[GPU] Fix implicit concat fusing blocked for feature-axis and multi-user predecessors#36032

Open

deepaks2 wants to merge 9 commits into

openvinotoolkit:masterfrom

deepaks2:gpu-optimize-concat-buffer-fusing

Commits on Jun 2, 2026

[GPU] Update concat_no_implicit_gpu_onednn_4d_f16 test expectation for batch>1
deepaks2
committed
[GPU] Improve prepare_buffer_fusing onednn concat test coverage
deepaks2
committed
[GPU] Use narrower type set for multi-user concat predecessor safety check
deepaks2
committed
[GPU] Use natural-number test data in multi_user_safe_type concat test
deepaks2
committed
[GPU] Use natural-number test data in conv_as_user concat test

deepaks2
and
Copilot
committed
[GPU] Use oneDNN conv in multi_user_conv_as_user concat test

deepaks2
and
Copilot
committed
[GPU] Replace clamp with abs in multi_user_safe_type concat test

deepaks2
and
Copilot
committed
[GPU] Harden conv_as_user concat test with asymmetric inputs

deepaks2
and
Copilot
committed

Commits on Jun 3, 2026

[GPU] Updating the comments in the source code
deepaks2
committed