Skip to content

Common/CPU

-iree-codegen-cpu-lower-to-ukernelslink

Separate out parts of the IR that lower to a micro-kernel

Optionslink

-skip-intermediate-roundings : Allow skipping intermediate roundings, e.g. in f16 ukernels internally doing f32 arithmetic.

-iree-codegen-cpu-prepare-ukernelslink

Rank reduce operations to fit existing ukernels requirements.For example, batch_mmt4d ops are decomposed to mmt4d ops