Increase number of cores used to calculate upsample for YOLO and Stable diffusion#16351

Draft

nkpatel-tt wants to merge 15 commits intomainfrom nkpatel/upsample_perf_improve

+289-115

Commits on Nov 26, 2024

Remove restriction of input_nsticks_per_core % w == 0 for height sharded tensor inputs.
nkpatel-tt
committed

Commits on Nov 28, 2024

Add support for block sharding for upsample.
nkpatel-tt
committed

Commits on Dec 10, 2024

Add support for block sharding.
nkpatel-tt
committed

Commits on Dec 11, 2024

Commonize code for block and height sharding.
nkpatel-tt
committed

Commits on Dec 18, 2024

Commits on Dec 19, 2024

Merge branch 'main' into nkpatel/upsample2d-generalization-remove-restrictions-input_width
nkpatel-tt
committed

Commits on Dec 20, 2024

Address review comments.
nkpatel-tt
committed

Commits on Dec 24, 2024

Reduce size of required CB for config buffer by half.
nkpatel-tt
committed

Commits on Dec 27, 2024

Merge remote-tracking branch 'origin/nkpatel/upsample2d-generalization-remove-restrictions-input_width' into nkpatel/upsample_perf_improve
nkpatel-tt
committed

Commits on Dec 30, 2024

Increase number of cores used for calculations for upsample for stable diffusion and yolo.
nkpatel-tt
committed