Skip to content

[GPU] integration of OneDNN grouped_gemm#34153

Open
e-ddykim wants to merge 18 commits intoopenvinotoolkit:masterfrom
e-ddykim:gpu_onednn_grouped_gemm
Open

[GPU] integration of OneDNN grouped_gemm#34153
e-ddykim wants to merge 18 commits intoopenvinotoolkit:masterfrom
e-ddykim:gpu_onednn_grouped_gemm

Conversation

@e-ddykim
Copy link
Contributor

@e-ddykim e-ddykim commented Feb 16, 2026

Details:

  • This PR integrates OneDNN grouped gemm that replaces the micro_gemm based ocl_v2 moe_gemm impl.
    • expects improved performance and supports int8 weight quantized moe models.

Tickets:

  • 177922

@github-actions github-actions bot added the category: GPU OpenVINO GPU plugin label Feb 16, 2026
@e-ddykim e-ddykim force-pushed the gpu_onednn_grouped_gemm branch from 9630fa7 to 06b1e7d Compare February 16, 2026 19:37
@e-ddykim e-ddykim force-pushed the gpu_onednn_grouped_gemm branch from ffa7991 to f09db82 Compare February 25, 2026 14:13
@github-actions github-actions bot added the category: build OpenVINO cmake script / infra label Feb 25, 2026
@e-ddykim e-ddykim force-pushed the gpu_onednn_grouped_gemm branch 3 times, most recently from f3e95d0 to ec2c9ed Compare March 4, 2026 08:57
@e-ddykim e-ddykim marked this pull request as ready for review March 4, 2026 10:45
@e-ddykim e-ddykim requested review from a team as code owners March 4, 2026 10:45
@e-ddykim e-ddykim added this to the 2026.1 milestone Mar 5, 2026
@e-ddykim e-ddykim force-pushed the gpu_onednn_grouped_gemm branch from 42d2c29 to 25adf30 Compare March 9, 2026 03:14
@e-ddykim e-ddykim force-pushed the gpu_onednn_grouped_gemm branch from 3b27044 to 4ce7257 Compare March 11, 2026 00:57
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

category: build OpenVINO cmake script / infra category: GPU OpenVINO GPU plugin Code Freeze priority: high High piority under_perf_check

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants