Benchmarks compare performance of existing implementations of Sequence::toSet and Sequence::toList extension functions with optimized version that avoids allocation of underlying collection when possible.
Optimized implementations are based on proposal from KT-55091.
Measurements were performed using JDK-8 and JDK-11 on a host with Apple M2 Max CPU.
The only varying parameter size affect the size of a sequence transformed to either a list, or a set depending on a particular benchmark.
Optimized implementations of Sequence::toList and Sequence::toSet should either superior, or the same performance as existing implementation. Optimized implementations also reduce allocation rate (as expected).