][WS] Remove the references towards original weights sooner during init execution#34072
Merged
razvanapetroaie merged 4 commits intoopenvinotoolkit:masterfrom Feb 16, 2026
Conversation
Move the OV model through compilation pipeline to allow earlier deletion of it. In theory, this should enable early deletion of original weights and in turn reduce the RAM footprint of WS schedule. This seems possible due to OV model being always cloned internally for the compilation.
PatrikStepan
approved these changes
Feb 12, 2026
MirceaDan99
approved these changes
Feb 12, 2026
Naseer-010
pushed a commit
to Naseer-010/openvino
that referenced
this pull request
Feb 18, 2026
…g init execution (openvinotoolkit#34072) ### Details: - This is an optimization that should decrease the memory consumed by the execution of the init schedules. - Note that, since in these changes we are freeing `shared_ptr` objects sooner, the benefit of this optimization can be observed only if the NPU plugin was the only component to hold the original weights pointers. Therefore, in cases such as the compilation flow, we don't expect to see any impact (the user holds a pointer towards the model, and implictly all original weights). But there might be an impact on the import flow, if the `weights_path` config option is used. Atm, no meaningful measurements have been taken. --------- Co-authored-by: Golubev, Andrey <andrey.golubev@intel.com>
wilson-seok
pushed a commit
to wilson-seok/openvino
that referenced
this pull request
Feb 19, 2026
…g init execution (openvinotoolkit#34072) ### Details: - This is an optimization that should decrease the memory consumed by the execution of the init schedules. - Note that, since in these changes we are freeing `shared_ptr` objects sooner, the benefit of this optimization can be observed only if the NPU plugin was the only component to hold the original weights pointers. Therefore, in cases such as the compilation flow, we don't expect to see any impact (the user holds a pointer towards the model, and implictly all original weights). But there might be an impact on the import flow, if the `weights_path` config option is used. Atm, no meaningful measurements have been taken. --------- Co-authored-by: Golubev, Andrey <andrey.golubev@intel.com>
mvafin
pushed a commit
to mvafin/openvino
that referenced
this pull request
Feb 20, 2026
…g init execution (openvinotoolkit#34072) ### Details: - This is an optimization that should decrease the memory consumed by the execution of the init schedules. - Note that, since in these changes we are freeing `shared_ptr` objects sooner, the benefit of this optimization can be observed only if the NPU plugin was the only component to hold the original weights pointers. Therefore, in cases such as the compilation flow, we don't expect to see any impact (the user holds a pointer towards the model, and implictly all original weights). But there might be an impact on the import flow, if the `weights_path` config option is used. Atm, no meaningful measurements have been taken. --------- Co-authored-by: Golubev, Andrey <andrey.golubev@intel.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Details:
shared_ptrobjects sooner, the benefit of this optimization can be observed only if the NPU plugin was the only component to hold the original weights pointers. Therefore, in cases such as the compilation flow, we don't expect to see any impact (the user holds a pointer towards the model, and implictly all original weights). But there might be an impact on the import flow, if theweights_pathconfig option is used. Atm, no meaningful measurements have been taken.