0.3 has our own variation of Immix algorithm [1] [2]. On current development snapshots we've also made it parallel and partially concurrent which is expected to bring major speedups over previous single-thread implementation on applications with medium to large-sized heaps [3].

[1] https://dl.acm.org/citation.cfm?id=1375586

[2] https://dl.acm.org/citation.cfm?id=2660198

[3] https://github.com/scala-native/scala-native/pull/1423