With the introduction of the Allocate trait in #155 we introduced a "lazy" API for preprocessing OTs. Upon further thought, this pattern can be applied to all offline preprocessing. It could potentially yield significant performance improvements by optimally batching and paralleling workloads.
Let's take a stab at this during the upcoming refactoring of mpz-garble
With the introduction of the
Allocate
trait in #155 we introduced a "lazy" API for preprocessing OTs. Upon further thought, this pattern can be applied to all offline preprocessing. It could potentially yield significant performance improvements by optimally batching and paralleling workloads.Let's take a stab at this during the upcoming refactoring of
mpz-garble