perf(diffusion): FLUX.1 performance optimization kernels (#187)#189

Closed

m96-chan wants to merge 2 commits intomainfrom

Commits on Jan 1, 2026

perf(diffusion): add native CUDA kernels for FLUX.1 (#187)

m96-chan
and
claude
committed
perf(flux): optimize blocks.py and attention.py to use GPU kernels

m96-chan
and
claude
committed