cpu usage is very low during parsing.
with multithreaded option on - no difference, same long parsing.
after all testing i found it is a problem only of laubwerk trees. they do not render fast in corona, i dont know why but they are parsing very slow.
tested with other hq tree models converted to proxy. it parses in 10-12 sec. 100000 instances using multiscatter. and cpu is loaded fully with generic models.
that is why laubwerk so slow , it dows not use cpu fully when parsing.