[Release] Set environment variable on LNL CPUs before compression model compilation #3233
Conversation
This file contains 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
nikita-savelyevv
changed the title
github-actions
bot
added
NNCF Common
| @@ -115,6 +117,17 @@ def clear_ov_model_cache(): | |||
| OV_MODEL_CACHE.clear() | |||
|
|
|||
|
|
|||
| def _compile_ov_model(model: ov.Model, device_name: str, config: Dict[str, str]) -> ov.CompiledModel: | |||
| if is_lnl_cpu(): | |||
There was a problem hiding this comment.
Why do we need to set it for LNL CPU particularly and not in all cases? Is there any drawback?
There was a problem hiding this comment.
I was recommended not to set it always. @dmitry-gorokhov could you comment on the possible drawbacks?
There was a problem hiding this comment.
It has to be done on LNL only. Exporting this varibable on systems w/o AVX-VNNI support will result in illegal instruction errors. Besides that it will disable AMX/AVX512 isa on Xeons.
MaximProshin
requested changes
Jan 31, 2025
AlexanderDokuchaev
approved these changes
Jan 31, 2025
alexsu52
reviewed
Feb 3, 2025
nikita-savelyevv
commented
Feb 3, 2025
nikita-savelyevv
commented
Feb 3, 2025
| def is_lnl_cpu() -> bool: | ||
| global _IS_LNL_CPU | ||
| if _IS_LNL_CPU is None: | ||
| _IS_LNL_CPU = re.search(r"Ultra \d 2\d{2}", cpuinfo.get_cpu_info()["brand_raw"]) is not None |
There was a problem hiding this comment.
alexsu52
approved these changes
Feb 3, 2025
Co-authored-by: Alexander Suslov <[email protected]>
MaximProshin
approved these changes
Feb 3, 2025
nikita-malininn
merged commit f94cd1e
into
openvinotoolkit:release_v2150
15 checks passed
nikita-savelyevv
added a commit
to nikita-savelyevv/nncf
that referenced
this pull request
Feb 4, 2025
…el compilation (openvinotoolkit#3233) ### Changes Set `export DNNL_MAX_CPU_ISA=AVX2_VNNI` during weights compression on LNL CPUs. ### Reason for changes Fix compression on LNL CPUs. ### Related tickets 161336 ### Tests - https://github.com/openvinotoolkit/nncf/actions/runs/13075940656 ✅ - https://github.com/openvinotoolkit/nncf/actions/runs/13075938396 ✅ - NNCF/job/nightly/job/sdl/job/nightly_trigger/426/ ✅ --------- Co-authored-by: Alexander Suslov <[email protected]>
This was referenced Feb 4, 2025
nikita-malininn
pushed a commit
that referenced
this pull request
Feb 4, 2025
### Changes Follow-up to #3233 . Fix an issue arising during documentation building. ### Reason for changes Caught here: https://github.com/openvinotoolkit/nncf/actions/runs/13131916886/job/36638718655 . This test is run on PRs to develop only. That's why it was not caught during merging to release branch. ### Tests https://github.com/openvinotoolkit/nncf/actions/runs/13132576726/job/36640708447?pr=3246
nikita-savelyevv
added a commit
to nikita-savelyevv/nncf
that referenced
this pull request
Feb 4, 2025
…el compilation (openvinotoolkit#3233) ### Changes Set `export DNNL_MAX_CPU_ISA=AVX2_VNNI` during weights compression on LNL CPUs. ### Reason for changes Fix compression on LNL CPUs. ### Related tickets 161336 ### Tests - https://github.com/openvinotoolkit/nncf/actions/runs/13075940656 ✅ - https://github.com/openvinotoolkit/nncf/actions/runs/13075938396 ✅ - NNCF/job/nightly/job/sdl/job/nightly_trigger/426/ ✅ --------- Co-authored-by: Alexander Suslov <[email protected]>
nikita-savelyevv
added a commit
to nikita-savelyevv/nncf
that referenced
this pull request
Feb 4, 2025
### Changes Follow-up to openvinotoolkit#3233 . Fix an issue arising during documentation building. ### Reason for changes Caught here: https://github.com/openvinotoolkit/nncf/actions/runs/13131916886/job/36638718655 . This test is run on PRs to develop only. That's why it was not caught during merging to release branch. ### Tests https://github.com/openvinotoolkit/nncf/actions/runs/13132576726/job/36640708447?pr=3246
AlexanderDokuchaev
pushed a commit
that referenced
this pull request
Feb 4, 2025
…tion (#3246) Copy of #3233 and #3247 to develop branch --------- Co-authored-by: Alexander Suslov <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Changes
Set
export DNNL_MAX_CPU_ISA=AVX2_VNNIduring weights compression on LNL CPUs.Reason for changes
Fix compression on LNL CPUs.
Related tickets
161336
Tests