vulkan: optimize and reenable split_k (#10637)
Use vector loads when possible in mul_mat_split_k_reduce. Use split_k when there aren't enough workgroups to fill the shaders.
J
Jeff Bolz committed
cc98896db858df7aa40d0e16a505883ef196a482
Parent: 91c36c2
Committed by GitHub <noreply@github.com>
on 12/3/2024, 7:29:54 PM