Skip to content

CPU: fixed issue #30098 #30143

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Open
wants to merge 1 commit into
base: master
Choose a base branch
from
Open

CPU: fixed issue #30098 #30143

wants to merge 1 commit into from
+11 −4

Conversation

ilya-lavrenov
Copy link
Contributor

Details:

  • Don't use FP16 instructions w/o guard

Tickets:

@ilya-lavrenov ilya-lavrenov added this to the 2025.2 milestone Apr 15, 2025
@ilya-lavrenov ilya-lavrenov requested review from a team as code owners April 15, 2025 09:08
@github-actions github-actions bot added the category: CPU OpenVINO CPU plugin label Apr 15, 2025
@ilya-lavrenov ilya-lavrenov mentioned this pull request Apr 15, 2025
Open
2 tasks
@ilya-lavrenov
Copy link
Contributor Author

This PR fixes code from #28182
@NishantPrabhuFujitsu please, have a look

@ilya-lavrenov ilya-lavrenov added category: build OpenVINO cmake script / infra platform: arm OpenVINO on ARM / ARM64 labels Apr 16, 2025
@ilya-lavrenov ilya-lavrenov enabled auto-merge April 17, 2025 14:04
allnes
allnes reviewed Apr 17, 2025
View reviewed changes
src/plugins/intel_cpu/src/nodes/kernels/scaled_attn/mha_single_token.cpp
@@ -822,7 +825,7 @@ static ov::float16 dot_product_fp16(ov::float16* a,
float16_t sum_2 = svaddv_f16(pg, sum2);
float16_t sum_3 = svaddv_f16(pg, sum3);
sum = static_cast<float>(sum_0 + sum_1 + sum_2 + sum_3);
# else
# elif defined(HAVE_NEON_FP16)
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I checked this on macOS, and the define is not enabled.
Perhaps we can use defined(__ARM_NEON) && defined(__ARM_FEATURE_FP16_VECTOR_ARITHMETIC) instead for macOS.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@allnes allnes allnes left review comments

At least 1 approving review is required to merge this pull request.

Assignees

@alvoron alvoron

@allnes allnes

Labels
category: build OpenVINO cmake script / infra category: CPU OpenVINO CPU plugin platform: arm OpenVINO on ARM / ARM64
Projects
None yet
Milestone
2025.2
Development

Successfully merging this pull request may close these issues.

[Build]: target specific option mismatch in softmax kernels for aarch64
3 participants
@ilya-lavrenov @allnes @alvoron